scan
Skill0 starsChanged 2 months agoApache-2.0View on GitHub
Run an audit scanner against ODD repositories. Reads scanner definition, picks unscanned items from coverage manifest, writes findings, updates navigation.
39 other projects use a version of this.
Adopted by
Most recently taken up by
- XORwell/archeflowtheir own version, first seen Nov 15, 2026
- willcoliveira/mac-cleanup-agenttheir own version, first seen Nov 15, 2026
- onur-yuksekkaya/cve-deltatheir own version, first seen Nov 15, 2026
- bitrix-tools/marketplace-security-skillstheir own version, first seen Nov 15, 2026
- cfgaudit/cfgaudittheir own version, first seen Nov 10, 2026
- jskswamy/skillstheir own version, first seen Nov 8, 2026
Used with
Projects with it also use metrics, report and 8 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
GoogleSkillmetrics
Expertise in analyzing time-series repository health metrics, investigating root causes, and proposing proactive workflow improvements.
Also in 3 projects that use this
MicrosoftSkillreport
Turn an exploration (threads, findings, charts) into a single Markdown report — note, blog post, executive summary, KPI dashboard, slide brief, or multi-section analytical report, with embedded charts.
Also in 5 projects that use this
anombyte93Skilldiscover
Phase 1 of the prd-taskmaster pipeline: brainstorm-driven discovery. Delegates to superpowers:brainstorming in Interactive Mode (one adaptive question at a time), or self-brainstorms in Autonomous Mode when no user is present. Intercepts before the brainstorming chain hands off to writing-plans — this skill owns the exit. Extracts constraints, calibrates scale (Solo / Team / Enterprise), and advances the pipeline to GENERATE.
Also in 3 projects that use this
tuistSkillmigrate
Migrates existing Xcode projects to Tuist generated workspaces with build and run validation, external dependency mapping, and migration checklists. Use when adopting Tuist for an existing app or converting a hand-edited Xcode project to generated projects.
Also in 3 projects that use this
Google CloudSkillfix
Plan a remediation strategy after a test failure.
Also in 3 projects that use this
MicrosoftSkillinit
Also in 3 projects that use this
Other versions39 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 39
- andrew-shwetzer/career-ops-plugin★ 501same name
- andrew-shwetzer/career-ops-plugin-do-not-fork-currently-updating-v2-★ 500same name
- cwinvestments/memstack★ 433same name
- buildomator/buildomator★ 102same name
- AsiaOstrich/universal-dev-standards★ 84same name
- mbwsims/claude-universe★ 71same name
- ratomirjovanovic/pandaward★ 50same name
- fivetaku/tikeytaka★ 40same name
- cfgaudit/cfgaudit★ 27same name
- thatmike1/claude-skills★ 10same name
29 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository27 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- alignSkill
- code-walkSkill
- conceptsSkill
- contributeSkill
- coverageSkill
- doc-gap-checkSkill
- enrichSkill
- enumerateSkill
- find-implicit-adrsSkill
- implementSkill
- ingest-docsSkill
- log-issueSkill
- navigateSkill
- next-batchSkill
- orientSkill
- panelSkill
- probe-defineSkill
- probe-runSkill
- probeSkill
- reflect-featureSkill
- retrieveSkill
- reviewSkill
- statusSkill
- streamsSkill
- test-coverageSkill
- triageSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
23 rules. 0 are common, 6 are distinctive, and 17 are very specific to this project.
0
Common
Most of the 34 edited versions we read keep these.
6
Distinctive
This author's own choices. Worth a look.
17
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 42
Read lineage/{substrate_repo}/concepts/index.yaml + relevant detail shards (only when scanner enables concepts.yaml consultation).
- Very specific: line 43
Read lineage/{substrate_repo}/shoebox/detail/SHB-*.md with Category: open | clustering (only when scanner enables shoebox consultation).
- Very specific: line 44
Read lineage/{substrate_repo}/doc-gaps/index.md + per-finding details — DEDUP/PRIORITY HINT ONLY.
- Very specific: line 47
If manifest exists: read it, pick next batch of not-scanned or changed-since-scan items
- Very specific: line 48
If no manifest exists: run enumeration first (follow /enumerate protocol inline), then pick first batch
- Distinctive: line 58
Follow the scanner's method systematically for EACH item in the batch
- Distinctive: line 65
Do NOT modify any files in target repos (read-only scan)
- Very specific: line 78
Compare
descriptionfield against the live doc: If the feature's description paragraph and the live doc materially disagree → emit a … - Very specific: line 86
If the finding pins a per-node defect (e.g.
- Very specific: line 87
If the same gap already exists as a DOC-GAP-NNN in doc-gaps.md: append a corroborated_by_scanner: block to its per-finding detail file — NO …
- Very specific: line 88
If the F-NNN's description disagrees with the chain (intent-vs-implementation drift) AND the disagreement is structural → append a NEW …
- Very specific: line 94
Coherence check: before emit: grep lineage/{substrate_repo}/{refactoring-scopes,doc-gaps,test-map,implicit-adrs}.md for any existing …
- Very specific: line 95
Verdict tagging: every emission tagged STATIC-INFERRED or PROBE-VERIFIED (rev-13 Rule 21 D7).
- Very specific: line 116
After writing, update the original finding file: append a - Cross-ref: enriched by F-NNN in \findings/{this-scanner-dir}/{file}.md\`` line …
- Very specific: line 124
Every file path discovered during scanning → add to relevant navigation/domains/*.md
- Very specific: line 125
Every repo name or integration pattern discovered (a new collector, a push-client, a platform module the scanner touched) → update …
- Distinctive: line 129
If scanner is now 100% covered, mark as completed
- Distinctive: line 132
Always emit (even if zero ontology clues were consumed — absence is informative).
- Distinctive: line 178
Always check coverage manifest before scanning — never re-scan already covered items
- Very specific: line 179
If manifest shows 100% coverage and no changed-since-scan items, report "fully scanned" and suggest re-enumeration if it's been >7 days
- Very specific: line 180
If the scanner path doesn't exist, list available scanners from scanners/ and ask which to run
- Distinctive: line 181
Prefer false positives over missed gaps (triager will filter later)
- Very specific: line 183
If an item can't be scanned (file missing, access issue), mark status as error with note
“Common” is measured: kept by at least 6 in 10 of the 34 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Run Scanner
Execute the audit scanner at $ARGUMENTS.
Protocol
- Orient — Read these files (do NOT skip):
CLAUDE.md(system overview — nine-gate Quality Bar and Gate 9 Source-of-Truth table)navigation/features.yaml(feature index)navigation/architecture.md(integration-pattern + canonical-repo map; every repo name encountered during the scan is classified against this file)- The relevant
navigation/domains/*.mdfor the scanner's target domain
- Load scanner definition at
$ARGUMENTS. Extract:- Target repo and scope
- Method (step-by-step)
- Criteria (what counts as a finding)
- Output format
- Chunking strategy (if applicable)
ontology_feed:block (if present) — determines run mode (see step 2.5)
2.5. Determine run mode (rev 13):
- If
ontology_feed:is absent ORontology_feed.enabled: false→ Mode A (standalone). Continue with steps 3-10 unchanged. This is backward-compatible — every existing scanner runs as before. - If
ontology_feed.enabled: true→ Mode B (ontology-fed). Execute mode B sub-protocol below before continuing to step 3. Mode is LOCKED for the run — never switched mid-session.
Mode B sub-protocol (rev 13, per APPROACH.md §20 + scanners/README.md "Ontology-fed mode")
2.5.1. Substrate-presence check: verify lineage/{substrate_repo}/ exists with manifest.yaml, nodes.jsonl, feature-flows/index.yaml. If ANY is missing → ABORT with INSUFFICIENT-ONTOLOGY-COVERAGE: substrate not found at lineage/{substrate_repo}/. NO silent fallback to mode A — explicit mode-A re-run (with the ontology_feed: block disabled or absent) is the only path to standalone behaviour.
2.5.2. Stamp ontology commit: read lineage/{substrate_repo}/manifest.yaml:current_substrate_commit (or equivalent — last_scan_commit). Record as ontology_commit_consulted: on every emission in this run. If current_substrate_commit < HEAD - staleness_threshold_commits (default 50) OR last-modified >48h, emit WARNING: stale-ontology — trust downgraded and set ontology_corroborated: stale_warning on every emission for this run. (Per-scanner override: ontology_feed.staleness_action: abort blocks the run instead.)
2.5.3. Load coverage check: enumerate intended F-NNN entries per the scanner's ontology_feed.feature_scope_filter: (e.g. pillar_id: [P-07] or target_repo_overlap). If the filtered set is empty AND the scanner declares feature_scope_filter: → ABORT with ZERO-FEATURES-IN-SCOPE. If the filtered set covers <50% of the maintainer-expected count for this scanner's scope → emit coverage_gap_for_scan: <missing entities> to the scan-feed log and continue.
2.5.4. Load clue sources in the order declared by ontology_feed.clue_sources: (always feature-flows/detail/F-*.yaml first per Rule 21):
- Glob
lineage/{substrate_repo}/feature-flows/detail/F-*.yaml; filter byfeature_scope_filter:; sort byfeature_id. This is the iteration set for steps 5.B. - Read
lineage/{substrate_repo}/concepts/index.yaml+ relevant detail shards (only when scanner enablesconcepts.yamlconsultation). - Read
lineage/{substrate_repo}/shoebox/detail/SHB-*.mdwithCategory: open | clustering(only when scanner enables shoebox consultation). - Read
lineage/{substrate_repo}/doc-gaps/index.md+ per-finding details — DEDUP/PRIORITY HINT ONLY. A feature absent from doc-gaps is NOT presumed documented.
- Check coverage manifest at
state/coverage/{scanner-id-dashed}.yaml:- If manifest exists: read it, pick next batch of
not-scannedorchanged-since-scanitems - If no manifest exists: run enumeration first (follow
/enumerateprotocol inline), then pick first batch - Batch size: 10-15 items per session (adjust based on item complexity)
- If manifest exists: read it, pick next batch of
- Load existing findings — Before scanning, read ALL existing findings files in
findings/:- Scan every
findings/*/directory (not just the current scanner's directory — gaps cross scanner boundaries) - Build a mental index of: finding ID, location, short title, severity
- During scanning, if you discover a gap that matches an existing finding (same location, same issue), do NOT create a duplicate — instead note an enrichment (see step 5)
- Match broadly: same file + same general issue = match, even if wording differs or the scanner that found it was different
- Scan every
Execute the scan on the selected batch:
- Follow the scanner's method systematically for EACH item in the batch
- Apply criteria to each item
- Every finding must cite a Source of Truth by Gate 9 class (Repo / Integration / Config / Builder / Spec / Term / Lifecycle / Dep / Handler / Cross-repo / Backlog / Ontology (rev 13, additive)) — full table in
pillars/{active}/gates.mdGate 9 + executable procedure inplaybooks/claim-inventory.md. "The doc says X but the code/SoT says Y" is a finding; "the doc might be wrong" is not. If you cannot cite an SoT, the gap is speculation — either find the SoT or drop the finding. - For mode B (rev 13): every finding sourced from an ontology clue cites BOTH the ontology and the underlying primary-source class — e.g.
Source-of-truth: Ontology[F-001:hop-1] → Repo[DataEntityDetails.tsx:56-64]. A finding citing ONLYOntology[...](without a primary-source class) is rejected by Gate 9 — same shape as Gate 9 already rejects "the doc might be wrong" without an SoT. TheOntologyclass is additive (12th class after the existing 11); it does NOT replace any existing class. - For SDK-backed integrations, run
playbooks/unset-parameter-audit.md(Gate 5) — every unset builder parameter with an unsafe SDK default is a finding (retrospectives/LSN-002is the canonical case). - For every outbound URL referenced in the doc under scan (
github.com/opendatadiscovery/*,docs.opendatadiscovery.org/*, external docs), resolve the URL againstnavigation/architecture.mdor (if missing) WebFetch /gh repo view. A broken or mis-targeted URL is a finding with SoT classRepoorIntegration(retrospectives/LSN-003is the canonical case). - Record findings as you go, each with its SoT citation inline
- Do NOT modify any files in target repos (read-only scan)
5.B. Mode B per-feature investigation (rev 13 — runs INSTEAD of step 5 when
ontology_feed.enabled: true)The scanner's PRIMARY investigation target is
lineage/{substrate_repo}/feature-flows/detail/F-*.yaml, not the scanner's traditional axes. For each in-scope F-NNN (the iteration set loaded in step 2.5.4):- Read F-NNN end-to-end. Extract:
feature_name,pillar_id,pillar_anchored_feature_name,description,contributing_nodes[],chain[].evidence,observed_vs_expected.facets[],status,seeded_from(if any). - Derive expected doc location. From
pillar_id+pillar_anchored_feature_name+system-mission.md's pillar-to-doc-path map. Check whether the page exists atdocumentation/docs/{expected_doc_path}and (if scanner is the canonical-accuracy variant) WebFetch the live URL. - Verify against code (4-tier clue ladder):
file_exists(sidecar saysX.java:139) → Read the cited file ±5 lines. Verified iff cited symbol present at cited line ±5.assertion_about_code(sidecar says "method returns 200 OK") → Read the cited region + verify mechanically (regex / direct match). Verified iff textual claim and read evidence agree.cross_layer_behaviour(chain-hop multiplicity, end-to-end DB-delta) → ifprobe_verifications:populated, run the cited probe; else file aprobe-neededfinding.doc_drift(sidecar'sdocs_link_semanticclaim) → fresh WebFetch in this session. Verified iff live status + key excerpt match sidecar within 30 days.
- Compare
descriptionfield against the live doc. If the feature'sdescriptionparagraph and the live doc materially disagree → emit adriftfinding. - Emit findings (per finding, cite Source-of-truth dual-class):
missing-pageif the doc doesn't exist.driftif description and live doc diverge.missing-caveatperobserved_vs_expected.facets[]entry not mentioned in doc.ontology-driftper hop whoseevidenceline moved or vanished (the code shifted since the sidecar was authored).
- Write back (mandatory; per scanners/README.md "Ontology-fed mode" + APPROACH.md §20.5):
- Append a
scanner_reviews:entry tofeature-flows/detail/F-{NNN}.yaml. APPEND-ONLY (never overwrite). Idempotency key:(scanner_id, scan_run_date). - If the finding pins a per-node defect (e.g.
@Valueconsumer with unsafe default) → also annotate the relevant sidecar underlineage/{substrate_repo}/understanding/*.md(max 5 sidecars per run; above that, the cluster belongs on the F-NNN flow). - If the same gap already exists as a DOC-GAP-NNN in
doc-gaps.md: append acorroborated_by_scanner:block to its per-finding detail file — NO new finding emitted (dedup). - If the F-NNN's description disagrees with the chain (intent-vs-implementation drift) AND the disagreement is structural → append a NEW shoebox thread at
lineage/{substrate_repo}/shoebox/detail/SHB-NNN-{slug}.mdwithCategory: open(the correction surface; feature-flow-builder picks it up on next pass).
- Append a
- Per-scan-run consultation budget enforced (hard caps from
ontology_feed.consultation_budget:):graph-retriever≤5: spawn for ad-hoc "does ontology cover {topic}?" queries beyond enumerated axes.feature-reflector≤3: spawn when a feature-flow's product framing disagrees with its chain AND the scanner can't independently judge.odd-sme≤2: spawn when pillar-mapping or industry-vocabulary alignment is ambiguous.- Above budget → backlog escalation entry (
escalation: pending-sme-review), never silent drop.
- Coherence check before emit: grep
lineage/{substrate_repo}/{refactoring-scopes,doc-gaps,test-map,implicit-adrs}.mdfor any existing finding matching the same(file:line)evidence. Duplicates extend via back-link, never create parallel entries. - Verdict tagging: every emission tagged
STATIC-INFERREDorPROBE-VERIFIED(rev-13 Rule 21 D7). A scan-run emitting onlySTATIC-INFERRED OKrows getsverification_class: descriptive-onlyand does not count toward "feature audited" status. ontology-confirmedvsontology-extendeddistinction: each finding marks whether it verified what the feature-flow already named (ontology-confirmed) or surfaced new behaviour beyond it (ontology-extended).
Mode B's findings file shape is the same as mode A; the
Source-of-truth:field carries the dual-class citation.- Write findings — Create the output file:
- Path:
findings/{scanner-id-dashed}/YYYY-MM-DD[-batch-N].md - Format: follow
scanners/README.mdoutput format exactly - Include summary counts at the top
- Note which specific items were covered in this run
- Dedup rules:
- New gap with no prior match → new finding ID (F-NNN), written normally
- Gap matches an existing finding from the SAME scanner → skip (already covered)
- Gap matches an existing finding from a DIFFERENT scanner → create an enrichment entry (see format below)
- Enrichment format (append to the findings file):
### F-NNN ← enriches F-XXX ({original-scanner-id}) - **Original**: F-XXX in `findings/{original-scanner-dir}/{file}.md` - **New evidence**: {what this scanner found that adds to the original} - **Severity adjustment**: {unchanged | escalate to X | de-escalate to X} — {reason} - After writing, update the original finding file: append a
- **Cross-ref**: enriched by F-NNN in \findings/{this-scanner-dir}/{file}.md`` line to the original finding
- Path:
- Update coverage manifest:
- For each item scanned: set
status: scanned, recordscanned_dateandscanned_commit - Set
findings_refto the findings file path - Recalculate
scanned_itemsandcoverage_pct
- For each item scanned: set
- Update navigation (MANDATORY):
- Every file path discovered during scanning → add to relevant
navigation/domains/*.md - Every repo name or integration pattern discovered (a new collector, a push-client, a platform module the scanner touched) → update
navigation/architecture.mdso future Gate 9 verifications can resolve the repo in O(1) without a fresh WebFetch - This is a core output, not optional bookkeeping
- Every file path discovered during scanning → add to relevant
- Update progress — Edit
state/PROGRESS.md:- If scanner is now 100% covered, mark as completed
9.5. Mode B only — write scanner-feed log (rev 13):
- Always emit (even if zero ontology clues were consumed — absence is informative).
- Path:
lineage/{substrate_repo}/scanner-feed/{YYYY-MM-DD}-{scan_run_id}.yaml. - Shape (per
adrs/drafts/research/scanner-ontology-fusion/INTEROP.md§2.5):artefact: scanner-feed scanner_id: <scanner id from frontmatter> scan_run_id: SR-{ISO8601-compact} scan_run_date: 2026-05-27 ontology_commit_consulted: <sha> mode: B # A | B | mixed clues_consumed: # ordered by consumption time - source: feature-flow id: F-NNN fields_read: [...] verified_against_code: true findings_produced: [F-NNN, F-MMM] - source: doc-gap id: DOC-GAP-NNN fields_read: [...] findings_produced: [] dedup_action: wrote_back_corroboration - source: concept | shoebox | graph-search ... agent_consultations: feature_reflector_calls: <int> odd_sme_calls: <int> graph_retriever_calls: <int> write_backs: feature_flows: [F-NNN, ...] sidecars: [<slug>, ...] doc_gaps: [DOC-GAP-NNN, ...] shoebox: [SHB-NNN, ...] warnings: - "<staleness / coverage gap / budget exceeded entries>" - Verification-class verdict (per Rule 21 D13): if
clues_consumed[]contains zerosource: feature-flowentries within scope → setverification_class: corroboration-onlyand emit a warning. The scan-run does NOT count toward "feature audited" status until the next mode-B run completes the per-feature iteration.
- Report:
- Items scanned this session: N
- Findings this session: M
- Coverage progress: X% (Y/Z items total)
- Remaining items: list next batch to scan
- Suggested: "Run
/scan $ARGUMENTSagain to continue (N items remaining)"
Rules
- Always check coverage manifest before scanning — never re-scan already covered items
- If manifest shows 100% coverage and no
changed-since-scanitems, report "fully scanned" and suggest re-enumeration if it's been >7 days - If the scanner path doesn't exist, list available scanners from
scanners/and ask which to run - Prefer false positives over missed gaps (triager will filter later)
- If you find a bug in target code, note it in findings but do NOT fix it. If the defect is clearly actionable upstream (concrete file:line, operator-visible failure mode, suggested fix obvious), the triager who picks up this scan will draft it as an
issues/{repo}/{PREFIX}-NNN.mdupstream issue (seeissues/README.mdand/log-issue); record enough detail in the finding (file:line, severity, repro shape) that the triager doesn't have to re-read the code. - If an item can't be scanned (file missing, access issue), mark status as
errorwith note
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

