Wiki

Persistent, self-maintained markdown knowledge base for project and session knowledge. Inspired by Karpathy's LLM Wiki concept.

Operations

Ingest

Process knowledge into wiki pages. A single ingest can touch multiple pages.

wiki_ingest({ title: "Auth Architecture", content: "...", tags: ["auth", "architecture"], category: "architecture" })

Query

Search across all wiki pages by keywords and tags. Returns matching pages with snippets — YOU (the LLM) synthesize answers with citations from the results.

wiki_query({ query: "authentication", tags: ["auth"], category: "architecture" })

Lint

Run health checks on the wiki. Detects orphan pages, stale content, broken cross-references, oversized pages, and structural contradictions.

wiki_lint()

Quick Add

Add a single page quickly (simpler than ingest).

wiki_add({ title: "Page Title", content: "...", tags: ["tag1"], category: "decision" })

List / Read / Delete

wiki_list()           # Show all pages (reads index.md)
wiki_read({ page: "auth-architecture" })  # Read specific page
wiki_delete({ page: "outdated-page" })    # Delete a page

Log

View wiki operation history by reading .omc/wiki/log.md.

Categories

Pages are organized by category: architecture, decision, pattern, debugging, environment, session-log

Storage

• Pages: .omc/wiki/*.md (markdown with YAML frontmatter)
• Index: .omc/wiki/index.md (auto-maintained catalog)
• Log: .omc/wiki/log.md (append-only operation chronicle)

Cross-References

Use [[page-name]] wiki-link syntax to create cross-references between pages.

Auto-Capture

At session end, significant discoveries are automatically captured as session-log pages. Configure via wiki.autoCapture in .omc-config.json (default: enabled).

Hard Constraints

• NO vector embeddings — query uses keyword + tag matching only
• Wiki pages are git-ignored by default (.omc/wiki/ is project-local)

Model Routing

• haiku — quick lookups, lightweight inspection, narrow docs work
• sonnet — standard implementation, debugging, and review
• opus — architecture, deep analysis, consensus planning, and high-risk review
• fable — Claude Fable 5 (above Opus); pass it explicitly on the Task call or pin it per agent with agents.<name>.model
• The session model chosen with /model applies to the main loop only. Delegated agents run on their pinned tier unless the Task call passes model explicitly or the agent is overridden via agents.<name>.model. To run delegated work on Fable, use one of those two surfaces; selecting Fable in /model alone does not change delegation.