From llm-wiki
End-to-end vault workflow — status, research/fetch, raw prepare, wiki ingest, lint, build-site, validate — with optional user gates between stages.
npx claudepluginhub skinnnyjay/wiki-llm --plugin llm-wikiThis skill uses the workspace's default tool permissions.
Run the standard **llm-wiki** path: **Source → Ingest → Prepare → Maintain → Build → Validate**, with explicit **gates** where the user can review before continuing. This is **domain-specific** to the vault.
Creates new Angular apps using Angular CLI with flags for routing, SSR, SCSS, prefixes, and AI config. Follows best practices for modern TypeScript/Angular development. Use when starting Angular projects.
Generates Angular code and provides architectural guidance for projects, components, services, reactivity with signals, forms, dependency injection, routing, SSR, ARIA accessibility, animations, Tailwind styling, testing, and CLI tooling.
Executes ctx7 CLI to fetch up-to-date library documentation, manage AI coding skills (install/search/generate/remove/suggest), and configure Context7 MCP. Useful for current API refs, skill handling, or agent setup.
Run the standard llm-wiki path: Source → Ingest → Prepare → Maintain → Build → Validate, with explicit gates where the user can review before continuing. This is domain-specific to the vault.
See skills/references/pipeline-artifacts.md for what each stage writes and what reads it.
Progress in chat — At each stage boundary, say which stage you are entering (e.g. “Stage: raw prepare”, “Stage: wiki-ingest merge”). After each invoked skill or CLI, give a one-paragraph recap (what ran, what paths changed, exit OK or not) before moving on. Do not run the full pipeline without visible checkpoints unless the user explicitly asked for a silent / unattended run.
_meta.setup_completed is false, stop and offer wiki-setup.llm-wiki/.agent-memory.md if wiki-learn is in use; if memory.enabled, optionally llm-wiki memory recall for relevant past sessions (wiki-session-memory).$ARGUMENTS)Parse the user request:
| User says | Start at |
|---|---|
| (default) / “full run” | status → research → … → validate |
| “from raw” / “merge only” | prepare (if needed) → ingest → … |
| “from wiki” / “lint and ship” | lint → build → validate |
| “status only” | status only |
Stages are skipped when not applicable (e.g. no new raw/ files → skip prepare if user confirms raw is already clean).
| Order | Stage | Invoke | Gate (pause for user?) |
|---|---|---|---|
| 1 | status | wiki-status | Stop if vault missing or setup incomplete |
| 2 | research | wiki-research (topic/URL) or wiki-fetch (quick URL) | Gate: “N file(s) in raw/. Review before wiki merge?” — skip auto-continue if user wants to inspect |
| 3 | prepare | wiki-raw-prepare / llm-wiki raw validate / raw finish | Usually no gate if validation passes |
| 4 | ingest | wiki-ingest then wiki-maintainer | Gate: “Updated wiki pages: …. Run lint?” (or auto-continue if user asked for full unattended run) |
| 4b | kg update | llm-wiki kg rebuild (if knowledge_graph.auto_update_on_ingest) | No gate — runs automatically after ingest |
| 5 | lint | wiki-lint | Gate if issues found: list orphans/contradictions; offer fixes via wiki-maintainer before build |
| 6 | build | llm-wiki build-site | No gate on success |
| 7 | validate | llm-wiki validate and llm-wiki validate --wikilinks if wiki has links | Done |
Unattended mode: If the user says “run all” / “no stops”, only stop on status failure or validate failure; log gate decisions as skipped in wiki/log.md.
Run wiki-status integration + config checks. If setup is incomplete, offer wiki-setup and stop.
If the user supplied a topic, URL, or batch intent, run wiki-research or wiki-fetch per wiki-research dispatch rules. After new files land in raw/, apply gate unless unattended.
For messy HTML/PDF/OCR: wiki-raw-prepare + llm-wiki raw finish as in wiki-raw-prepare. If raw validate already passes with no LLM cleanup needed, skip. raw/ validation skips raw/memory/ (machine-generated session files).
Run wiki-ingest then wiki-maintainer. Append wiki/log.md with a pipeline | … entry. Gate unless unattended.
If knowledge_graph.auto_update_on_ingest is true, run llm-wiki kg rebuild after wiki merges; the CLI also runs kg rebuild after post_ingest on new raw/ files — see skills/references/mcp-and-kg.md § "KG auto-update after ingest".
Run wiki-lint. If issues exist, gate: propose fixes or run wiki-maintainer / new pages.
llm-wiki build-site
llm-wiki validate
llm-wiki validate --wikilinks # when wiki has wikilinks
Report pass/fail clearly.
wiki/log.md records the pipeline run (date, stages, notable files).wiki/.og/ is and how to serve it (see WORKFLOWS.md).Reads: llm-wiki/config.json, raw/, wiki/.
Writes: per pipeline-artifacts table; pipeline primarily orchestrates other skills.
llm-wiki validate and llm-wiki build-site as appropriate to the pipeline stage under test.Optional: skills/references/context-persona.md; persona.name in config (default Gennie).