{ "title": "Factory Atlas", "asOf": "2026-09-22", "coverage": "18 identified noncrypto builders; selected primary sources, not a comprehensive census.", "cases": [ { "asOf": "2026-09-22", "priority": 1, "imageStatus": "generated", "id": "warp", "name": "Zach Lloyd / Warp", "title": "Warp Factories", "category": "Factory", "summary": "Cloud agents form a development loop with explicit human decisions at specifications and release.", "verdict": "Detailed build documentation and a recent, scoped operating claim.", "evidence": "Documented architecture", "status": "September 15: vendor reports roughly 75% automation of marketing-site changes; broader product automation remains work in progress.", "projects": [ { "name": "Warp Factories", "role": "Cloud execution, shared context and factory operations.", "url": "https://www.warp.dev/blog/adopting-the-software-factory-model-crawl-walk-run" }, { "name": "Cloud factory build guide", "role": "Triage, specification, review and verification skills.", "url": "https://www.warp.dev/blog/software-factory-build-guide" }, { "name": "warp.dev factory", "role": "Reported operating example on the marketing website.", "url": "https://www.warp.dev/blog/adopting-the-software-factory-model-crawl-walk-run" } ], "orchestration": { "trigger": "Issue, human request or monitoring alert.", "controller": "Workflow triggers connect specialized cloud agents.", "workers": "Triage, specification, implementation, review, verification and monitoring agents; a separate outer-loop agent proposes review-skill improvements from human PR feedback.", "memory": "Shared repository context, specifications and traces; human review feedback can become a proposed change to the review skill.", "verification": "Review agent, computer-use checks and reviewable artifacts.", "human": "Approves specifications and final change evidence.", "output": "PR through CI/CD to a released change.", "failure": "Ambiguous issues wait for human input; failed checks return work for revision." }, "stages": [ { "name": "Triage", "owner": "Triage agent", "action": "Sort into ready, needs specification or needs input.", "certainty": "documented" }, { "name": "Specify", "owner": "Spec agent + human", "action": "Agree product behavior and technical approach.", "certainty": "documented" }, { "name": "Implement and review", "owner": "Coding + review agents", "action": "Produce and critique a patch.", "certainty": "documented" }, { "name": "Verify", "owner": "Verification agent", "action": "Exercise behavior and collect evidence.", "certainty": "documented" }, { "name": "Approve and ship", "owner": "Human + CI/CD", "action": "Decide release; monitoring starts another issue when needed.", "certainty": "documented" } ], "timeline": [ { "date": "2026-06-18", "event": "Warp announces internal factory-engineering direction.", "source": "https://www.warp.dev/blog/we-are-now-factory-engineers-not-product-engineers" }, { "date": "2026-07-23", "event": "Publishes staged cloud-factory build guide.", "source": "https://www.warp.dev/blog/software-factory-build-guide" }, { "date": "2026-08-18", "event": "Announces factory infrastructure closed beta.", "source": "https://www.warp.dev/blog/open-infrastructure-for-building-a-software-factory" }, { "date": "2026-09-15", "event": "Reports marketing-site automation and adoption stages.", "source": "https://www.warp.dev/blog/adopting-the-software-factory-model-crawl-walk-run" } ], "metrics": [ { "label": "Marketing-site changes automated", "value": "~75%", "date": "2026-09-15", "source": "https://www.warp.dev/blog/adopting-the-software-factory-model-crawl-walk-run", "caveat": "Vendor report; scope is warp.dev, not the whole Rust terminal." } ], "sources": [ { "title": "July build guide", "url": "https://www.warp.dev/blog/software-factory-build-guide", "supports": "Triage branches, human-approved specifications, review and evidence." }, { "title": "September adoption guide", "url": "https://www.warp.dev/blog/adopting-the-software-factory-model-crawl-walk-run", "supports": "End-to-end loop, shared context, monitoring and scoped automation claim." }, { "title": "June direction", "url": "https://www.warp.dev/blog/we-are-now-factory-engineers-not-product-engineers", "supports": "Dated strategy from prior evidence review." }, { "title": "August infrastructure launch", "url": "https://www.warp.dev/blog/open-infrastructure-for-building-a-software-factory", "supports": "Closed-beta milestone from prior evidence review." } ], "unknowns": [ "Independent denominator audit and failure rate.", "Automation percentage for Warp's complete product.", "Which guide stages each customer actually operates.", "The July guide marks its full Part 4 verification tutorial forthcoming; computer-use capabilities and a starting validation skill are documented." ], "image": "images/warp.png", "imageAlt": "Warp triage branches, human specification gate, agent implementation and verification, human release and monitoring loop.", "sceneContract": "Issue → triage → optional human-approved specification → implementation/review → behavior evidence → human approval → CI/CD; monitoring returns issues.", "example": "WP-01 pricing-layout fix; illustrative.", "takeaway": "Improve both the product and the review process: human feedback can produce a PR changing the factory’s own review skill.", "checked": "Re-read July and September primary guides; historical milestones come from the saved primary-source review. No deployment tested." }, { "asOf": "2026-09-22", "priority": 2, "imageStatus": "generated", "id": "strongdm", "name": "Justin McCarthy, Jay Taylor & Navan Chauhan / StrongDM", "title": "StrongDM Software Factory", "category": "Verification", "summary": "Specified work runs through agent graphs and is judged against scenarios outside the implementation.", "verdict": "The clearest verification-first architecture in this sample; current production outcomes remain unmeasured here.", "evidence": "Documented architecture", "status": "Architecture is public; latest dated operating model report retrieved is June 4, 2026.", "projects": [ { "name": "Attractor", "role": "Graph of prompt-governed work phases and conditional transitions.", "url": "https://factory.strongdm.ai/products/attractor" }, { "name": "Digital Twin Universe", "role": "Simulated third-party services for behavioral scenarios.", "url": "https://factory.strongdm.ai/" }, { "name": "CXDB", "role": "Context store with branching histories and trace inspection.", "url": "https://factory.strongdm.ai/products" }, { "name": "StrongDM ID", "role": "Identity and scoped access for people, workloads and agents.", "url": "https://factory.strongdm.ai/products" } ], "orchestration": { "trigger": "Fully specified work with scenarios.", "controller": "Attractor traverses a graph until completion or termination conditions.", "workers": "Coding and specialist agents selected by task.", "memory": "CXDB context; inspectable graph transitions and checkpoints.", "verification": "Held-out scenarios, simulated services and behavioral satisfaction.", "human": "Defines intent and evaluation requirements; reported design excludes routine human code writing/review.", "output": "A candidate that meets the graph's completion conditions.", "failure": "Unmet conditions route refinement or termination; production release policy remains unspecified." }, "stages": [ { "name": "Specify", "owner": "Human team", "action": "Define the intended outcome and scenarios.", "certainty": "documented" }, { "name": "Run the graph", "owner": "Attractor", "action": "Implement and refine through work nodes.", "certainty": "documented" }, { "name": "Exercise twins", "owner": "Test harness", "action": "Run the candidate against simulated dependencies.", "certainty": "documented" }, { "name": "Judge behavior", "owner": "Scenario evaluator", "action": "Assess trajectories; return inadequate work for refinement.", "certainty": "documented" }, { "name": "Complete or stop", "owner": "Pipeline", "action": "Reach convergence or termination.", "certainty": "documented" } ], "timeline": [ { "date": "2026-02-06", "event": "Public factory account describes scenarios and service twins.", "source": "https://factory.strongdm.ai/" }, { "date": "2026-03-12", "event": "Planning and critique model choices updated.", "source": "https://factory.strongdm.ai/weather-report" }, { "date": "2026-06-04", "event": "Latest retrieved operating model configuration update.", "source": "https://factory.strongdm.ai/weather-report" } ], "metrics": [], "sources": [ { "title": "Factory account", "url": "https://factory.strongdm.ai/", "supports": "February 6: scenarios, twins and no routine human code review." }, { "title": "Attractor", "url": "https://factory.strongdm.ai/products/attractor", "supports": "Work graphs, conditional edges and checkpoint properties." }, { "title": "Components", "url": "https://factory.strongdm.ai/products", "supports": "CXDB and identity capabilities." }, { "title": "Weather report", "url": "https://factory.strongdm.ai/weather-report", "supports": "Dated model routing history through June 4." } ], "unknowns": [ "September throughput and quality.", "The team's production-release policy.", "Independent reproduction and actual spend; the $1,000/day statement is a heuristic." ], "image": "images/strongdm.png", "imageAlt": "StrongDM agent work graph with held-out scenarios, service twins, behavioral evaluation and an unknown production-release boundary.", "sceneContract": "Specified work → Attractor → candidate → twin scenarios → behavioral evaluation → refine, converge or terminate.", "example": "SD-01 access-revocation behavior; illustrative.", "takeaway": "Keep evaluation outside the code the agent is allowed to change.", "checked": "Read story, products, Attractor and dated weather report; no test harness executed." }, { "asOf": "2026-09-22", "priority": 3, "imageStatus": "generated", "id": "steve-yegge", "name": "Steve Yegge; Gas City by Julian Knutsen and Chris Sells", "title": "Gas Town, Gas City & Wheelhouse", "category": "Factory", "summary": "A durable work ledger supports three related but distinct approaches to agent coordination.", "verdict": "Public orchestration designs plus detailed first-person accounts of operating and repairing a personal factory.", "evidence": "Reported operation", "status": "September 15: Wheelhouse reportedly improved after a major simplification, but mostly idles under token constraints.", "projects": [ { "name": "Beads", "role": "Persistent work ledger shared across the ecosystem.", "url": "https://yegge.ai/gastown" }, { "name": "Gas Town", "role": "Mayor dispatches workers; Witness observes; Refinery serializes integration.", "url": "https://yegge.ai/gastown" }, { "name": "Gas City", "role": "Knutsen/Sells SDK: configurable packs and a supervisory control layer.", "url": "https://yegge.ai/essays/welcome-to-gas-city/" }, { "name": "Wasteland", "role": "Federated work board and validation reputation, separate from the local controller.", "url": "https://yegge.ai/gastown" }, { "name": "Wheelhouse", "role": "Persistent roles coordinate work on Wyvern; not Gas Town's fixed topology.", "url": "https://yegge.ai/essays/seats-and-sunsets/" }, { "name": "Wyvern", "role": "Game receiving Wheelhouse's features and fixes.", "url": "https://play.ghosttrack.com" } ], "orchestration": { "trigger": "A work item or product direction.", "controller": "Gas Town: Mayor. Gas City: configurable packs. Wheelhouse: persistent roles; exact scheduler undisclosed.", "workers": "Gas Town Polecats; Wheelhouse role agents and delegated implementers.", "memory": "Beads; Wheelhouse adds role context, authority and history.", "verification": "Gas Town integration machinery; Wheelhouse reviewer pool and builds.", "human": "Sets direction, audits work, limits costs and release pace.", "output": "Integrated patches; Wyvern changes.", "failure": "Interrupted work resumes from records; Wheelhouse seats pause when fuel runs out." }, "stages": [ { "name": "Define work", "owner": "Human + Mayor", "action": "Create a durable task.", "certainty": "documented" }, { "name": "Dispatch", "owner": "Mayor", "action": "Assign a Polecat.", "certainty": "documented" }, { "name": "Build", "owner": "Polecat", "action": "Produce a patch.", "certainty": "documented" }, { "name": "Observe", "owner": "Witness", "action": "Watch work and recover interruptions.", "certainty": "documented" }, { "name": "Integrate", "owner": "Refinery", "action": "Serialize merges.", "certainty": "documented" } ], "timeline": [ { "date": "2026-01-01", "event": "Gas Town opens publicly.", "source": "https://yegge.ai/gastown" }, { "date": "2026-03", "event": "Wasteland adds federated work exchange.", "source": "https://yegge.ai/gastown" }, { "date": "2026-04-24", "event": "Gas City v1.0 announcement credits Knutsen and Sells.", "source": "https://yegge.ai/essays/welcome-to-gas-city/" }, { "date": "2026-09-15", "event": "Wheelhouse update describes August scale, policy simplification and fuel limits.", "source": "https://yegge.ai/essays/seats-and-sunsets/" } ], "metrics": [ { "label": "Actual Claude Max accounts", "value": "21", "date": "2026-09-15", "source": "https://yegge.ai/essays/seats-and-sunsets/", "caveat": "Self-report. 55 was estimated need, not ownership." }, { "label": "Meaningful commits per day", "value": "250–300", "date": "2026-09-15", "source": "https://yegge.ai/essays/seats-and-sunsets/", "caveat": "Reported capability after repair, not a sustained audited rate." } ], "sources": [ { "title": "Gas Town overview", "url": "https://yegge.ai/gastown", "supports": "Roles, ledger, January launch and March federation." }, { "title": "Gas City announcement", "url": "https://yegge.ai/essays/welcome-to-gas-city/", "supports": "April 24 attribution, packs and topology distinction." }, { "title": "Seats and Sunsets", "url": "https://yegge.ai/essays/seats-and-sunsets/", "supports": "September 15 operating account; roles, limits and metrics." } ], "unknowns": [ "Independent quality and throughput measurements.", "Wheelhouse's exact scheduler and private implementation.", "Reliability of arbitrary Gas City configurations." ], "image": "images/steve-yegge.png", "imageAlt": "Gas Town's durable task route and Gas City's configurable packs.", "detailImage": "images/steve-yegge-detail.png", "detailImageAlt": "Wheelhouse persistent seats, review, human release pace and fuel pause.", "sceneContract": "Task → Mayor → Polecat → Witness observation → Refinery integration; Beads persists work. Separate map: direction → persistent seat → implementer → reviewer → paced release.", "example": "GT-01 / WY-01 inventory fix; illustrative, not a reconstructed real task.", "takeaway": "Persist work independently of sessions; keep governance from stalling or overwhelming production.", "checked": "Read the three primary pages on September 22; no factory runtime was installed or benchmarked." }, { "id": "ben-guo", "name": "Ben Guo / Hraness", "title": "Hra and the personal software factory", "category": "Factory", "summary": "Repeated human dispatch, shared knowledge and repeatable UI checks support long coding sessions across personal projects.", "asOf": "2026-09-22", "priority": 4, "image": "images/ben-guo.png", "imageAlt": "Hra and the personal software factory shown as a miniature workplace with owners, evidence and exception paths.", "imageStatus": "generated", "timeline": [ { "date": "2026-08-21", "event": "Article describes 15 Codex subscriptions, roughly a dozen projects, kb, Direct and Hra.", "source": null, "sourceLabel": "User-supplied article, 21 August 2026" }, { "date": "2026-09-12", "event": "Self-report: 28 subscriptions and 253B tokens over the preceding month.", "source": "https://x.com/hraness/status/2098804050070286712" }, { "date": "2026-09-16", "event": "Self-report: 22 Devin and four Claude Code instances; instances are not subscription inventory.", "source": "https://x.com/hraness/status/2100283814551474335" } ], "metrics": [ { "label": "All-time submitted tokens", "value": "540.9B", "date": "2026-09-22", "source": "https://tokscale.ai/leaderboard", "caveat": "Public submission snapshot; cache-inclusive; not actual spend, account count or output attributable to these projects." }, { "label": "Reported subscriptions", "value": "28: 19 Codex, 5 Claude, 3 Devin, 1 Cursor", "date": "2026-09-12", "source": "https://x.com/hraness/status/2098804050070286712", "caveat": "First-person claim preserved in local archive; not a world record." }, { "label": "Reported preceding-month tokens", "value": "253B", "date": "2026-09-12", "source": "https://x.com/hraness/status/2098804050070286712", "caveat": "Different period from leaderboard total." } ], "sources": [ { "title": "Building a software factory — user-provided article", "url": null, "supports": "August 21 account of tools, conventions, dispatch and personal projects.", "kind": "user-supplied" }, { "title": "Hraness public profile", "url": "https://github.com/0thernet", "supports": "Current pinned portfolio and its descriptions, read September 22." }, { "title": "Local first-person evidence", "url": null, "supports": "Archived September token, subscription and instance statements.", "kind": "local-archive" } ], "unknowns": [ "No verified January–July history.", "No audited success rate, actual spend or production throughput.", "Hra→Oompa and kb→Wordcell lineage not established.", "Autonomous merge, deployment and failure recovery contracts not disclosed." ], "projects": [ { "name": "Hra", "role": "Metaharness and subscription pooling, described in August.", "url": "https://github.com/0thernet" }, { "name": "kb", "role": "Markdown/SQLite plans and decisions.", "url": "https://github.com/0thernet" }, { "name": "Oompa", "role": "Persistent sessions, isolated accounts and control.", "url": "https://github.com/hraness/oompa" }, { "name": "Wordcell", "role": "Agent knowledge with backlinks, semantic search and Git context.", "url": "https://github.com/hraness/wordcell" }, { "name": "Direct", "role": "Repeatable frontend scenarios and browser verification.", "url": "https://github.com/hraness/direct" }, { "name": "Wrench", "role": "Browser workflows and website SDK creation.", "url": "https://github.com/0thernet" }, { "name": "Ghostget", "role": "Typed web capabilities and captures.", "url": "https://github.com/hraness/ghostget" }, { "name": "Slopcamera", "role": "Source-retaining visual production.", "url": "https://github.com/hraness/slopcamera" }, { "name": "AICharts", "role": "Model comparison charts.", "url": "https://github.com/hraness/aicharts" }, { "name": "Atet", "role": "Iterative media toolkit reported in August.", "url": "https://github.com/0thernet" }, { "name": "Read", "role": "Reading summaries and knowledge ingestion.", "url": "https://github.com/0thernet" }, { "name": "PeopleBlade", "role": "Contact aggregation and enrichment.", "url": "https://github.com/0thernet" }, { "name": "Stripe History", "role": "Stripe news archive and monitoring.", "url": "https://github.com/0thernet" }, { "name": "Rough.day / News", "role": "News aggregation and daily summaries.", "url": "https://github.com/0thernet" }, { "name": "SEO Bot", "role": "Iterative site SEO work.", "url": "https://github.com/0thernet" }, { "name": "act60", "role": "Administrative tool suite.", "url": "https://github.com/0thernet" }, { "name": "Invest Bot", "role": "Reported personal portfolio automation; execution unverified.", "url": "https://github.com/0thernet" } ], "checked": "Primary descriptions read; architecture not installed or runtime-tested. Images use fictional work items.", "verdict": "Reported operation, backed by a public tool portfolio; unattended delivery quality is not independently established.", "evidence": "Reported operation", "status": "August article describes the operating loop; September posts report more subscriptions and agents. Current profile exposes Oompa, Wordcell and Direct, but their exact succession from Hra and kb is unverified.", "orchestration": { "trigger": "Ben sends new prompts several times daily.", "controller": "Human dispatch plus Hra/account tooling; no scheduler contract established.", "workers": "Coding sessions across projects; September also reports Devin instances.", "memory": "Agents search and write plans/decisions in kb; current Wordcell has a related public role.", "verification": "Direct supplies repeatable UI scenarios; conventions include property-based testing.", "human": "Chooses goals and tools; precise merge/release approval policy undisclosed.", "output": "Software changes and personal automation tools; not every listed project shares a proven pipeline.", "failure": "UI-check repair loop is an illustrative synthesis; retry budgets and operational recovery are undisclosed." }, "stages": [ { "name": "Dispatch goals", "owner": "Ben", "action": "Give a bounded feature to coding sessions.", "certainty": "reported" }, { "name": "Recall decisions", "owner": "Agents", "action": "Read shared plans and record new decisions.", "certainty": "reported" }, { "name": "Build changes", "owner": "Agents", "action": "Use shared stack and project conventions.", "certainty": "reported" }, { "name": "Check the interface", "owner": "Agents + Direct", "action": "Exercise repeatable UI states.", "certainty": "reported" }, { "name": "Review delivery", "owner": "Ben", "action": "Human review/release boundary remains incompletely described.", "certainty": "unknown" } ], "sceneContract": "F-17 → Ben dispatch → agents recall kb → workers implement → Direct checks → unknown review/release boundary; failed checks illustratively return to workers.", "example": "F-17 is an invented website feature, not an actual Ben project.", "takeaway": "Shared knowledge and self-checks explain sustained agent work more than account counts do." }, { "id": "jeff-emanuel", "name": "Jeff Emanuel", "title": "Agent Flywheel / ACFS", "category": "Harness", "summary": "A reproducible coding environment combines persistent sessions, task dependencies, coordination, memory and guardrails.", "asOf": "2026-09-22", "priority": 5, "image": "images/jeff-emanuel.png", "imageAlt": "Agent Flywheel / ACFS shown as a miniature workplace with owners, evidence and exception paths.", "imageStatus": "generated", "timeline": [ { "date": "2026-09-22", "event": "Current documentation reviewed; no month-by-month release chronology established.", "source": "https://github.com/Dicklesworthstone/agentic_coding_flywheel_setup" } ], "metrics": [ { "label": "All-time submitted tokens", "value": "1.829T", "date": "2026-09-22", "source": "https://tokscale.ai/leaderboard", "caveat": "Public submission snapshot; cache-inclusive; not actual spend, account count or output attributable to these projects." } ], "sources": [ { "title": "ACFS README", "url": "https://raw.githubusercontent.com/Dicklesworthstone/agentic_coding_flywheel_setup/main/README.md", "supports": "Installer, manifest, tool roster, advisory planning, memory and service lifecycle." }, { "title": "ACFS coordination integration", "url": "https://github.com/Dicklesworthstone/agentic_coding_flywheel_setup#mcp-agent-mail-integration", "supports": "Message identities, file leases and Git-persisted coordination artifacts." }, { "title": "Tokscale leaderboard", "url": "https://tokscale.ai/leaderboard", "supports": "All-time submitted usage snapshot." }, { "title": "MCP Agent Mail source", "url": "https://github.com/Dicklesworthstone/mcp_agent_mail", "supports": "Coordination identities, messaging, file reservations and persistence." }, { "title": "Named Tmux Manager", "url": "https://github.com/Dicklesworthstone/ntm", "supports": "Persistent coding-agent sessions and cockpit." }, { "title": "Coding Agent Session Search", "url": "https://github.com/Dicklesworthstone/coding_agent_session_search", "supports": "Unified history search." } ], "unknowns": [ "Reference workflow assembles documented components; not one observed production trace.", "No universal auto-merge/release controller established.", "No independent project-output attribution for leaderboard tokens.", "January–August evolution not reconstructed." ], "projects": [ { "name": "ACFS", "role": "Environment installer and doctor checks.", "url": "https://github.com/Dicklesworthstone/agentic_coding_flywheel_setup" }, { "name": "NTM", "role": "Named tmux agent cockpit.", "url": "https://github.com/Dicklesworthstone/agentic_coding_flywheel_setup" }, { "name": "MCP Agent Mail", "role": "Messages and advisory file reservations.", "url": "https://github.com/Dicklesworthstone/agentic_coding_flywheel_setup" }, { "name": "BeadsRust / br", "role": "Dependency-aware work records.", "url": "https://github.com/Dicklesworthstone/agentic_coding_flywheel_setup" }, { "name": "Beads Viewer / bv", "role": "Task graph inspection.", "url": "https://github.com/Dicklesworthstone/agentic_coding_flywheel_setup" }, { "name": "CASS", "role": "Cross-agent session search.", "url": "https://github.com/Dicklesworthstone/agentic_coding_flywheel_setup" }, { "name": "CM", "role": "Persistent agent memory.", "url": "https://github.com/Dicklesworthstone/agentic_coding_flywheel_setup" }, { "name": "UBS", "role": "Bug scanning.", "url": "https://github.com/Dicklesworthstone/agentic_coding_flywheel_setup" }, { "name": "DCG", "role": "Destructive-command guard.", "url": "https://github.com/Dicklesworthstone/agentic_coding_flywheel_setup" }, { "name": "SLB", "role": "Risky-action review guardrail.", "url": "https://github.com/Dicklesworthstone/agentic_coding_flywheel_setup" }, { "name": "CAAM", "role": "Account management.", "url": "https://github.com/Dicklesworthstone/agentic_coding_flywheel_setup" }, { "name": "RU", "role": "Repository management.", "url": "https://github.com/Dicklesworthstone/agentic_coding_flywheel_setup" }, { "name": "RCH", "role": "Build execution / capacity input.", "url": "https://github.com/Dicklesworthstone/agentic_coding_flywheel_setup" } ], "checked": "Primary descriptions read; architecture not installed or runtime-tested. Images use fictional work items.", "verdict": "Inspectable infrastructure with defined tool roles; its installed components do not themselves prove unattended production.", "evidence": "Documented architecture", "status": "Current ACFS docs expose a substantial integrated tool inventory and advisory swarm planning. Actual factory throughput and monthly history are not established here.", "orchestration": { "trigger": "Operator selects work and launches sessions.", "controller": "NTM manages sessions; ACFS swarm planning is advisory, not an automatic dispatcher.", "workers": "Installed coding CLIs in persistent tmux sessions.", "memory": "Beads work records, Agent Mail threads/reservations, CASS history and CM memory.", "verification": "Project checks plus UBS and command guardrails; installer doctor checks are a separate concern.", "human": "Provides intent, architecture and review; release policy is project-specific.", "output": "Coordinated coding work and an installed environment.", "failure": "Agents can repair findings; coordination services require operational care. CM/CASS tmux services do not automatically survive reboot." }, "stages": [ { "name": "Select ready work", "owner": "Operator + Beads", "action": "Choose an unblocked task.", "certainty": "inferred" }, { "name": "Start sessions", "owner": "Operator / NTM", "action": "Provide agents with bounded startup context.", "certainty": "documented" }, { "name": "Coordinate", "owner": "Coding agents", "action": "Exchange messages and reserve files; retrieve memory.", "certainty": "documented" }, { "name": "Check changes", "owner": "Agents + tools", "action": "Run project checks and guard dangerous operations.", "certainty": "inferred" }, { "name": "Review delivery", "owner": "Human", "action": "Assess code and evidence before project-specific release.", "certainty": "inferred" } ], "sceneContract": "F-17 ready task → operator/NTM starts workers → Mail and memory support coordination → checks find problems → human reviews; inter-room route is explicitly illustrative.", "example": "F-17 Add export is fictional.", "takeaway": "A shared task graph and coordination state matter; installing a toolkit is not the same as running a factory." }, { "id": "feng-gao", "name": "Feng Gao", "title": "One Person Lab / OPL Flow", "category": "Domain production", "summary": "Durable work records, a supervisor and specialized foundry agents organize long research and software tasks.", "asOf": "2026-09-22", "priority": 6, "image": "images/feng-gao.png", "imageAlt": "One Person Lab / OPL Flow shown as a miniature workplace with owners, evidence and exception paths.", "imageStatus": "generated", "timeline": [ { "date": "2026-09-22", "event": "Current Flow and Framework documentation reviewed; historical release progression not verified.", "source": "https://github.com/gaofeng21cn/opl-flow" } ], "metrics": [ { "label": "All-time submitted tokens", "value": "775.7B", "date": "2026-09-22", "source": "https://tokscale.ai/leaderboard", "caveat": "Public submission snapshot; cache-inclusive; not actual spend, account count or output attributable to these projects." } ], "sources": [ { "title": "OPL Flow", "url": "https://github.com/gaofeng21cn/opl-flow", "supports": "Ledger/Linear/Fleet authority, supervisor, dispatch readback and explicitly unfinished migration capabilities." }, { "title": "OPL Framework", "url": "https://github.com/gaofeng21cn/one-person-lab", "supports": "Staged domain work, review/revision and traceable deliverables." }, { "title": "Feng Gao profile", "url": "https://github.com/gaofeng21cn", "supports": "Domain project inventory." }, { "title": "Tokscale leaderboard", "url": "https://tokscale.ai/leaderboard", "supports": "All-time usage snapshot." } ], "unknowns": [ "No independently validated scientific or clinical outcomes.", "No verified deployed fleet size or unattended completion rate.", "Different domain agents need their own quality criteria; they are not one mandatory pipeline.", "January–August history not established." ], "projects": [ { "name": "One Person Lab Framework", "role": "Shared runtime and contracts.", "url": "https://github.com/gaofeng21cn" }, { "name": "One Person Lab App", "role": "User-facing workbench.", "url": "https://github.com/gaofeng21cn" }, { "name": "OPL Flow", "role": "Codex workflow and coordination.", "url": "https://github.com/gaofeng21cn" }, { "name": "OPL Ledger", "role": "Beads/Dolt work authority.", "url": "https://github.com/gaofeng21cn" }, { "name": "OPL Fleet / Cockpit", "role": "Optional distributed execution and visibility.", "url": "https://github.com/gaofeng21cn" }, { "name": "OPL Instance", "role": "Private owner state and policy.", "url": "https://github.com/gaofeng21cn" }, { "name": "OPL Skills", "role": "Independent capability packages.", "url": "https://github.com/gaofeng21cn" }, { "name": "med-autoscience", "role": "Research foundry.", "url": "https://github.com/gaofeng21cn" }, { "name": "OPL Meta Agent", "role": "Agent-building workflow.", "url": "https://github.com/gaofeng21cn" }, { "name": "Med Auto Grant", "role": "Proposal drafting/review/export.", "url": "https://github.com/gaofeng21cn" }, { "name": "RedCubeAI", "role": "Presentation production.", "url": "https://github.com/gaofeng21cn" }, { "name": "OPL Book Forge", "role": "Book drafting and owner-gated export.", "url": "https://github.com/gaofeng21cn" }, { "name": "HealthPlatform", "role": "Hospital-oriented system.", "url": "https://github.com/gaofeng21cn" }, { "name": "OPL Med Cast", "role": "Research communication media.", "url": "https://github.com/gaofeng21cn" }, { "name": "OPL NetFleet", "role": "Network administration.", "url": "https://github.com/gaofeng21cn" } ], "checked": "Primary descriptions read; architecture not installed or runtime-tested. Images use fictional work items.", "verdict": "Documented control plane and domain-agent ecosystem; demonstrated research quality and deployment scale remain unverified.", "evidence": "Documented architecture", "status": "Current Flow describes optional Ledger, Linear and Fleet layers. It explicitly identifies parts of cross-machine workspace/owner migration as active source work, not current public behavior.", "orchestration": { "trigger": "Owner instruction or authorized due/event trigger.", "controller": "One OPL Flow Supervisor coordinates; native Codex tasks are replaceable executors.", "workers": "Codex and specialist domain agents; optional Fleet dispatch checks resources and leases.", "memory": "Beads/Dolt owns objective and execution state; Git holds code/checkpoints; private Instance owns operational configuration.", "verification": "Outcome readback, artifacts and stage review; an online runner or created task is never completion.", "human": "Defines intent and delivery decisions; waiting-user and blocked remain explicit states.", "output": "Inspectable files, reports and deliverables across domain systems.", "failure": "Classify blocker, retain objective, release unused executor; reconcile uncertain dispatch before retry." }, "stages": [ { "name": "Record objective", "owner": "Owner + Ledger", "action": "Preserve task identity and intent.", "certainty": "documented" }, { "name": "Supervise", "owner": "OPL Flow", "action": "Read state and dispatch bounded work.", "certainty": "documented" }, { "name": "Execute", "owner": "Codex / domain agents", "action": "Work locally or through optional leased capacity.", "certainty": "documented" }, { "name": "Read back", "owner": "Controller + reviewers", "action": "Inspect outcome evidence, not activity alone.", "certainty": "documented" }, { "name": "Decide delivery", "owner": "Owner", "action": "Resolve missing input or authorization and accept delivery.", "certainty": "documented" } ], "sceneContract": "R-17 owner intent → Beads objective → Supervisor dispatch → local/optional leased agents → result evidence → owner delivery; failures retain objective and route to recovery or human.", "example": "R-17 Evidence brief is fictional; not medical advice or a validated research result.", "takeaway": "Keep task identity and evidence durable while allowing the executor to change." }, { "id": "yeongyu-kim", "name": "YeonGyu Kim", "title": "Oh My OpenAgent / OmO", "category": "Harness", "summary": "One main agent coordinates specialist workers while plans, work state and evidence survive interrupted sessions.", "asOf": "2026-09-22", "priority": 7, "image": "images/yeongyu-kim.png", "imageAlt": "Oh My OpenAgent / OmO shown as a miniature workplace with owners, evidence and exception paths.", "imageStatus": "generated", "timeline": [ { "date": "2026-09-22", "event": "Reviewed current three-edition distribution and Senpi orchestration contract; beta remains explicitly labelled.", "source": "https://github.com/code-yeongyu/oh-my-openagent" } ], "metrics": [ { "label": "All-time submitted tokens", "value": "966.8B", "date": "2026-09-22", "source": "https://tokscale.ai/leaderboard", "caveat": "Public submission snapshot; cache-inclusive; not actual spend, account count or output attributable to these projects." } ], "sources": [ { "title": "OmO current README", "url": "https://github.com/code-yeongyu/oh-my-openagent", "supports": "Edition boundaries, component inventory and beta status." }, { "title": "Orchestration guide", "url": "https://raw.githubusercontent.com/code-yeongyu/oh-my-openagent/dev/docs/guide/orchestration.md", "supports": "One orchestrator, worker delegation, persistent state, evidence gates, continuation and delivery controls." }, { "title": "Tokscale leaderboard", "url": "https://tokscale.ai/leaderboard", "supports": "Submitted usage snapshot." } ], "unknowns": [ "No independent success-rate or valuable-output measurement.", "No verified month-by-month January–August timeline.", "Edition-specific behavior must not be generalized across all installations." ], "projects": [ { "name": "Oh My OpenAgent", "role": "Multi-harness orchestration toolkit.", "url": "https://github.com/code-yeongyu/oh-my-openagent" }, { "name": "lazycodex / Light", "role": "Codex installer and portable components.", "url": "https://github.com/code-yeongyu/lazycodex" }, { "name": "Senpi / omo-ai beta", "role": "Standalone runtime and OMO extension.", "url": "https://github.com/code-yeongyu/oh-my-openagent" }, { "name": "Ultrawork Planner", "role": "Interviews and reviewed plan.", "url": "https://github.com/code-yeongyu/oh-my-openagent" }, { "name": "ulw-execute", "role": "Delegated implementation with evidence gates.", "url": "https://github.com/code-yeongyu/oh-my-openagent" }, { "name": "mass-ulw", "role": "Dependency-graph task execution.", "url": "https://github.com/code-yeongyu/oh-my-openagent" }, { "name": "Team Mode", "role": "Communicating parallel members.", "url": "https://github.com/code-yeongyu/oh-my-openagent" }, { "name": "ulw-loop", "role": "Goal execution and reviewer gates.", "url": "https://github.com/code-yeongyu/oh-my-openagent" }, { "name": "Kibitzer", "role": "Read-only memory recall sidecar.", "url": "https://github.com/code-yeongyu/oh-my-openagent" }, { "name": "Jobdori / Dori", "role": "Maintainer-assistant claim and related product waitlist.", "url": "https://github.com/code-yeongyu/oh-my-openagent" } ], "checked": "Primary descriptions read; architecture not installed or runtime-tested. Images use fictional work items.", "verdict": "Detailed inspectable orchestration contracts; no independent production-throughput claim.", "evidence": "Documented architecture", "status": "Current docs distinguish OpenCode Ultimate, Codex Light/lazycodex and standalone Senpi beta. The map explains the documented Senpi planned-execution path, not identical behavior across editions.", "orchestration": { "trigger": "Prompt, ulw, planned execution or dependency-graph mode.", "controller": "Main session agent retains orchestration; category selection determines worker model.", "workers": "Category workers implement/test; curated researchers and independent reviewers have separate roles.", "memory": "Plans, boulder work registry, todos and evidence ledger; Kibitzer sends recall hints.", "verification": "Per-task verification and independent reviewer evidence precede completion.", "human": "Explicit planning approval; PR handoff or ship behavior selected by instruction.", "output": "Verified changes integrated by orchestrator, plus requested PR lifecycle.", "failure": "Repair cited failures, retry graph nodes and resume from durable plan/ledger; continuation has limits." }, "stages": [ { "name": "Agree a plan", "owner": "User + main agent", "action": "Set scope and approve the plan; execution is a separate action.", "certainty": "documented" }, { "name": "Track work", "owner": "Main agent", "action": "Bind plan, work state, todos and worktree.", "certainty": "documented" }, { "name": "Delegate", "owner": "Category workers", "action": "Execute independent units in parallel.", "certainty": "documented" }, { "name": "Verify evidence", "owner": "Independent reviewers", "action": "Check worker claims against evidence.", "certainty": "documented" }, { "name": "Integrate", "owner": "Main agent", "action": "Land verified work and follow requested PR behavior.", "certainty": "documented" } ], "sceneContract": "F-17 plan → main-agent durable registry → workers produce code/tests → independent review → orchestrator integrates; failed evidence returns to workers; shipping requires requested scope.", "example": "F-17 Add export is an invented planning example.", "takeaway": "Persist the objective and its evidence, then treat workers as replaceable execution units." }, { "id": "weiyangzen", "name": "weiyangzen", "title": "b3ehive", "category": "Harness", "summary": "Portable swarm skills generate bounded, evidence-bearing workflows with isolated workers and explicit master acceptance.", "verdict": "A genuine documented orchestration design; deployed output and reliability are not independently verified.", "evidence": "Documented architecture", "status": "September review finds substantially more orchestration detail than the initial profile scan: DAG execution, explicit lifecycle policy, worker identity and request budgets.", "priority": 8, "projects": [ { "name": "b3ehive / compete-cron-builder", "role": "Proposal competition, selection, synthesis and coverage/repair queues.", "url": "https://github.com/weiyangzen/b3ehive" }, { "name": "execution-cron-builder", "role": "Generates repository-specific controllers for blueprint/DAG work and isolated agent claims.", "url": "https://github.com/weiyangzen/b3ehive/tree/main/execution-cron-builder" }, { "name": "learn-cron-builder", "role": "Source-to-target notes, transformation and translation.", "url": "https://github.com/weiyangzen/b3ehive/tree/main/learn-cron-builder" }, { "name": "optimization-cron-builder", "role": "Produces architecture-refinement research and blueprints; implementation is downstream.", "url": "https://github.com/weiyangzen/b3ehive/tree/main/optimization-cron-builder" }, { "name": "looper-cron-builder", "role": "Resource-aware repeated attempts with evidence, budgets and pause/resume rules.", "url": "https://github.com/weiyangzen/b3ehive/tree/main/looper-cron-builder" }, { "name": "mbti-coding-agents", "role": "Agent parameter/persona experiments; separate from b3ehive.", "url": "https://github.com/weiyangzen/mbti-coding-agents" }, { "name": "cc-research-k2p5", "role": "Research tooling repository; internal orchestration not examined here.", "url": "https://github.com/weiyangzen/cc-research-k2p5" }, { "name": "life-evolve-skills", "role": "Evidence-oriented repository maintenance skills.", "url": "https://github.com/weiyangzen/life-evolve-skills" }, { "name": "OpenRateVC", "role": "Public-data venture screening helper; separate domain tool.", "url": "https://github.com/weiyangzen/OpenRateVC" }, { "name": "openfastmd", "role": "Public FastMD release repository; active development is elsewhere.", "url": "https://github.com/weiyangzen/openfastmd" }, { "name": "guizang-ppt-skill-mod", "role": "Presentation skill variant; separate workflow.", "url": "https://github.com/weiyangzen/guizang-ppt-skill-mod" } ], "orchestration": { "trigger": "Operator supplies a bounded question or one authoritative repository blueprint.", "controller": "Generated repo-specific controller admits DAG work under frozen ownership, lifecycle and resource policy.", "workers": "Isolated claims. Codex transport uses a separate tmux/TUI process and private state per claim.", "memory": "Blueprint checklist, durable claim/handoff artifacts, evidence logs and read-only Gantt projection.", "verification": "Worker self-test remains provisional; canonical Master integrates and reruns acceptance gates.", "human": "Chooses scope, lifecycle and resource policy; unresolved policy or reset conditions return to the operator.", "output": "Accepted work and checkpoint commits, with evidence and resolved handoffs.", "failure": "Failed acceptance returns work for repair. Cleanup waits for unresolved work; controller enforces request and worker caps." }, "stages": [ { "name": "Freeze the blueprint", "owner": "Operator + controller", "action": "Set DAG, owned paths, gates and budgets.", "certainty": "documented" }, { "name": "Admit a claim", "owner": "Controller", "action": "Assign ready work within explicit caps.", "certainty": "documented" }, { "name": "Work in isolation", "owner": "Agent worker", "action": "Produce a self-tested durable handoff.", "certainty": "documented" }, { "name": "Integrate and validate", "owner": "Canonical Master", "action": "Run acceptance checks; only this role marks accepted.", "certainty": "documented" }, { "name": "Checkpoint or repair", "owner": "Controller + Master", "action": "Keep failed work open; reconcile before cleanup.", "certainty": "documented" } ], "timeline": [ { "date": "2026-09-22", "event": "Direct repository review establishes five swarm skills and a master-only acceptance boundary; observation date, not project launch.", "source": "https://github.com/weiyangzen/b3ehive" } ], "metrics": [ { "label": "All-time submitted tokens", "value": "944.7B · leaderboard #9", "date": "2026-09-22", "source": "https://tokscale.ai/leaderboard", "caveat": "Self-submitted usage snapshot; includes cached input. Not project-attributable output, monthly burn, money spent or account count." } ], "sources": [ { "title": "b3ehive", "url": "https://github.com/weiyangzen/b3ehive", "supports": "Five skills and dual-cursor checklist." }, { "title": "Core concepts", "url": "https://raw.githubusercontent.com/weiyangzen/b3ehive/main/docs/concepts.md", "supports": "Blueprint authority, workflow outputs, inspectable evidence." }, { "title": "Execution skill", "url": "https://raw.githubusercontent.com/weiyangzen/b3ehive/main/execution-cron-builder/SKILL.md", "supports": "Controller generation contract, identity, lifecycle, acceptance, caps." }, { "title": "Profile", "url": "https://github.com/weiyangzen", "supports": "Separate project inventory." } ], "unknowns": [ "Measured success rates and deployments of generated controllers.", "Which lifecycle is actually used in the author's own current factory.", "Nine-month release trajectory and which tools account for the submitted tokens." ], "example": "Illustrative TASK-01: add a search filter. Its checkbox is unfinished, then self-tested, then accepted only after Master integration.", "sceneContract": "Freeze → operator/controller → TASK-01 blueprint → set dependencies/gates → ready item; admit → controller → item → lease capacity and isolate claim → worker; work → worker → claim → self-test and durable handoff → provisional [_]; accept → Master → handoff → integrate and rerun gates → [x] or repair; checkpoint → controller → accepted state → reconcile and clean. Human reset desk at failed policy boundary.", "takeaway": "Worker completion is provisional; only integration plus acceptance makes the work done.", "checked": "Read README, concepts and the execution skill as research evidence, not as instructions to run. No controller was installed or executed.", "imageAlt": "b3ehive workshop with a blueprint desk, bounded claim gate, isolated agent bay and master-only acceptance desk showing unfinished, self-tested and accepted states.", "asOf": "2026-09-22", "image": "images/weiyangzen.png", "imageStatus": "generated" }, { "id": "vincent-koc", "name": "Vincent Koc / OpenClaw maintainers", "title": "Clawpatch and Clownfish", "category": "Verification", "summary": "Explicit finding repair and conservative issue cleanup provide two inspectable maintenance workflows.", "asOf": "2026-09-22", "priority": 9, "image": "images/vincent-koc.png", "imageAlt": "Clawpatch and Clownfish shown as a miniature workplace with owners, evidence and exception paths.", "imageStatus": "generated", "timeline": [ { "date": "2026-06-16", "event": "Clownfish dashboard reports reviewed clusters and still flags failures requiring inspection; it is dated operational evidence, not a current success guarantee.", "source": "https://github.com/openclaw/clownfish" }, { "date": "2026-09-22", "event": "Current Clawpatch and Clownfish contracts reviewed; Clawbench link now redirects to Shellbench.", "source": "https://github.com/openclaw/clawpatch" } ], "metrics": [ { "label": "All-time submitted tokens", "value": "624.9B", "date": "2026-09-22", "source": "https://tokscale.ai/leaderboard", "caveat": "Public submission snapshot; cache-inclusive; not actual spend, account count or output attributable to these projects." } ], "sources": [ { "title": "Clawpatch README", "url": "https://github.com/openclaw/clawpatch", "supports": "Feature mapping, persistent findings, explicit repair/revalidation/open-PR; no merge." }, { "title": "Clownfish README", "url": "https://github.com/openclaw/clownfish", "supports": "Proposal-first cluster cleanup, promotion, fresh-state checks and dated dashboard." }, { "title": "Vincent project inventory", "url": "https://github.com/vincentkoc", "supports": "Published, maintained or contributed projects; not sole-authorship attribution." }, { "title": "Shellbench", "url": "https://github.com/openclaw/shellbench", "supports": "Current destination of Clawbench link." }, { "title": "Tokscale leaderboard", "url": "https://tokscale.ai/leaderboard", "supports": "Usage snapshot." } ], "unknowns": [ "No evidence that all listed tools form one connected production pipeline.", "Clownfish dashboard freshness ends in June for the reviewed evidence.", "No token-to-output attribution or measured unattended factory rate." ], "projects": [ { "name": "Clawpatch", "role": "Finding-based review and repair.", "url": "https://github.com/openclaw/clawpatch" }, { "name": "Clownfish", "role": "Cluster cleanup.", "url": "https://github.com/openclaw/clownfish" }, { "name": "Clawsweeper", "role": "Backlog scanning.", "url": "https://github.com/vincentkoc" }, { "name": "Gitcrawl", "role": "Local issue/PR indexing.", "url": "https://github.com/vincentkoc" }, { "name": "Clawbench → Shellbench", "role": "Trace-based agent benchmark.", "url": "https://github.com/openclaw/shellbench" }, { "name": "Crabbox", "role": "Remote targeted tests.", "url": "https://github.com/vincentkoc" }, { "name": "Crabpot", "role": "Plugin testbed.", "url": "https://github.com/vincentkoc" }, { "name": "Plugin Inspector", "role": "Compatibility checks.", "url": "https://github.com/vincentkoc" }, { "name": "Kitchen Sink", "role": "Plugin fixtures.", "url": "https://github.com/vincentkoc" }, { "name": "Tokenjuice", "role": "Terminal output compaction.", "url": "https://github.com/vincentkoc" }, { "name": "Clawrouter", "role": "Provider gateway.", "url": "https://github.com/vincentkoc" }, { "name": "ReleaseBar", "role": "Release dashboard.", "url": "https://github.com/vincentkoc" }, { "name": "Agentrinse", "role": "Cleanup planning.", "url": "https://github.com/vincentkoc" }, { "name": "Libterminal", "role": "Terminal primitives.", "url": "https://github.com/vincentkoc" }, { "name": "Tideglass", "role": "Portable intent.", "url": "https://github.com/vincentkoc" }, { "name": "Graincrawl / Crawlbar / Crawlkit", "role": "Local archive tools.", "url": "https://github.com/vincentkoc" }, { "name": "Slacrawl / Discrawl / Notcrawl", "role": "Local message/document archives.", "url": "https://github.com/vincentkoc" }, { "name": "ACPX", "role": "Stateful agent sessions.", "url": "https://github.com/vincentkoc" }, { "name": "Openamnesia", "role": "Memory backfill.", "url": "https://github.com/vincentkoc" }, { "name": "Opik / Opik Optimizer", "role": "Evaluation and optimization; maintained/contributed.", "url": "https://github.com/vincentkoc" }, { "name": "Opik OpenClaw / Opik MCP", "role": "Observability integrations.", "url": "https://github.com/vincentkoc" }, { "name": "GEPA", "role": "Optimization contribution.", "url": "https://github.com/vincentkoc" }, { "name": "TinyQA", "role": "Evaluation smoke tests.", "url": "https://github.com/vincentkoc" }, { "name": "NemoClaw / OpenClaw", "role": "Security/runtime contributions.", "url": "https://github.com/vincentkoc" }, { "name": ".skills / awesome-openclaw", "role": "Skills and references.", "url": "https://github.com/vincentkoc" } ], "checked": "Primary descriptions read; architecture not installed or runtime-tested. Images use fictional work items.", "verdict": "Concrete documented maintenance machinery; separate tools must not be presented as one proven autonomous factory or solely authored by Vincent.", "evidence": "Documented architecture", "status": "Clawpatch separates review, repair and PR creation and never merges. Clownfish defaults to proposals and checks current GitHub state before promoted actions. Its visible operations dashboard is dated June 16, not September.", "orchestration": { "trigger": "Maintainer selects a repository/finding or curated issue cluster.", "controller": "Clawpatch CLI manages feature/finding state; Clownfish separately validates proposed issue actions.", "workers": "Installed coding harnesses; Clownfish uses a Codex classification worker.", "memory": "Clawpatch persists findings/attempts/history under .clawpatch; Clownfish retains job proposals and action reports.", "verification": "Revalidate current code; cleanup applicator rechecks live GitHub state.", "human": "Explicit repair/publication commands; Clownfish jobs require promotion. Clawpatch never merges.", "output": "Evidence-backed findings and PRs; separately, narrowly authorized cleanup receipts.", "failure": "Dirty worktree or failed checks block repair/publication; uncertain issues need humans; security reports leave ordinary cleanup." }, "stages": [ { "name": "Map features", "owner": "Clawpatch", "action": "Build durable review slices.", "certainty": "documented" }, { "name": "Review", "owner": "Coding agents", "action": "Run bounded parallel reviews.", "certainty": "documented" }, { "name": "Choose repair", "owner": "Maintainer + agent", "action": "Apply one explicit finding patch.", "certainty": "documented" }, { "name": "Revalidate", "owner": "Agent + checks", "action": "Confirm the finding against current code.", "certainty": "documented" }, { "name": "Open PR", "owner": "Maintainer command", "action": "Publish a recorded patch; merging remains outside Clawpatch.", "certainty": "documented" } ], "detailImage": "images/vincent-koc-detail.png", "detailImageAlt": "Clownfish cluster cleanup with proposal, human promotion, live-state checks and escalation.", "sceneContract": "F-17 feature → agent review → maintainer-selected patch → revalidation → explicit PR publication, never merge. Separate C-17 cleanup: cluster → classify → propose → promote → fresh-state action.", "example": "F-17 Export bug and C-17 issue cluster are fictional.", "takeaway": "Separate evidence, permission and mutation; each is a different step." }, { "id": "geoffrey-huntley", "name": "Geoffrey Huntley", "title": "Loom, Ralph and Latent Patterns", "category": "Factory", "summary": "Experiments span repeat-until-done coding loops, a custom agent engine and development embedded inside the product.", "asOf": "2026-09-22", "priority": 10, "image": "images/geoffrey-huntley.png", "imageAlt": "Loom experimental engine: a core state machine coordinates provider calls, tools and persisted threads; production release remains unknown.", "imageStatus": "generated", "timeline": [ { "date": "2025-07-14", "event": "Historical foundation, outside the nine-month window: Ralph article describes a single-process loop, one task per iteration and tests as backpressure.", "source": "https://ghuntley.com/ralph/" }, { "date": "2026-03-09", "event": "Latent Patterns demonstration: edit inside designer mode and launch Cursor Cloud Agents/Workflow Automations using a risk-based shipping approach.", "source": "https://ghuntley.com/rad/" }, { "date": "2026-09-22", "event": "Reviewed Loom README still labels it experimental, unstable and proprietary.", "source": "https://github.com/ghuntley/loom" } ], "metrics": [], "sources": [ { "title": "Loom repository", "url": "https://github.com/ghuntley/loom", "supports": "State machine, provider proxy, tools, thread persistence, optional Weaver and explicit limitations." }, { "title": "Embedded factory demonstration", "url": "https://ghuntley.com/rad/", "supports": "March 9 product-as-IDE and Cursor Cloud Agent path." }, { "title": "Ralph technique", "url": "https://ghuntley.com/ralph/", "supports": "Repeated coding-agent loop; separate predecessor context." } ], "unknowns": [ "No verified September Latent Patterns operating metrics.", "Risk thresholds, failure recovery and production approval rules undisclosed.", "No proof Loom executes the shown Latent Patterns workflow.", "Do not treat visible Loom source as an open-source license." ], "projects": [ { "name": "Loom", "role": "Rust coding engine with state machine and tools.", "url": "https://github.com/ghuntley/loom" }, { "name": "Weaver", "role": "Loom remote execution via Kubernetes.", "url": "https://github.com/ghuntley/loom" }, { "name": "Ralph", "role": "Repeated coding-agent loop technique.", "url": "https://ghuntley.com/ralph/" }, { "name": "Latent Patterns", "role": "Product with embedded designer/agent workflow.", "url": "https://ghuntley.com/rad/" } ], "checked": "Primary descriptions read; architecture not installed or runtime-tested. Images use fictional work items.", "verdict": "Inspectable experimental agent engine and a dated product demonstration; these are distinct mechanisms, not a verified common runtime.", "evidence": "Reported operation", "status": "Loom currently warns that it is experimental and unstable, under proprietary licensing. March Latent Patterns demo used Cursor Cloud Agents; September production operation is not established.", "orchestration": { "trigger": "Loom CLI/REPL prompt; separately, a Latent Patterns designer-mode change.", "controller": "Loom core state machine orchestrates tool/model calls. Latent Patterns launches a Cursor Cloud Agent.", "workers": "Loom tool registry, optional Weaver environments; separate Cursor workflow workers.", "memory": "Loom thread persistence and FTS5 search; Latent Patterns state details undisclosed.", "verification": "Loom includes build/test/lint tooling; product shipment described as risk-based, without a disclosed complete gating contract.", "human": "Sets task or edits design intent; exact production approval thresholds unknown.", "output": "Tool-run results/code; separate product demo reports production updates.", "failure": "Loom describes retries and logging; end-to-end release/recovery policies not established." }, "stages": [ { "name": "Enter request", "owner": "Human", "action": "Pass a development task to Loom.", "certainty": "documented" }, { "name": "Drive loop", "owner": "Loom core", "action": "Select model/tool interactions.", "certainty": "documented" }, { "name": "Execute tools", "owner": "Tool registry / optional Weaver", "action": "Return filesystem/code-analysis results.", "certainty": "documented" }, { "name": "Persist thread", "owner": "Thread system", "action": "Retain searchable conversation state.", "certainty": "documented" }, { "name": "Release boundary", "owner": "Unknown", "action": "No production-release contract established.", "certainty": "unknown" } ], "detailImage": "images/geoffrey-huntley-detail.png", "detailImageAlt": "Latent Patterns designer mode launches a cloud agent, with undisclosed risk and recovery boundaries.", "sceneContract": "L-17 CLI request → Loom state machine ↔ proxy/tools → persisted thread → unknown release boundary. Separate P-17 designer change → launch Cursor agent → reported risk-based shipment.", "example": "L-17 Inspect export bug and P-17 Improve a lesson page are illustrative.", "takeaway": "A product can expose its own development interface; reliability still depends on the undisclosed gates.", "images": [ { "path": "images/geoffrey-huntley.png", "title": "Loom: experimental engine", "alt": "Loom experimental engine: a core state machine coordinates provider calls, tools and persisted threads; production release remains unknown.", "status": "generated" }, { "path": "images/geoffrey-huntley-detail.png", "title": "Latent Patterns: embedded factory", "alt": "Latent Patterns designer mode launches a cloud agent, with undisclosed risk and recovery boundaries.", "status": "generated" }, { "path": "images/geoffrey-huntley-ralph.png", "title": "Ralph: historical loop technique", "alt": "One task per iteration, with plan context, tests and operator tuning.", "status": "generated" } ] }, { "id": "factory-ai", "name": "Matan Grinberg, Eno Reyes / Factory", "title": "Factory Missions", "category": "Factory", "summary": "A commercial orchestration system separates planning, implementation and fresh validation agents.", "verdict": "Documented end-to-end mission architecture; throughput examples are vendor reports.", "evidence": "Documented architecture", "status": "Missions is documented as available. September vendor update describes enterprise expansion and model routing; independent production reliability is unverified.", "priority": 11, "projects": [ { "name": "Droid", "role": "The coding agent and shared runtime exposed through CLI and app.", "url": "https://docs.factory.ai/droid-cli/overview" }, { "name": "Missions", "role": "Plans, feature queue, milestone coordination and validation.", "url": "https://docs.factory.ai/missions/overview" }, { "name": "Droid Exec", "role": "Headless entry for scripts and CI; not identical to the mission orchestrator.", "url": "https://docs.factory.ai/droid-cli/overview" }, { "name": "Custom Droids, Skills, Hooks, MCP", "role": "Scoped delegation, reusable procedures, lifecycle automation and connected tools.", "url": "https://docs.factory.ai/droid-cli/overview" } ], "orchestration": { "trigger": "User goal, refined with the orchestrator; user approves the mission plan.", "controller": "Orchestrator maintains a feature plan; programmatic runner dispatches fresh workers by feature.", "workers": "Feature implementers; separate scrutiny and black-box user-testing validators.", "memory": "Validation contract, feature list, shared instructions, services and accumulated knowledge.", "verification": "Milestone validators report gaps; orchestrator creates fix features and revalidates.", "human": "Approves plan, monitors, intervenes; blocked work returns to the user.", "output": "Implemented features satisfying a behavioral contract.", "failure": "Pause and return to user when implementation or validation is blocked." }, "stages": [ { "name": "Define success", "owner": "User + orchestrator", "action": "Approve the goal and behavioral contract.", "certainty": "documented" }, { "name": "Queue features", "owner": "Orchestrator", "action": "Group bounded features into milestones.", "certainty": "documented" }, { "name": "Implement features", "owner": "Fresh worker", "action": "Write tests and code for one assigned feature.", "certainty": "documented" }, { "name": "Validate behavior", "owner": "Fresh validators", "action": "Inspect implementation and exercise user flows.", "certainty": "documented" }, { "name": "Repair or finish", "owner": "Orchestrator", "action": "Queue fixes, repeat gates, or complete the milestone.", "certainty": "documented" } ], "timeline": [ { "date": "2026-04-10", "event": "Missions architecture published: distinct roles, external state and milestone repair loop.", "source": "https://factory.com/news/missions-architecture" }, { "date": "2026-07-14", "event": "Release notes introduce a unified mission side panel.", "source": "https://docs.factory.ai/changelog/release-notes" }, { "date": "2026-09-15", "event": "Vendor update describes enterprise growth, routing and effectiveness measurement.", "source": "https://factory.ai/news/5-billion-valuation" } ], "metrics": [ { "label": "Published mission example", "value": "16.5 hours; 185 agent runs", "date": "2026-04-10", "source": "https://factory.com/news/missions-architecture", "caveat": "Vendor-reported Slack-clone example, not a general throughput benchmark; publication may have been updated." } ], "sources": [ { "title": "How Missions Work", "url": "https://factory.com/news/missions-architecture", "supports": "April 10 architecture and example; fetched September 22." }, { "title": "Missions documentation", "url": "https://docs.factory.ai/missions/overview", "supports": "Current approval, monitoring, QA prerequisites and recovery." }, { "title": "Droid CLI", "url": "https://docs.factory.ai/droid-cli/overview", "supports": "Surfaces and extensions." }, { "title": "Release notes", "url": "https://docs.factory.ai/changelog/release-notes", "supports": "Dated July mission UI change." }, { "title": "September company update", "url": "https://factory.ai/news/5-billion-valuation", "supports": "September 15 vendor status; not independent validation." } ], "unknowns": [ "Independently measured success rates on production repositories.", "How September model routing changes worker selection in each customer deployment.", "Public evidence does not establish a universal account or token-burn record." ], "example": "Illustrative LOGIN-01: valid credentials should open the dashboard. This is a teaching example, not a claim about a customer deployment.", "sceneContract": "Approve → user/orchestrator → LOGIN-01 goal → define observable success → contract → queue; queue → orchestrator → contract → make feature → worker; build → fresh worker → feature → tests/code → milestone; validate → fresh validators → app/contract → check login → pass or issue; repair → orchestrator → issue → fix feature → worker; blockage → human.", "takeaway": "The worker that builds a feature does not get the final vote on correctness.", "checked": "Read primary architecture and current Missions/CLI docs, plus dated release notes. No runtime test.", "imageAlt": "Isometric Factory Missions workshop with approved contract, feature queue, fresh worker, independent validator and repair loop.", "asOf": "2026-09-22", "image": "images/factory-ai.png", "imageStatus": "generated" }, { "asOf": "2026-09-22", "priority": 12, "imageStatus": "generated", "id": "theo", "name": "Theo Browne & T3 team", "title": "T3 Code", "category": "Harness", "summary": "One interface directs several coding harnesses while execution and credentials stay with the workspace machine.", "verdict": "A source-available control surface with agent-assisted shipping evidence; unattended production is not established.", "evidence": "Documented architecture", "status": "September source describes remote, desktop and mobile control; documentation still warns the product is early.", "projects": [ { "name": "T3 Code", "role": "Agent control interface and workspace server.", "url": "https://github.com/pingdotgg/t3code" }, { "name": "Provider adapters", "role": "Normalize commands and events from supported harnesses.", "url": "https://github.com/pingdotgg/t3code/blob/main/docs/internals/overview.md" }, { "name": "Source-control integrations", "role": "Repository publication, PR creation and review actions.", "url": "https://github.com/pingdotgg/t3code/blob/main/docs/user/source-control.md" }, { "name": "T3 Connect", "role": "Remote access reported generally available in August.", "url": "https://x.com/theo/status/2085639979011891445" } ], "orchestration": { "trigger": "Human starts or continues a task.", "controller": "T3 server records intent; reactors perform side effects through provider adapters.", "workers": "Installed coding harnesses, including Codex and Claude Code.", "memory": "Durable event log, projections, provider sessions and Git checkpoints.", "verification": "Underlying harness checks plus reviewable changes; factory-wide quality policy is not established.", "human": "Directs work, handles attention requests and reviews/merges changes.", "output": "Workspace edits and PRs.", "failure": "Recorded state supports replay; rollback coordinates checkpoint and conversation." }, "stages": [ { "name": "Choose work", "owner": "Human", "action": "Give a task through phone, web or desktop.", "certainty": "documented" }, { "name": "Record intent", "owner": "T3 server", "action": "Commit commands before side effects.", "certainty": "documented" }, { "name": "Run provider", "owner": "Adapter + harness", "action": "Operate in the workspace environment.", "certainty": "documented" }, { "name": "Inspect", "owner": "Human", "action": "Read diffs, logs and checkpoints.", "certainty": "documented" }, { "name": "Review and merge", "owner": "Human + repository checks", "action": "Use source-control actions.", "certainty": "documented" } ], "timeline": [ { "date": "2026-07-24", "event": "Theo reports 76 PRs merged since Monday.", "source": "https://x.com/theo/status/2080776679035896166" }, { "date": "2026-08-07", "event": "Reports 250+ PRs in two weeks and T3 Connect GA.", "source": "https://x.com/theo/status/2085639979011891445" }, { "date": "2026-09-22", "event": "Current source documents server-owned execution and persisted orchestration intent.", "source": "https://github.com/pingdotgg/t3code/blob/main/docs/internals/overview.md" } ], "metrics": [ { "label": "Reported merged PRs", "value": "250+ / two weeks", "date": "2026-08-07", "source": "https://x.com/theo/status/2085639979011891445", "caveat": "Shipping volume does not establish autonomous contribution or quality." }, { "label": "Claimed token value", "value": ">$200k / prior month", "date": "2026-07-10", "source": "https://x.com/theo/status/2075439348267577790", "caveat": "Truncated saved post; not proven cash spend or token count." } ], "sources": [ { "title": "Repository", "url": "https://github.com/pingdotgg/t3code", "supports": "Supported harnesses and app surfaces." }, { "title": "Architecture", "url": "https://github.com/pingdotgg/t3code/blob/main/docs/internals/overview.md", "supports": "Execution boundary, event log, reactors and checkpoints." }, { "title": "Source control", "url": "https://github.com/pingdotgg/t3code/blob/main/docs/user/source-control.md", "supports": "PR/review workflow." }, { "title": "August shipping report", "url": "https://x.com/theo/status/2085639979011891445", "supports": "Saved first-person tweet in events-x.jsonl:13043." } ], "unknowns": [ "How much T3 development is unattended.", "A centralized autonomous scheduler or measured factory completion rate.", "Actual token spend and subscription count." ], "image": "images/theo.png", "imageAlt": "T3 human task control, persisted server intent, provider execution and human review.", "sceneContract": "Human task → persisted intent → provider adapter → workspace changes → review/merge; attention returns to human.", "example": "T3-01 mobile-sidebar fix; illustrative.", "takeaway": "A good control surface can coordinate agents without being an autonomous factory.", "checked": "Read repository and architecture/source-control docs; shipping claims checked against the local tweet archive." }, { "asOf": "2026-09-22", "priority": 13, "imageStatus": "generated", "id": "daniel-miessler", "name": "Daniel Miessler", "title": "LifeOS / PAI & media workflow", "category": "Harness", "summary": "Personal intent, memory and reusable skills guide work; a separate video workflow is reported to publish end to end.", "verdict": "Inspectable general-purpose harness; zero-touch media production is a first-person claim with partial internals.", "evidence": "Documented architecture", "status": "September source exposes LifeOS 7.x; media workflow's latest detailed evidence is August 28.", "projects": [ { "name": "LifeOS / PAI", "role": "Goal-aware harness with skills and persistent context.", "url": "https://github.com/danielmiessler/LifeOS" }, { "name": "Algorithm + ISA", "role": "Define testable outcomes and close claims with evidence.", "url": "https://ourlifeos.ai/" }, { "name": "Cortex + TELOS", "role": "Memory and personal goals.", "url": "https://ourlifeos.ai/" }, { "name": "Router + skills", "role": "Choose specialist work and model effort.", "url": "https://ourlifeos.ai/" }, { "name": "Pulse / Atlas / Ledger", "role": "Operations visibility, asset relationships and change tracking.", "url": "https://ourlifeos.ai/" }, { "name": "Arbol / Bunker", "role": "Composable execution and shared application support.", "url": "https://ourlifeos.ai/" }, { "name": "Video publishing workflow", "role": "Descript/Underlord editing, Remotion visuals, packaging and YouTube publication.", "url": "https://x.com/DanielMiessler/status/2093450595520966777" } ], "orchestration": { "trigger": "A person's intent; recording input for the separate media workflow.", "controller": "LifeOS Algorithm and router on a capable harness; media scheduler undisclosed.", "workers": "Selected skills and tools; media names Descript/Underlord and Remotion.", "memory": "Goals, context and Cortex; outcomes recorded against an ISA.", "verification": "Tool evidence closes claims. Media-specific checks are not disclosed.", "human": "Supplies goals, context and authority; media is claimed zero-touch after input.", "output": "Task artifacts; separately, published videos, clips and social drafts.", "failure": "Unproven claims remain open; media recovery and retries unknown." }, "stages": [ { "name": "Articulate", "owner": "Person + assistant", "action": "Record a checkable desired state.", "certainty": "documented" }, { "name": "Route", "owner": "LifeOS router", "action": "Choose skills and effort using context.", "certainty": "documented" }, { "name": "Act", "owner": "Harness + skills", "action": "Take a concrete step.", "certainty": "documented" }, { "name": "Verify", "owner": "Tools + Algorithm", "action": "Require evidence against the claim.", "certainty": "documented" }, { "name": "Learn", "owner": "Memory", "action": "Carry outcomes into future work.", "certainty": "documented" } ], "timeline": [ { "date": "2026-01-23", "event": "PAI v2.4 introduces the Algorithm.", "source": "https://github.com/danielmiessler/LifeOS" }, { "date": "2026-04-30", "event": "v5 adds Life Operating System structure.", "source": "https://github.com/danielmiessler/LifeOS" }, { "date": "2026-07-02", "event": "v6 adopts LifeOS name and one-skill distribution.", "source": "https://github.com/danielmiessler/LifeOS" }, { "date": "2026-08-28", "event": "Claims an operational zero-touch video publishing workflow.", "source": "https://x.com/DanielMiessler/status/2093450595520966777" }, { "date": "2026-09-22", "event": "Current website presents the articulate/climb/verify/learn loop.", "source": "https://ourlifeos.ai/" } ], "metrics": [], "sources": [ { "title": "LifeOS repository", "url": "https://github.com/danielmiessler/LifeOS", "supports": "Release history, architecture and personal context." }, { "title": "LifeOS current design", "url": "https://ourlifeos.ai/", "supports": "Current loop and subsystem roles." }, { "title": "Media workflow report", "url": "https://x.com/DanielMiessler/status/2093450595520966777", "supports": "Saved first-person August 28 tweet at events-x.jsonl:26880." } ], "unknowns": [ "Measured software-factory throughput.", "Media pipeline source, validation, retry and approval mechanics.", "September media reliability and publication volume." ], "image": "images/daniel-miessler.png", "imageAlt": "LifeOS context, routing, action, evidence and learning loop.", "detailImage": "images/daniel-miessler-detail.png", "detailImageAlt": "Reported video editing and publishing capabilities with an explicit unknown recovery boundary.", "sceneContract": "Intent → checkable ISA → routed skills → artifact → evidence → memory. Separate media map: recording → edit → visual/package → publish; recovery unknown.", "example": "LO-01 personal dashboard and VID-01 recorded episode; illustrative.", "takeaway": "Make intent and proof persistent; distinguish a general harness from a particular production pipeline.", "checked": "Read current repository and website; checked media capabilities against saved first-person text." }, { "asOf": "2026-09-22", "priority": 14, "imageStatus": "generated", "id": "danny-postma", "name": "Danny Postma", "title": "Private game-development pipeline", "category": "Factory", "summary": "A morning specification enters a background coding pipeline; Danny reviews and merges the result later.", "verdict": "Concrete first-person operating workflow, with private controller and verification details.", "evidence": "Partial evidence", "status": "Latest detailed workflow evidence is August 3; September scale and reliability are not established.", "projects": [ { "name": "Game-development factory", "role": "Spec-to-build daily loop; game title unestablished.", "url": "https://x.com/dannypostma/status/2084251032712872182" }, { "name": "Agent inbox", "role": "MCP messaging tied to agent-session context.", "url": "https://x.com/dannypostma/status/2082275703710679344" }, { "name": "Private harness", "role": "Implementation pipeline using loops/templates.", "url": "https://x.com/dannypostma/status/2086417507192291769" } ], "orchestration": { "trigger": "Danny writes a game-feature specification.", "controller": "Private harness; routing and scheduler details not established.", "workers": "Coding agents; model mix and count unestablished.", "memory": "Inbox replies feed session context; durable task-state design unknown.", "verification": "Danny reviews and adjusts; automated tests not established.", "human": "Writes specification, responds asynchronously and merges.", "output": "Merged game features, including a reported inventory system.", "failure": "Adjustments return to implementation; unattended failure recovery unknown." }, "stages": [ { "name": "Specify", "owner": "Danny", "action": "Define a feature in the morning.", "certainty": "reported" }, { "name": "Queue", "owner": "Private harness", "action": "Accept the work; internal routing unknown.", "certainty": "reported" }, { "name": "Build", "owner": "Coding agents", "action": "Work while Danny focuses elsewhere.", "certainty": "reported" }, { "name": "Communicate", "owner": "Inbox + Danny", "action": "Exchange messages tied to the session.", "certainty": "reported" }, { "name": "Review and merge", "owner": "Danny", "action": "Inspect and adjust the result.", "certainty": "reported" } ], "timeline": [ { "date": "2026-07-29", "event": "Describes an MCP inbox for asynchronous agent communication.", "source": "https://x.com/dannypostma/status/2082275703710679344" }, { "date": "2026-08-03", "event": "Describes the daily spec/build/review loop and inventory feature.", "source": "https://x.com/dannypostma/status/2084251032712872182" }, { "date": "2026-08", "event": "Advocates a dedicated harness with loops and templates.", "source": "https://x.com/dannypostma/status/2086417507192291769" } ], "metrics": [], "sources": [ { "title": "Daily game pipeline", "url": "https://x.com/dannypostma/status/2084251032712872182", "supports": "Saved first-person tweet at events-x.jsonl:10530." }, { "title": "Agent inbox", "url": "https://x.com/dannypostma/status/2082275703710679344", "supports": "Saved first-person tweet at events-x.jsonl:8569; ending truncated." }, { "title": "Harness and templates", "url": "https://x.com/dannypostma/status/2086417507192291769", "supports": "Saved first-person tweet at events-x.jsonl:13524." } ], "unknowns": [ "Game identity, repository and controller implementation.", "Automated verification and recovery.", "Agent count, account inventory and current September operation." ], "image": "images/danny-postma.png", "imageAlt": "Danny's morning feature spec, private background agents, async inbox and evening human review.", "sceneContract": "Morning spec → private pipeline → coding → asynchronous messages → human review/adjust/merge; private routing drawn dashed.", "example": "DP-01 inventory feature; illustrative reconstruction of the reported feature type.", "takeaway": "An inbox makes agent work asynchronous; the public evidence still ends at a human merge decision.", "checked": "Read local archived first-person posts and their preserved tweet IDs; no private pipeline inspected." }, { "id": "max-ghenis", "name": "Max Ghenis / PolicyEngine", "title": "PolicyEngine and personal tools", "category": "Domain production", "summary": "An open-source policy simulation organization with a separate portfolio of practical personal software.", "verdict": "Concrete domain software and contributor checks are public; an unattended software factory is not established.", "evidence": "Partial evidence", "status": "September snapshot shows PolicyEngine country models and active public tools. Nine-month factory progression cannot be reconstructed from the reviewed sources.", "priority": 15, "projects": [ { "name": "PolicyEngine Core", "role": "Simulation engine, parameter system and reform machinery, derived from OpenFisca.", "url": "https://github.com/PolicyEngine/policyengine-core" }, { "name": "PolicyEngine US", "role": "US rules and microdata for tax-benefit modelling.", "url": "https://github.com/PolicyEngine/policyengine-us" }, { "name": "OpenMessage", "role": "Local Google Messages and WhatsApp inbox with Mac/web/MCP interfaces.", "url": "https://github.com/MaxGhenis/openmessage" }, { "name": "rambar", "role": "Mac RAM dashboard with app and process breakdowns.", "url": "https://github.com/MaxGhenis/rambar" }, { "name": "scrollywood", "role": "Smooth scrolling video capture extension.", "url": "https://github.com/MaxGhenis/scrollywood" }, { "name": "mystquarto", "role": "MyST and Quarto document conversion.", "url": "https://github.com/MaxGhenis/mystquarto" }, { "name": "usage-data", "role": "Daily coding-tool usage aggregates.", "url": "https://github.com/MaxGhenis/usage-data" } ], "orchestration": { "trigger": "Product use: an analyst supplies household inputs and a policy scenario.", "controller": "The documented runtime is a simulation engine. Coding-agent scheduler: Not established in reviewed sources.", "workers": "Country-model formulas and parameter calculations; these are software components, not asserted autonomous agents.", "memory": "Country rules, parameter tree, simulation inputs and microdata.", "verification": "Contributor docs describe unit and integration tests, source references, formatting and PR workflow.", "human": "Analyst interprets results; contributors maintain models. Agent supervision details remain unknown.", "output": "Computed household or population policy effects; no numerical result is asserted here.", "failure": "Bad inputs or model defects need correction. Autonomous recovery: Not established in reviewed sources." }, "stages": [ { "name": "Choose a scenario", "owner": "Analyst", "action": "Select a country model, household and reform.", "certainty": "documented" }, { "name": "Load rules", "owner": "Country model", "action": "Combine formulas and parameters.", "certainty": "documented" }, { "name": "Calculate effects", "owner": "PolicyEngine Core", "action": "Evaluate the simulation.", "certainty": "documented" }, { "name": "Inspect results", "owner": "Analyst", "action": "Interpret model outputs; review assumptions.", "certainty": "documented" } ], "timeline": [ { "date": "2026-09-22", "event": "Research snapshot confirms public core/country packages and personal tool portfolio; this is an observation date, not a launch.", "source": "https://github.com/MaxGhenis" } ], "metrics": [ { "label": "All-time submitted tokens", "value": "664.1B · leaderboard #14", "date": "2026-09-22", "source": "https://tokscale.ai/leaderboard", "caveat": "Self-submitted usage snapshot; includes cached input. Not project-attributable output, monthly burn, money spent or account count." } ], "sources": [ { "title": "Max Ghenis profile", "url": "https://github.com/MaxGhenis", "supports": "Organization role and named public tools; checked September 22." }, { "title": "PolicyEngine Core", "url": "https://github.com/PolicyEngine/policyengine-core", "supports": "Simulation architecture and testing commands." }, { "title": "PolicyEngine US", "url": "https://github.com/PolicyEngine/policyengine-us", "supports": "Rules and microdata purpose." }, { "title": "US development guide", "url": "https://raw.githubusercontent.com/PolicyEngine/policyengine-us/master/CLAUDE.md", "supports": "Contributor testing and evidence conventions; retrieved copy is older, not a current orchestration audit." } ], "unknowns": [ "How agents are scheduled, delegated and accepted across PolicyEngine repositories.", "Which share of submitted token usage belongs to PolicyEngine.", "Dated January–September milestones for an autonomous factory." ], "example": "Illustrative CASE-01: an analyst compares the baseline with a hypothetical policy reform; no real eligibility or financial advice.", "sceneContract": "Choose → analyst → CASE-01 → select model and inputs → scenario → rules; load → country model → scenario → select formulas/parameters → simulation → core; calculate → core → simulation → evaluate → results → analyst; inspect → analyst → results → interpret assumptions. Separate gray boundary: coding-agent factory unknown.", "takeaway": "A substantive application can be visible even when its software-production machinery is not.", "checked": "Read profile, core repository, US repository and contributor guide. Did not execute a policy model or assess legal accuracy.", "imageAlt": "PolicyEngine miniature modelling office: analyst inputs, country rules, core calculations and interpreted results, with coding-factory automation explicitly unknown.", "asOf": "2026-09-22", "image": "images/max-ghenis.png", "imageStatus": "generated" }, { "id": "maple-gao", "name": "Maple Gao / MapleEve", "title": "Private audio workspace and coding governance", "category": "Tool portfolio", "summary": "A connected recording and transcription workflow, alongside separate coding-attribution and home-automation tools.", "verdict": "Product pipeline is documented; the process that builds these products is not a proven autonomous factory.", "evidence": "Documented architecture", "status": "BetterAINote is described as 0.6.2-preview; VoScript and aitrack have public implementation docs. Snapshot checked September 22.", "priority": 16, "projects": [ { "name": "BetterAINote", "role": "Collects recordings into a private library, archives audio, manages transcripts and review.", "url": "https://github.com/MapleEve/BetterAINote" }, { "name": "VoScript", "role": "Self-hosted transcription and persistent speaker voiceprints; integrates with BetterAINote.", "url": "https://github.com/MapleEve/VoScript" }, { "name": "company-aitrack", "role": "Signed edit evidence, local usage collection and server validation for coding attribution.", "url": "https://github.com/MapleEve/company-aitrack" }, { "name": "lifesmart-for-homeassistant", "role": "LifeSmart integration for Home Assistant; separate project.", "url": "https://github.com/MapleEve/lifesmart-for-homeassistant" } ], "orchestration": { "trigger": "Product workflow: configured recording-source sync or an uploaded audio file.", "controller": "BetterAINote web app and background worker manage the library; VoScript is a separate transcription service.", "workers": "Audio/transcription services; not evidence of coding-agent worker orchestration.", "memory": "Local audio archive and SQLite records; VoScript retains enrolled voiceprints.", "verification": "Speaker review is exposed to the user; unknown speakers retain generic labels.", "human": "Configures providers and enrolled speakers, reviews the resulting transcript.", "output": "Searchable recording records, transcripts and speaker labels.", "failure": "Unavailable source audio cannot be assumed retrievable; unenrolled voices keep generic IDs. Coding-factory recovery is unknown." }, "stages": [ { "name": "Collect recordings", "owner": "BetterAINote worker", "action": "Sync accessible source records into a private library.", "certainty": "documented" }, { "name": "Transcribe audio", "owner": "VoScript service", "action": "Convert audio into timestamped words.", "certainty": "documented" }, { "name": "Match speakers", "owner": "Voiceprint service", "action": "Match enrolled voices; retain generic IDs for unknowns.", "certainty": "documented" }, { "name": "Review and find", "owner": "User + BetterAINote", "action": "Review speakers and browse the recording archive.", "certainty": "documented" } ], "timeline": [ { "date": "2026-09-22", "event": "Reviewed README describes BetterAINote 0.6.2-preview and its VoScript integration; observation date only.", "source": "https://github.com/MapleEve/BetterAINote" }, { "date": "2026-09-22", "event": "aitrack documents signed edit evidence, status heartbeats and local usage scans; no factory release chronology established.", "source": "https://github.com/MapleEve/company-aitrack" } ], "metrics": [ { "label": "All-time submitted tokens", "value": "672.1B · leaderboard #13", "date": "2026-09-22", "source": "https://tokscale.ai/leaderboard", "caveat": "Self-submitted usage snapshot; includes cached input. Not project-attributable output, monthly burn, money spent or account count." } ], "sources": [ { "title": "BetterAINote README", "url": "https://raw.githubusercontent.com/MapleEve/BetterAINote/main/README.md", "supports": "Preview status, library, worker and review surface." }, { "title": "VoScript", "url": "https://github.com/MapleEve/VoScript", "supports": "Audio API, enrolled speaker matching and generic unknown-speaker labels." }, { "title": "aitrack", "url": "https://github.com/MapleEve/company-aitrack", "supports": "Separate signed-evidence client/server workflow; usage is distinct from edit proof." }, { "title": "Maple Gao profile", "url": "https://github.com/MapleEve", "supports": "Project inventory." } ], "unknowns": [ "Coding-agent scheduler, acceptance gates and unattended shipping behavior.", "Actual adoption and measured transcript/speaker accuracy.", "Complete January–September release history." ], "example": "Illustrative AUDIO-01: one meeting recording yields timestamped words; a voice without enrolment remains SPEAKER_XX.", "sceneContract": "Collect → BetterAINote worker → AUDIO-01 → archive authorized audio → file → VoScript; transcribe → VoScript → file → decode speech → words → matcher; match → VoScript → words/voiceprints → attach enrolled names or generic speaker IDs → transcript → user; review → user → transcript → inspect speakers → searchable record. Separate products shelf: aitrack, LifeSmart integration. Gray boundary: coding factory unknown.", "takeaway": "The audio pipeline is visible; the factory that writes its software is not yet established.", "checked": "Read original Chinese product READMEs and profile. Did not execute services or test identification accuracy.", "imageAlt": "Private audio workshop showing recording intake, VoScript transcription, enrolled-speaker matching and human review, with separate products and unknown coding orchestration.", "asOf": "2026-09-22", "image": "images/maple-gao.png", "imageStatus": "generated" }, { "id": "kaptinlin", "name": "kaptinlin", "title": "Go validation and repair libraries", "category": "Tool portfolio", "summary": "A public library portfolio for repairing, validating and translating application data.", "verdict": "Inspectable libraries with tests; no public factory orchestration established in this review.", "evidence": "Partial evidence", "status": "September snapshot shows jsonrepair, jsonschema, gozod and go-i18n. Token scale cannot be attributed to a particular library.", "priority": 17, "projects": [ { "name": "jsonrepair", "role": "Repairs malformed JSON-like text into parseable JSON; Go port of Jos de Jong's library.", "url": "https://github.com/kaptinlin/jsonrepair" }, { "name": "jsonschema", "role": "JSON Schema Draft 2020-12 validation for Go.", "url": "https://github.com/kaptinlin/jsonschema" }, { "name": "gozod", "role": "Type-safe Go validation API inspired by Zod.", "url": "https://github.com/kaptinlin/gozod" }, { "name": "go-i18n", "role": "ICU MessageFormat localisation and fallback handling.", "url": "https://github.com/kaptinlin/go-i18n" } ], "orchestration": { "trigger": "Product workflow: application calls jsonrepair.Repair with malformed text.", "controller": "Library parser/repair function; coding-agent controller: Not established in reviewed sources.", "workers": "Deterministic library routines, not a documented agent fleet.", "memory": "Input text and parser state; persistent coding memory is unknown.", "verification": "Repository documents race-tested unit tests and lint. Valid syntax does not verify semantic truth.", "human": "Application developer handles return values and decides whether repaired content is usable.", "output": "A repaired JSON string or a structured error.", "failure": "Unrepairable input returns an error to the caller; no claimed autonomous recovery loop." }, "stages": [ { "name": "Submit malformed text", "owner": "Calling application", "action": "Pass DATA-01 to Repair.", "certainty": "documented" }, { "name": "Repair syntax", "owner": "jsonrepair library", "action": "Normalize quotes and missing JSON punctuation.", "certainty": "documented" }, { "name": "Return result", "owner": "Library API", "action": "Return valid JSON or a structured error.", "certainty": "documented" }, { "name": "Decide how to use it", "owner": "Application developer", "action": "Handle failure and validate application meaning.", "certainty": "documented" } ], "timeline": [ { "date": "2026-09-22", "event": "Reviewed current public library descriptions and jsonrepair API/test documentation; no nine-month factory timeline established.", "source": "https://github.com/kaptinlin/jsonrepair" } ], "metrics": [ { "label": "All-time submitted tokens", "value": "1.238T · leaderboard #3", "date": "2026-09-22", "source": "https://tokscale.ai/leaderboard", "caveat": "Self-submitted usage snapshot; includes cached input. Not project-attributable output, monthly burn, money spent or account count." } ], "sources": [ { "title": "jsonrepair", "url": "https://github.com/kaptinlin/jsonrepair", "supports": "API, repair operations, errors, tests and Go-port attribution." }, { "title": "kaptinlin profile", "url": "https://github.com/kaptinlin", "supports": "Named library portfolio; previously reviewed September 22." } ], "unknowns": [ "Agent controller, delegation, memory and approval process used to develop the libraries.", "Dated factory milestones and independently measured developer productivity.", "Account count and actual spending." ], "example": "Illustrative DATA-01: {name: 'Ada'} becomes valid JSON with name equal to Ada. This demonstrates syntax repair, not truth checking.", "sceneContract": "Submit → application → DATA-01 malformed JSON → call Repair → parser; repair → library → DATA-01 → normalize syntax → JSON or error; return → API → result → deliver to caller → developer; decide → developer → returned data → check application requirements. An error branch goes to caller. Separate library shelves, no arrows implying all libraries form one deployed pipeline.", "takeaway": "Repairing JSON syntax is a useful component, not evidence of an autonomous software factory.", "checked": "Read library API and tests section and existing profile evidence. Did not run the Go code.", "imageAlt": "A Go library miniature workshop teaching malformed JSON repair and its error route, with other libraries on an unconnected shelf.", "asOf": "2026-09-22", "image": "images/kaptinlin.png", "imageStatus": "generated" }, { "id": "mateo-palau", "name": "Mateo Palau / Nova Labs", "title": "MCP Doctor and B2B automation", "category": "Verification", "summary": "A small offline MCP configuration inspector, alongside a self-described B2B AI automation business.", "verdict": "The inspector's behavior is documented. A production software factory or client deployment scale is unverified.", "evidence": "Partial evidence", "status": "September snapshot shows MCP Doctor 0.1.0 source/package and Nova Labs founder profile; only three repository commits were displayed at review.", "priority": 18, "projects": [ { "name": "MCP Doctor", "role": "Offline CLI that checks local MCP JSON files and reports structural or safety findings.", "url": "https://github.com/mateopalau/mcp-doctor" }, { "name": "Nova Labs", "role": "Founder describes B2B AI systems and automation; public delivery architecture is not established.", "url": "https://github.com/mateopalau" } ], "orchestration": { "trigger": "User runs mcp-doctor on a file or standard input.", "controller": "A local deterministic CLI; agent orchestration: Not established in reviewed sources.", "workers": "JSON parsing, configuration structure checks and bounded safety detectors.", "memory": "Input configuration and finding locations; no persistent agent memory established.", "verification": "Reports errors, high-severity findings and machine-readable results.", "human": "Operator reads findings and edits configuration. Automatic repair is not claimed.", "output": "Human-readable or JSON report plus process exit code.", "failure": "Invalid JSON/structure is exit 1; high-severity findings exit 2; usage errors exit 3." }, "stages": [ { "name": "Choose a config", "owner": "Operator", "action": "Pass CONFIG-01 to the offline CLI.", "certainty": "documented" }, { "name": "Inspect structure", "owner": "MCP Doctor", "action": "Parse JSON and check server definitions.", "certainty": "documented" }, { "name": "Flag risky values", "owner": "Safety rules", "action": "Detect recognized broad filesystem paths and secret-shaped values.", "certainty": "documented" }, { "name": "Review findings", "owner": "Operator", "action": "Read redacted results; decide how to correct config.", "certainty": "documented" } ], "timeline": [ { "date": "2026-09-22", "event": "Reviewed public 0.1.0 tarball, CLI behavior and founder profile; observation date, not release date.", "source": "https://github.com/mateopalau/mcp-doctor" } ], "metrics": [ { "label": "All-time submitted tokens", "value": "933.1B · leaderboard #10", "date": "2026-09-22", "source": "https://tokscale.ai/leaderboard", "caveat": "Self-submitted usage snapshot; includes cached input. Not project-attributable output, monthly burn, money spent or account count." } ], "sources": [ { "title": "MCP Doctor", "url": "https://github.com/mateopalau/mcp-doctor", "supports": "Checks, redaction, output and exit codes." }, { "title": "Mateo Palau profile", "url": "https://github.com/mateopalau", "supports": "Nova Labs business description and public project." } ], "unknowns": [ "Nova Labs' private automation deployments and their operational state.", "Coding-agent controller, review process, account inventory and unattended shipping.", "How the all-time token total relates to this small public repository." ], "example": "Illustrative CONFIG-01 grants a recognized filesystem server the root directory. MCP Doctor flags the broad path; the operator decides the correction.", "sceneContract": "Choose → operator → CONFIG-01 → invoke offline CLI → JSON; inspect → parser → JSON → structural findings → detector; flag → rules → server config → broad-path/secret-shaped warnings → report; review → operator → redacted report → choose edit. Invalid JSON routes directly to error report. No network/server launch or automatic fixing implied.", "takeaway": "A configuration inspection tool is visible; the surrounding production factory remains unknown.", "checked": "Read public README, package presence and profile. Did not run the CLI or inspect private client systems.", "imageAlt": "Offline MCP configuration inspection workshop with file intake, structure checks, risk findings and human correction, explicitly not an autonomous factory.", "asOf": "2026-09-22", "image": "images/mateo-palau.png", "imageStatus": "generated" } ], "unresolved": [ { "name": "P-0001 / PT", "rank": 1, "tokens": "7.129T", "url": "https://tokscale.ai/leaderboard", "reason": "Highest displayed total. Profile retrieval did not establish the project, individual-versus-team scope, or factory." }, { "name": "kzquandary", "rank": 4, "tokens": "1.192T", "url": "https://tokscale.ai/u/kzquandary", "reason": "Current project unresolved. Its retrieved usage profile was approximately 89.5% cached reads; last submission shown May 6." }, { "name": "netsus", "rank": 5, "tokens": "1.172T", "url": "https://github.com/netsus", "reason": "Retrieved profile was stale. Older pinned projects do not explain the present token total." }, { "name": "guswn2521 / HyeonJu Son", "rank": 6, "tokens": "979.0B", "url": "https://tokscale.ai/leaderboard", "reason": "Public profile retrieval did not establish a current project or orchestration." }, { "name": "daylab-official / DayLab", "rank": 8, "tokens": "960.7B", "url": "https://tokscale.ai/leaderboard", "reason": "Public profile retrieval did not establish a current project or orchestration." }, { "name": "Gusarich / Daniil Sedov", "rank": 11, "tokens": "803.2B", "url": "https://gusarich.com", "reason": "Substantive 2025 compiler-fuzzing research was found, but its blockchain tooling falls outside the requested noncrypto shortlist. No current noncrypto factory established." } ] }