Agent
Phase-4.5 summary re-narrator for the inverted pipeline. For each distilled page (concept / entity / person) that gained claims this run, rewrites the.
Agent
Phase-4.5 summary re-narrator for the inverted pipeline. For each distilled page (concept / entity / person) that gained claims this run, rewrites the.
Agent
Phase-7 concepts-outline lead-in narrator for the inverted pipeline (concept-map auto-refresh). For each theme group in wiki/concepts/index.md, (re)writes the per-theme lead-in paragraph the ENGINE owns — framing prose (why the theme matters + what to read first), never a bullet dump. Reads a bundle (per theme…
Agent
Phase-4.5 cross-lingual claim merger for the inverted pipeline. Reads the script-flagged cross-lingual (DE↔EN) candidate claim PAIRS on a distilled page — two claims that share an article-number digit anchor but did not auto-merge — and confirms which pairs are the SAME fact restated in another language. Writes a…
Agent
Phase-7 portal lead-in narrator for the inverted pipeline (Knowledge Portal auto-refresh, option 4b). For each theme that grew this run, (re)writes the per-theme lead-in paragraph the ENGINE owns in wiki/index.md — framing prose (why the theme matters + what to read first), never a bullet dump — plus the "state of the…
Agent
Phase-6 corrective revisor for the inverted pipeline. Reads /.metadata/verify-vN.json::deviations[] + /output/draft-vN.md + each deviation's cited page's preextractedclaims, and produces draft-v{N+1}.md plus a raw-text citation-records file the orchestrator serializes into citation-manifest.json (never hand-built…
Agent
Phase-4 zero-network ingest-time contradiction scorer for the inverted pipeline. Dispatched once per qualifying question group at knowledge-ingest Step 4.6, AFTER the Step 4.5 question-node emission. Reads the freshly-ingested NEW source pages' preextractedclaims and the related PEER pages' claim frontmatter, and…
Agent
Phase-2 source curator for the inverted pipeline. Reads a sub-question, runs WebSearch, scores candidates on 5 dimensions, then fetches each surviving candidate's body via WebFetch (Option B) through the shared fetch-cache. Emits a per-batch JSON array of candidate objects (each carrying a fetch sub-object) for merge…
Agent
Phase-3 cobrowse agent, dispatched by knowledge-fetch --cobrowse, which picks the mode via MODE. recover (default) recovers WebFetch-miss URLs through the claude-in-chrome extension, emitting {fetched[], unavailable[]} through fetch-cache.py. topup re-reads thin primary-tier URLs past WebFetch's cap, superseding the…
Agent
Phase-4 source ingester for the inverted pipeline. Reads one fetched-source entry from fetch-manifest.json, reads the cached body via fetch-cache.py fetch, dispatches claim-extractor to identify pre-extracted claims, and writes one wiki/sources/ .md page with type:source frontmatter populated by the claim array. An…
Agent
Phase-5 draft composer for the inverted pipeline. Reads wiki/index.md + selected wiki/sources/.md + prior wiki/syntheses/.md + distilled wiki/{concepts,entities}/.md + wiki/questions/.md, and writes /output/draft-vN.md plus a raw-text citation-records file the orchestrator serializes into…
Agent
Phase-7 zero-network contradiction scorer for the inverted pipeline. Reads the just-deposited /syntheses/ .md and each cited page's claim frontmatter, walks the synthesis body sentence-by-sentence against every claim, and emits /.metadata/contradictor-vN.json (schema 0.1.0) with findings carrying kind ∈…
Agent
Phase-7 zero-network structural quality reviewer for the inverted pipeline. Reads the latest /output/draft-vN.md + .metadata/plan.json (sub-questions, outputlanguage, targetwords, prosedensity) + .metadata/ingest-manifest.json (source diversity), scores the draft on 5 weighted structural dimensions (Completeness 0.25…
Agent
Phase-6 zero-network claim verifier for the inverted pipeline. Reads /output/draft-vN.md + a citation manifest (full or a shard via CITATIONSPATH) + each cited page's claim frontmatter (preextractedclaims on wiki/sources/ .md; distilledclaims on wiki/{concepts,entities}/ .md; answerclaims on wiki/questions/ .md …
Skill Claude CodeCodex
Phase 5 of the inverted pipeline. Reads /.metadata/plan.json + /.metadata/ingest-manifest.json + the populated cogni-wiki, dispatches a wiki-composer pass (plus ONE bounded fail-soft zero-network coverage-gated re-dispatch when a sub-question has ingested evidence the draft left uncited), and lands /output/draft-vN.md…
Skill Claude CodeCodex
Phase 2 of the inverted pipeline. Reads plan.json, dispatches one source-curator agent per sub-question to WebSearch + score candidate sources AND fetch each survivor's body via WebFetch into the shared fetch-cache, then merges the per-sub-question batches into candidates.json via candidate-store.py. Each candidate…
Skill Claude CodeCodex
Render an HTML dashboard for a bound cogni-knowledge base — runs the vendored wiki-dashboard scripts (renderdashboard.py + buildgraph.py) natively against the bound wiki and writes a knowledge-overlay.md sidecar that lists deposited research projects and the latest lint-audit claimdrift count. Use this skill whenever…
Skill Claude CodeCodex
Phase 4.5 of the inverted pipeline (between ingest and compose). Distills the run's source claims into recurring type:concept / type:entity / type:person pages, creating-or-merging them under a lock with claim-level dedup so the bound wiki compounds across runs (distilled pages get enriched, not duplicated). Also…
Skill Claude CodeCodex
Phase 3 of the inverted pipeline. Builds fetch-manifest.json from the fetch results the curators already produced in Phase 2 (bodies are already fetched during curate). Cobrowse is OPT-IN (--cobrowse): the skill walks the user through enabling the Claude-in-Chrome extension, then dispatches source-fetcher sequentially…
Skill Claude CodeCodex
Phase 7 of the inverted pipeline — deposits the verified draft as a wiki synthesis page, closing the compounding loop. Reads the latest verified draft plus its verify and citation manifests, refuses self-citing loops, atomically writes the synthesis page into the bound wiki, updates the surrounding index, log, and…
Skill Claude CodeCodex
Run a read-only structural health check on a cogni-knowledge base — page/link/schema integrity plus entries-count and claim drift for the bound wiki. Use this skill whenever the user says 'check knowledge health', 'knowledge health', 'is my knowledge base healthy', 'audit the knowledge base structure', 'knowledge…
Skill Claude CodeCodex
Rebuild the curated root index + per-type sub-indexes of a cogni-knowledge base on demand, migrate an EXISTING pre-0.0.8 wiki to the curated layout (control files into wiki/meta/, overview folded into the index intro, flat root split into root-map + sub-indexes, schema 0.0.7→0.0.8), or repair drifted machine-owned…
Skill Claude CodeCodex
Standalone single-source ingest for cogni-knowledge — deposit ONE source directly into the bound wiki with no research run: a web page or PDF URL, a local file (.docx/.html/.txt), pasted text, a local PDF, or a local interview note. Lands one wiki/sources/ .md page (or wiki/interviews/ .md with type: interview for an…
Skill Claude CodeCodex
Phase 4 of the inverted pipeline. Reads fetch-manifest.json, dispatches source-ingester per fetched source to write wiki/sources/ .md pages with preextractedclaims:, merges per-source results into ingest-manifest.json, then consolidates the wiki integration (backlinks, thematic index, question nodes, sub-indexes) into…
Skill Claude CodeCodex
Run semantic lint on a cogni-knowledge base — surface stale pages/drafts, claim drift, and broken reverse links, optionally repairing the mechanical classes with --fix. Use this skill whenever the user says 'lint the knowledge base', 'knowledge lint', 'fix knowledge drift', 'clean up the wiki', 'repair reverse links'…