Docs traceability desk

Docs Traceability Desk

Suite workflow mode

This desk is part of the SDLC Command Desk workflow suite. When invoked from an end-to-end workflow, do not stop with only a bare next-desk instruction. Complete this desk's artifact, emit a workflow packet, and continue to the next stage when enough facts are available.

If the next stage cannot be completed because required facts, connector access, approval, or source evidence are missing, return Workflow Halt with specific resume requirements. Use references/suite-workflow-contract.md for the packet, continuation, and halt format.

Purpose

Use this skill to turn scattered documentation, repository facts, requirements, tests, issues, and pull requests into traceable documentation artifacts. The skill verifies claims against source evidence, identifies unsupported or stale docs, builds proof maps, and prepares downstream handoff notes for implementation-handoff-desk, verification-desk, review-quality-desk, or issue-planning-desk.

This skill is evidence-first. It must not invent source facts, test names, file paths, PR numbers, issue IDs, status claims, or implementation coverage.

Workflow

Outcome. The documentation artifact the request calls for: a proof map or claim map, a documentation update plan, a doc-code consistency report, a knowledge index or source catalog, an audit evidence packet, or a downstream handoff note for a PR, verification, or issue-planning workflow.

Grounding. Use GitHub for repo files, commits, PRs, issues, tests, code paths, and CI facts. Use docs connectors or uploaded docs for product, roadmap, architecture, status, parity, audit, and decision documents. Use communication connectors only for decision history or halt reports that are not captured in repo or docs.

Claim extraction. Use references/claim-map-template.md for claim-level mapping and references/proof-map-template.md for source-backed proof tables. Every claim must be assigned exactly one of supported, partially supported, unsupported, stale, conflicting, or unverified.

Doc-code consistency. Compare documentation claims against source files, tests, issue/PR state, and current repository structure. Use references/doc-code-consistency.md to classify drift and impact.

Parallel surface. Individual claims are independent units of work: each is checked against its own evidence and no claim's status depends on another's. Extract and evaluate claims in parallel, and retrieve the documents and repository paths backing them in parallel too. Fan out across documents rather than walking them one at a time.

Artifact production. Follow references/output-contract.md. For prompts, reports, plans, indexes, and handoff notes, create a downloadable Markdown artifact when file-writing tools are available. Include source facts and unverified assumptions.

Downstream handoff. Do not write implementation PR prompts unless asked. Produce handoff notes with target files, source evidence, risk, acceptance gates, and halt conditions that implementation-handoff-desk can convert into a PR prompt.

Acceptance bar. The artifact is done when every extracted claim carries a status and a named source artifact for that status; unsupported, stale, and conflicting claims are visible rather than smoothed over; conflicts are preserved instead of resolved silently; and each unsupported or unverified claim names the next evidence-gathering step. Do not invent source facts, test names, file paths, PR numbers, issue IDs, status claims, or implementation coverage.

Default artifact types

A traceability run delivers a set rather than the smallest artifact that satisfies the request:

• claim-map.md: every claim extracted from the documents in scope, each carrying exactly one status.
• proof-map.md: each claim tied to the specific file, test, PR, issue, or commit that proves it or fails to.
• doc-code-consistency-report.md: the drift those two maps expose, classified by impact.
• documentation-update-plan.md: the scoped change that would close the drift, per document.
• docs-handoff-notes.md: what implementation, verification, or issue planning inherits, with target files and acceptance gates.

Three artifact types are genuinely mode-specific and are produced instead of that set, or beside it only when asked for by name:

• knowledge-index.md when the request is for a durable source catalog rather than a claim audit.
• audit-evidence-packet.md when governance or release evidence is what was asked for.
• connector-diagnostic.md when grounding is insufficient: it replaces the set rather than shipping next to one built on nothing.

Depth is what makes the set worth producing. A claim entry names the document, the location inside it, the claim as written, the status, and the artifact that establishes that status. An update plan says what the document should say instead, not that it needs updating. A row reading "unverified, needs investigation" without naming the investigation is an unfinished row.

Claims are independent units, which is exactly the parallel surface the workflow describes: extract, evaluate, and write across the set concurrently.

None of that makes an unproven claim provable. A claim with no evidence stays unsupported or unverified however awkward it looks in the table, conflicting sources stay conflicting, and a proof map with more gaps than proofs is an accurate proof map. The set is complete when every claim has an honest status, not when every status is green.

Required references

Load only the reference needed for the current artifact:

• references/proof-map-template.md for proof maps.
• references/claim-map-template.md for claim classification.
• references/doc-code-consistency.md for drift and consistency checks.
• references/knowledge-index-template.md for documentation/source indexes.
• references/audit-evidence-template.md for audit packets.
• references/connector-routing.md for grounding requirements.
• references/source-hierarchy.md for truth precedence.
• references/output-contract.md for artifact structure.
• references/handoff-rules.md for downstream PR/verification handoff.
• references/halt-conditions.md for stop conditions.

Halt policy

Proceed by default. A claim that cannot be proven is a result, not a blocker: mark it unsupported or unverified, name the missing evidence, and continue mapping the rest. Reserve hard halts for the consequence classes in references/halt-taxonomy.md:

• Approval: the request would publish, overwrite, or retire documentation that a human owner must authorize.
• Production or destructive: the plan would delete or rewrite canonical source-of-truth documents rather than propose changes.
• Security or privacy: proving a claim would require reproducing secrets, credentials, or personal data in the artifact.
• Source conflict: documentation and repository state genuinely disagree on a load-bearing fact. Record it as conflicting and halt rather than picking a side.
• Release integrity: an audit or evidence packet would assert coverage that available evidence cannot establish.
• Connector unreachable: a required repo or doc source exists but cannot be read. Evidence that is merely absent is a soft gap: produce a connector diagnostic or continue with the claim marked unverified.

Quality bar

Every traceability artifact must separate facts from inference, cite or name source artifacts, mark unsupported claims clearly, and preserve conflicts rather than smoothing them over. When a doc says something that the repo cannot prove, label it as unsupported or unverified and recommend the next evidence-gathering step.

Continuity Kernel Adoption

Use references/continuity-kernel.md, references/capability-baseline.md, references/readiness-gates.md, references/halt-taxonomy.md, references/preflight-cache.md, and references/handoff-density-policy.md when participating in an SDLC Command Desk workflow. Preserve and update the continuity_packet instead of reasking for facts already present. Classify missing inputs as hard halts, soft gaps, or auto-routable upstream/downstream work. Use HANDOFF_BLOCKER when implementation handoff facts are insufficient for a coding agent.