| name | traceability-review |
| description | Review explicit architecture traceability metadata for gaps, inconsistencies, unsupported claims, stale links, dangling references, and metamodel violations. Use when Codex needs to inspect artifact relations, validate ADR-to-risk or quality-scenario coverage, propose relation updates, or produce an evidence-backed traceability review report. |
Traceability Review Skill
Purpose
Review explicit architecture traceability metadata for gaps, inconsistencies,
and unsupported claims. Treat every AI-suggested relation, finding, and impact
claim as proposed until a human reviewer accepts it in the repository.
This skill is guidance and reusable content only. Do not implement AI execution,
do not implement a generator, and do not add automation as part of using this
skill.
Apply the repository contract hierarchy: this skill narrows AGENTS.md, and
AGENTS.md adapts general-semantic-contracts.md. Do not duplicate general
metadata, anchor, Docs-as-Code, arc42, or traceability rules here unless the
traceability review workflow needs a stricter rule.
Core Workflow
- Inspect the requested documentation scope before writing findings.
- Read artifact metadata and relation metadata from repository source files.
- Inspect the relevant AsciiDoc prose only to verify relation rationale,
visible references, anchors, and evidence.
- Inspect relevant source code, tests, configuration, or operational evidence
only when relation claims depend on those artifacts.
- Validate relation endpoints, relation types, ownership, status, rationale,
and review state against the metamodel.
- Compare explicit metadata relations with visible AsciiDoc
xref links when
both are expected.
- Identify missing, weak, stale, dangling, conflicting, or unsupported
relations.
- Propose relation additions or changes as
proposed; do not silently create
accepted relations.
- Update affected artifacts with outgoing relations only when the user
requested edits. Never create reciprocal incoming relations; they are derived
automatically during generation.
- Prefer deterministic ordering by source ID, relation type, target ID, then
rationale.
- Prefer small, reviewable patches.
- Run the repository validator after changes if one is available.
Required Reading
Read these references as needed:
../../src/docs/arc42/04-solution-strategy/doc-04001-metamodel.adoc before reviewing
artifact metadata, artifact types, lifecycle states, or relation semantics.
../references/metadata-rules.md before reviewing YAML front matter.
../references/relation-rules.md before proposing or updating relations.
../references/adr-writing-guide.md before reviewing ADR relations.
../references/risk-writing-guide.md before reviewing risk relations.
../references/quality-scenario-guide.md before reviewing quality scenario
relations.
Use these templates:
../../templates/impact-report.adoc for review reports when the user wants
analysis without direct artifact changes.
Inputs
- Artifact metadata conforming to
metamodel/artifact.schema.yaml.
- Relation metadata conforming to
metamodel/relations.schema.yaml.
- Relevant AsciiDoc source documents.
- Review focus, such as ADR-to-risk coverage or quality-scenario coverage.
- Optional evidence sources, such as code, tests, configuration, deployment
files, issue references, or operational data.
- Review scope: report-only, update proposed relations, or validate a changed
artifact set.
Output
Produce a review report with:
- Missing or weak relations.
- Stale, dangling, or conflicting relations.
- Suggested relation additions marked as
proposed.
- Questions for human reviewers.
- Validation notes against the metamodel.
Inspection Rules
- Prefer reviewed repository source files over conversational memory.
- Do not load derived output as evidence or architecture context. This includes
generated/, build/, dist/, target/, out/, rendered HTML/PDF,
generated indexes, traceability views, and assembled documentation.
- Treat explicit metadata as authoritative over inferred prose.
- Treat visible
xref links as documentation references, not as substitutes for
metadata relations when traceability is required.
- Check the artifact and relation schemas before declaring a relation invalid.
- Check existing artifact IDs before suggesting new IDs.
- Never add reciprocal relations. Incoming relations are derived from outgoing
relations during generation.
- Preserve accepted or reviewed relations unless repository evidence justifies a
proposed update.
Output Rules
- Do not silently create accepted relations.
- Identify dangling references by stable ID.
- Mark AI-suggested additions as
status: proposed and reviewed: false.
- Prefer precise relation types over
relates_to.
- Prefer deterministic ordering by source ID, relation type, target ID, then
rationale.
- Keep findings actionable and evidence-backed.
- Check metadata, relation, anchor, and
xref rules from
general-semantic-contracts.md.
Review Checklist
- All relation endpoints exist.
- Relation types match the intended semantics.
- Accepted relations have evidence or rationale.
- AI-suggested additions are not marked as reviewed.
- Visible references to documentation artifacts use valid anchor-based
AsciiDoc
xref links where required.
- Visible references use the explicit target anchors from source files, not
raw numbered chapter file names.
- Open questions identify the human decision needed to resolve each uncertain
relation.
Validation
After edits, run the repository validator if available. Prefer existing commands
from the repository documentation or scripts. If no validator exists, report
that validation was not available and describe the manual checks performed.