skills/wiki-curator/SKILL.md
Curate this repo's hallouminate wiki (.hallouminate/wiki/, the repo:dotfiles:wiki corpus) — add or update architecture pages, per-harness docs, and gotchas. Use when the user says "update the wiki", "document this in the wiki", "refresh the harness docs", "add a wiki page", "curate the wiki", "the wiki is stale", or invokes /wiki-curator. Also use at session end to write back a non-obvious decision or gotcha worth preserving. Grounds the existing wiki first, follows one-topic-per-file conventions, verifies every external doc URL before writing, and reindexes. Do NOT use for general code search (that is cheez-search) or for editing AGENTS.md command reference.
npx skillsauth add paulnsorensen/dotfiles wiki-curatorInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Maintain the repo:dotfiles:wiki corpus at .hallouminate/wiki/. The wiki holds the cross-session why behind the what — architecture rationale, harness wiring, gotchas. AGENTS.md stays the command/structure reference; this skill never duplicates it.
Skip if the knowledge is already in the code, git history, AGENTS.md, or an existing wiki page (update that page instead of adding a duplicate).
Always read before writing — never author blind.
list_tree the repo:dotfiles:wiki corpus to see the current structure.ground (semantic search) or read_markdown the page(s) you're about to touch. Call read_markdown before any overwrite so you preserve what's there.The chunker splits on headings, so two unrelated topics in one file degrade retrieval. Map the knowledge to a single page:
.hallouminate/wiki/architecture/ (agents-dir.md, agent-profile.md)..hallouminate/wiki/harnesses/<harness>.md.index.md.If the new knowledge is a sub-topic of an existing page, add a heading there. If it's a distinct topic, make a new file.
Two hard rules that keep the wiki trustworthy:
WebFetch it and confirm it resolves with on-topic content. A removed link beats a dead/wrong one.cheez-search / tilth_read). Don't assert wiring you haven't confirmed. Tag genuine uncertainty with `<speculative>` rather than stating it flat.AGENTS.md say what things do. Capture rationale, trade-offs, "this not that", and gotchas.[[name]] (the page's path-stem, e.g. [[agent-profile]], [[../harnesses/index]]).add_markdown (overwrite: true for an existing file): it writes atomically AND refreshes ancestor index.md link trees + the LanceDB index in one step.hallouminate index afterward so the changes are searchable. Verify with a ground query.index.md Sections list pointing at the files under it.harnesses/index.md carries a capability support matrix (hooks / sub-agents / MCP / system prompt / settings / skills / isolated launch) and a "this repo → harness" mapping table. When a harness page changes, reconcile the matrix. When the upstream tool ships a new capability, add the row + the verified doc link on the harness page first, then the matrix.
The five harness pages are self-contained reference: each has a Capability | Official doc | This repo table. The official doc column must be live URLs (rule in step 3); the "This repo" column points at the ap renderer / registry that wires it (see [[architecture/agent-profile]]).
.hallouminate/wiki/ and .hallouminate/config.toml are git-tracked (carved out of the .hallouminate/ gitignore); the rest of .hallouminate/ (LanceDB caches) is local scratch. Wiki edits are committable changes — treat them like code.add_markdown may target the main checkout's corpus path rather than the worktree — prefer plain file writes + hallouminate index when working in a worktree, or pass an explicit corpus.tools
Reconstruct what a past coding-agent session was doing so you can resume it — goal, files touched, last verified state, and the next step — by querying the session logs. Use when the user says "what was I working on", "recover that session", "reconstruct where I left off", "resume my last session", "what did that session change", "rebuild context from logs", or invokes /work-recovery. Report-only — it never scores or judges. Do NOT use for usage scoring (that is /skill-improver, /tool-efficiency, /prompt-analytics) or one-off interactive log queries (that is /session-analytics).
tools
Audit how a tool, command, or MCP server is actually used across coding-agent sessions and produce calibrated recommendations — tool-vs-task fit, error forensics, fix recommendations, permission friction, MCP health, and token economics. Use when the user says "tool efficiency", "am I using X efficiently", "audit tool usage", "why does X keep failing", "how do I fix this error", "what should I change", "permission friction", "is this MCP worth it", "tool error rate", "fix recommendations", or invokes /tool-efficiency. Do NOT use for auditing a skill or agent definition (that is /skill-improver) or for one-off interactive log queries (that is /session-analytics).
tools
Analyze how prompts and skill routing behave across coding-agent sessions and produce calibrated recommendations — prompt-pattern analysis, routing accuracy, and knowledge gaps. Use when the user says "analyze my prompts", "prompt patterns", "is routing working", "which skill should have fired", "knowledge gaps", "what do I keep asking", or invokes /prompt-analytics. Do NOT use for auditing a single skill/agent definition (that is /skill-improver), tool/MCP efficiency (that is /tool-efficiency), or one-off interactive log queries (that is /session-analytics).
development
Diagnose and self-heal harness-config drift between live files (~/.claude, ~/.codex, opencode, Cursor, Copilot) and what `ap` renders from the dotfiles registries. Use when the user says "harness doctor", "check my harness config", "settings drifted", "why is this hook firing twice", or asks to audit agent config. In Codex, invoke via `$harness-doctor` or `/skills`, not `/harness-doctor`. Do NOT use for general code review (/age), single-file permission cleanup (/settings-clean), or app-level debugging.