How this skill is triggered — by the user, by Claude, or both
Slash command
/sf:documentThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Creates minimal, proportional documentation. Match doc weight to change weight.
Creates minimal, proportional documentation. Match doc weight to change weight.
/sf:document [SPECIFICATION_AND_IMPLEMENTATION_PATHS]
test -f .claude/.sf/implementation-summary.md && head -20 .claude/.sf/implementation-summary.md || echo "none"Input Resolution:
.claude/.sf/spec.md and .claude/.sf/implementation-summary.md exist: Use themAfter input resolution, run agents in 3 batches:
Batch 1 (Parallel):
Gate 1 — Post-Analysis: Check that each output file exists:
$SF_DIR/research/artifacts-summary.md$SF_DIR/research/implementation-summary.md$SF_DIR/research/docs-inventory.mdMissing file = agent failed to run → halt pipeline, report which file(s) are missing, preserve existing output for inspection. Empty or whitespace-only file = valid "no docs needed" signal → pass. Write gate results to $SF_DIR/research/gate-analysis.md (pass/fail per file, warnings for files under 5 lines).
Batch 2 (Parallel):
Gate 2 — Post-Generation: For each generated doc, verify:
TODO, TBD, PLACEHOLDER, [INSERT (case-insensitive)Block on: placeholder-only content. Empty files are valid — they mean the agent correctly determined no docs were needed. Write gate results to $SF_DIR/research/gate-generation.md.
Batch 3:
Gate 3 — Post-Integration:
Verify integration completed. If no docs were updated or created, that's valid — report "no documentation changes needed" and finish cleanly. Write gate results to $SF_DIR/research/gate-integration.md.
Output: Documentation updates (if any) + terminal summary
On any gate failure: halt immediately, do not proceed to the next batch. Report which gate failed and why. Preserve output for inspection.
npx claudepluginhub bitcraft-apps/spec-firstScans a codebase and generates structured documentation: READMEs, API references, setup guides, runbooks, ship logs, and release notes. Useful after PRs or for onboarding material.
Generates API, architecture, and user documentation from code using AI analysis. Automates documentation pipelines and maintains consistency across repositories.