Product: The capability factory, end to end

The projection pipeline

From canonical source to a managed file on disk.

Validate references. Every manifest reference is resolved before anything is written. A missing or inconsistent reference stops the run, and the v2 release gate fails closed.

Compose the bundle. Typed manifests decide which skills, agents, workflows, and services a given surface receives. Catalogs are per-surface, never one global dump.

Check drift. The sync engine snapshots projected-file hashes and compares them against what is actually on disk under the runtime root.

Project. Managed output is written into each declared runtime root, one way only. Surfaces subscribe to what the factory publishes.

Checkpoint. New hashes are recorded, so the next run can tell a fresh projection apart from a hand edit.

Drift detection is the honest part of that loop: the sync halts before touching a hand-edited live file and surfaces a diff, so the edit can be promoted into canonical source instead of clobbered.

Fail-closedDrift-detectedEvidence-backed

Some of this is mechanically enforced today (the VerificationGate hook, drift detection, fail-closed validators); the rest is honest self-attestation, and the repo grades each claim’s enforcement level by name.

01 / SourceCanonical sourceSkills, agents, workflows, contracts, and adapters, versioned in one repository.

02 / ManifestsTyped manifestsDeclare what each surface receives. A missing reference stops projection.

03 / SyncOne-way projectionManaged files written per surface, hashed and drift-checked.

04 / Surfaces12 declared surfaces

Claude CodeCodexHermesOpenClawPrime AgentPaperclip+ 6 more declared, four at experimental tier

evidence returns: doctor · readiness · tests · Work Packets

Diagram: canonical source projected through typed manifests into twelve declared surfaces, with evidence returning to source.

What the factory owns

One factory. Every surface.

Everything below is versioned source in one repository, projected per-surface through typed manifests.

Skills

177 active skills across 26 declared domains, authored once in canonical source and projected per surface.

Agents

Nearly 40 named agents, including cross-vendor wrappers: Claude models plus Gemini, Grok, Qwen, DeepSeek, Kimi and more.

Workflows

235 multi-step workflows, versioned alongside the capabilities they call.

Surfaces

12 declared surfaces, six of them primary executors. A declaration says what the factory can project, not that a given machine is healthy.

Runtime services

17 declared services: voice, browser, observability, WhatsApp, knowledge, and more.

Scheduled jobs

51 declared jobs: nightly ship, inbox checkers, memory health, secret scanning.

Memory federation

Paperclip owns the durable backlog. gbrain synthesizes cross-surface knowledge with lineage. Human-authored intent wins conflicts.

Verification machinery

The VerificationGate hook, the drift detector, doctor and readiness commands, and 34 gate validators whose registry is asserted against the validator file set in CI, so a new validator fails the build until it is registered.

Vendored lines

Paperclip and gbrain are vendored upstreams, carried with explicit patch boundaries rather than claimed as in-house work.

At a glance

The factory, counted.

Counts come from the repo’s generated indexes and manifests, August 2026.

177Active skills

235Workflows

10,900+Tests green

51Scheduled jobs

12 declared surfaces26 domains13 bundles17 runtime services34 gate validatorsnearly 40 agents

Private, single-operator, macOS. v1.0 and v1.1 tagged; v2.0.0 is a release candidate.

One canonical source. Every declared surface.