diff --git a/llm-utils/skills/llm-wiki/SKILL.md b/llm-utils/skills/llm-wiki/SKILL.md index 589f492..13b426d 100644 --- a/llm-utils/skills/llm-wiki/SKILL.md +++ b/llm-utils/skills/llm-wiki/SKILL.md @@ -43,19 +43,21 @@ any editor. No database, no special tooling required. ## Architecture: Three Layers ``` -wiki/ -├── SCHEMA.md # Conventions, structure rules, domain config -├── index.md # Sectioned content catalog with one-line summaries -├── log.md # Chronological action log (append-only, rotated yearly) -├── raw/ # Layer 1: Immutable source material -│ ├── articles/ # Web articles, clippings -│ ├── papers/ # PDFs, arxiv papers -│ ├── transcripts/ # Meeting notes, interviews -│ └── assets/ # Images, diagrams referenced by sources -├── entities/ # Layer 2: Entity pages (people, orgs, products, models) -├── concepts/ # Layer 2: Concept/topic pages -├── comparisons/ # Layer 2: Side-by-side analyses -└── queries/ # Layer 2: Filed query results worth keeping +llm-wiki/ +├── ingest/ # Input processing zone (raw files to parse) +├── wiki/ +│ ├── SCHEMA.md # Conventions, structure rules, domain config +│ ├── index.md # Sectioned content catalog with one-line summaries +│ ├── log.md # Chronological action log (append-only, rotated yearly) +│ ├── raw/ # Layer 1: Immutable source material (NEVER EDIT) +│ │ ├── articles/ # Web articles, clippings +│ │ ├── papers/ # PDFs, arXiv papers +│ │ ├── transcripts/# Meeting notes, interviews +│ │ └── assets/ # Images, diagrams referenced by sources +│ ├── concepts/ # Layer 2: Synthesized concept notes +│ ├── entities/ # Layer 2: Entity pages (people, orgs, products, models) +│ ├── comparisons/ # Layer 2: Side-by-side analyses +│ └── queries/ # Layer 2: Filed query results worth keeping ``` **Layer 1 — Raw Sources:** Immutable. The agent reads but never modifies these. @@ -92,8 +94,10 @@ at hand before creating anything new. When the user asks to create or start a wiki: -1. Determine the wiki path (from `$WIKI_PATH` env var, or ask the user; default `~/wiki`) -2. Create the directory structure above +1. Determine the wiki path (from `$WIKI_PATH` env var, or ask the user; default `~/llm-wiki`) +2. Create the directory structure above, including: + - `llm-wiki/ingest/` (input processing zone) + - `llm-wiki/wiki/raw/` subdirectories (`articles/`, `papers/`, `transcripts/`, `assets/`) 3. Ask the user what domain the wiki covers — be specific 4. Write `SCHEMA.md` customized to the domain (see template below) 5. Write initial `index.md` with sectioned header @@ -253,10 +257,10 @@ a `_meta/topic-map.md` that groups pages by theme for faster navigation. When the user provides a source (URL, file, paste), integrate it into the wiki: ① **Capture the raw source:** - - URL → use `web_extract` to get markdown, save to `raw/articles/` - - PDF → use `web_extract` (handles PDFs), save to `raw/papers/` - - Pasted text → save to appropriate `raw/` subdirectory - - Name the file descriptively: `raw/articles/karpathy-llm-wiki-2026.md` + - URL → use `web_extract` to get markdown, save to `wiki/raw/articles/` + - PDF → use `web_extract` (handles PDFs), save to `wiki/raw/papers/` + - Pasted text → save to appropriate `wiki/raw/` subdirectory (e.g., `transcripts/`) + - Name the file descriptively: `wiki/raw/articles/karpathy-llm-wiki-2026.md` - **Add raw frontmatter** (`source_url`, `ingested`, `sha256` of the body). On re-ingest of the same URL: recompute the sha256, compare to the stored value — skip if identical, flag drift and update if different. This is cheap enough to @@ -343,9 +347,9 @@ wiki = "" only a single source but has no confidence field set — these are candidates for either finding corroboration or demoting to `confidence: medium`. -⑧ **Source drift:** For each file in `raw/` with a `sha256:` frontmatter, recompute +- **Source drift:** For each file in `wiki/raw/` with a `sha256:` frontmatter, recompute the hash and flag mismatches. Mismatches indicate the raw file was edited - (shouldn't happen — raw/ is immutable) or ingested from a URL that has since + (shouldn't happen — `wiki/raw/` is immutable) or ingested from a URL that has since changed. Not a hard error, but worth reporting. ⑨ **Page size:** Flag pages over 200 lines — candidates for splitting.