Adopt an existing project

Turn a generic installed seed into a project-specific knowledge plant. Follow docs/graph/protocols/initialize.md; this skill supplies the existing-project discovery and authoring discipline.

Invariants

• docs/graph/ is the only maintained knowledge root. The router, nodes, LLM wiki, provenance, plans, runbooks, contracts, and deep dives are layers of this one graph (rule.knowledge).
• Executable source outranks prose. Manifests, entry points, routes, schemas, migrations, config, deployment, tests, CI, prompts, and evals are primary evidence. Existing docs are corroborating evidence only.
• Make additive knowledge changes. Preserve application code and unrelated existing files. Leave competing AI configurations in place; deleting or relocating one is the owner's call.
• Adoption ONLY records build and test commands: extract each exact command and label it discovered, not executed, with where it was found (path:line) and, when they hold, target absent or declared, non-functional (the states are verify.gate-states; the growth gate is grow.gate.executed). An inherited suite stays untrusted until proven RED by mutation (docs/graph/protocols/test-first.md), so record it as discovered.
• Git stays read-only: record the repository revision as provenance; fetch, pull, switch, commit, and push each need a separate explicit request.
• Record observed behaviour and observed choices as node facts. Specs and ADRs record intent, which only an existing record or the owner supplies. Unknown means unknown.

Scout pass

Establish the governed boundary first: one repo, workspace/monorepo, or an umbrella of sibling repos. For each repository record path, branch, HEAD, worktree state, role, manifests, stack, and the natural language of its identifiers and domain vocabulary. A non-English (or otherwise non-default) language is an explicit graph fact, because downstream agents grep and reason in it and an unrecorded mismatch silently defeats every later search. A glossary entry for a term in that language is checked against the fields and code that use the term: the obvious translation, or a plausible cognate, is a hypothesis until the code agrees, and a wrong gloss sends every later search to code that does not exist. Boundaries follow capabilities, not repository count.

Inventory cheaply, excluding generated/vendor/cache/build output; the inventory of source-index.py build --json, run once, is the scouts' file list. Then open the smallest authoritative files needed to trace:

1. bootstrap and runtime entry points;
2. module/service/package boundaries and imports;
3. inbound APIs, messages, scheduled work, and outbound integrations;
4. entities, schemas, migrations, storage, and data movement;
5. config/secrets interfaces, deployment, and observability;
6. tests, CI, scripts, prompts, evaluations, and operational commands;
7. direct dependencies and evidence of their actual usage.

Return facts with exact paths and symbols. Find source corroboration for each claim prose makes, or record it as untrusted/unverified; repetition across central docs adds no corroboration.

Every delegated worker's brief embeds the canonical block from docs/graph/templates/prompts/graph-session-bootstrap.md.

Librarian pass

Normalize the evidence into single fact owners:

• configure ROOT_ID and KINDS in docs/graph/graph-lint.py;
• author a root node and capability/subsystem nodes;
• factor shared stack, platform, data, domain, and cross-cutting facts into their own nodes only where this reduces duplication;
• give nodes concrete load_when triggers and source paths;
• keep requires minimal and acyclic; use peers for boundaries, and composes where an expertise.* node offers sub-expertises the router should descend into only when the task names one;
• link detailed leaves using artifacts: and dependency wiki pages using libraries:.

Then enrich the leaf collections under docs/graph/:

Keep specs/ and decisions/ indexes, and fill them only from genuine intent records that already exist and can be preserved with provenance. Put implementation observations in nodes or architecture leaves.

When the sweep confirms zero test or gate infrastructure, write an absent (YYYY-MM-DD) — <reason> row for each standard gate in the verification runbook: a blank runbook is indistinguishable from one nobody checked, and a row for each gate turns an unknown into a stated finding.

When a legacy or parallel documentation source predates and conflicts with the evidence-derived graph, give the exclusion a real, routable node that states (a) the source is excluded as evidence, (b) what supersedes it, and (c) that this is a trust/evidence decision only: the artifact itself stays untouched (Invariants). A silent skip reads to a later agent as no decision, and the artifact gets re-trusted. The Handoff's excluded-docs report then points at this node rather than standing in for it. An agent-operations system is not such a source: protocol.grow Phase 4 says what each of its parts yields.

Refreshing an existing graph

Treat graph prose as a read model to verify against current source:

1. compare repository revisions with the graph changelog/provenance;
2. scout changed areas and their dependency/data/API blast radius (source-index.py impact, run once over the moved paths);
3. update the existing fact owner instead of creating a duplicate;
4. preserve valid hand-authored context;
5. remove or supersede stale seed-owned claims only with cited contrary evidence;
6. record conflicts and unanswered questions in the plan.

Report a refresh as partial, naming each unavailable source, whenever part of the governed source was out of reach.

Validation

Run knowledge checks only:

python3 docs/graph/graph-lint.py
python3 docs/graph/graph-lint.py --plan "change a representative capability"

Verify all graph links resolve; every Tier-3 leaf is reachable from an owning node; no duplicate fact homes exist; representative task routes are small and relevant; commands say whether they were executed; the verification runbook carries the absent-gate rows where no infrastructure was found (Librarian pass); each recorded command carries its found-at, not-executed and target labels (Invariants); and no template placeholders, fabricated success, inferred specs, or invented rationale remain.

Use known-answer navigation questions from several capabilities plus an adversarial false-premise question. A wrong or bulk-read answer is a graph defect: improve ownership, routing, or leaf content and rerun, for at most two fix-and-rerun rounds per defect; with the authoring pass they are the three attempts docs/graph/protocols/recover.md allows. A defect that survives them goes to the user as an honest unknown.

Handoff (the stopping condition)

Adoption is done when validation passes within its bounded rounds and the open defects are recorded; completeness is measured by reliable progressive discovery. End with the payload from docs/graph/templates/prompts/handback-payload.md, and report: repository revisions, evidence inspected, graph artifacts created or refreshed, validation outcomes, docs deliberately excluded as untrusted (each named by its exclusion node), remaining unknowns, and one highest-leverage next action with its tier.