Adrian Rogala
f9c7456900
Squashed replay of the original 9-commit branch onto current main. The original branch was 30+ commits behind, forked from before the skills refactor (PR #42) and workflow template expansion (PR #36), so a standard rebase hit heavy add/add conflicts on workflows/*.toml that are unrelated to the wiki hand. This replay keeps only the final hands/wiki/ tree state, which is the actual intent of the PR (the author iterated several times on the same files; squashing matches that). Implements the "LLM Wiki" pattern (Andrej Karpathy) for building a personal, Obsidian-compatible knowledge base. Instead of on-the-fly RAG, the wiki hand incrementally maintains a Markdown vault: hands/wiki/ ├── HAND.toml # hand manifest + [agents.*] sections ├── README.md # user-facing docs ├── SKILL-main.md # Librarian (coordinator) routing + FS ops ├── SKILL-ingestor.md # Source extraction + [[wikilink]] writing ├── SKILL-analyst.md # Synthesis with provenance citations └── SKILL-linter.md # Broken link / orphan / contradiction audit Closes librefang/librefang-registry#44 (via replay, not merge).
116 lines
4.1 KiB
Markdown
116 lines
4.1 KiB
Markdown
---
|
|
name: wiki-analyst
|
|
version: "1.0.0"
|
|
description: Wiki Analyst skills — synthesis methodology, gap identification, contradiction handling, and query resolution.
|
|
author: Leszek3737
|
|
tags: [wiki, knowledge-base, analyst, synthesis, reasoning]
|
|
runtime: prompt_only
|
|
---
|
|
|
|
# Wiki Hand — SKILL-analyst.md
|
|
|
|
## 1. Obsidian Conventions
|
|
|
|
### Wikilinks
|
|
- Internal cross-references: `[[page-name]]` (no path, no extension)
|
|
- Display text override: `[[page-name|Display Text]]`
|
|
- Never use markdown `[text](path)` for internal links
|
|
- External URLs: standard markdown `[text](https://...)`
|
|
- Image embeds: `![[filename.png]]`
|
|
- Images live in `raw/assets/`
|
|
|
|
### File Naming
|
|
- Format: `kebab-case.md`
|
|
- Max 40 characters (excluding `.md`)
|
|
- ASCII only — transliterate non-ASCII characters (ü → ue, ñ → n, ś → s, etc.)
|
|
- Entities: canonical name (`john-doe.md`, not `dr-john-doe-phd.md`)
|
|
- Concepts: noun phrase (`knowledge-base.md`, not `about-knowledge-bases.md`)
|
|
- Sources: derived from title (`the-future-of-ai.md`)
|
|
- Syntheses: derived from query (`tradeoffs-x-vs-y.md`)
|
|
- No version suffixes (_v2, _revised, _updated)
|
|
|
|
### Frontmatter
|
|
- Valid YAML between `---` delimiters at the top of every page
|
|
- All fields lowercase with underscores
|
|
- Dates: ISO 8601 (`YYYY-MM-DD`)
|
|
- Lists: YAML sequences (not comma-separated strings)
|
|
- Dataview-compatible: all fields are queryable
|
|
|
|
### Dataview Compatibility
|
|
Useful queries the user can run in Obsidian:
|
|
```dataview
|
|
TABLE confidence, source_count, last_updated
|
|
FROM "pages/entities"
|
|
SORT source_count DESC
|
|
```
|
|
```dataview
|
|
LIST
|
|
FROM "pages/concepts"
|
|
WHERE confidence = "disputed"
|
|
```
|
|
```dataview
|
|
TABLE claim_count, date_ingested
|
|
FROM "pages/sources"
|
|
SORT date_ingested DESC
|
|
```
|
|
|
|
---
|
|
|
|
## 4. Provenance and Confidence
|
|
|
|
### Inline Provenance Syntax
|
|
|
|
Every key claim (bullet point in Key Claims, Key Points, Key Facts sections) requires a provenance tag at the end of the line:
|
|
|
|
```markdown
|
|
- The company reported $10M ARR in Q4 2025 ([[annual-report-2025]], extracted)
|
|
- This suggests a 40% year-over-year growth rate (inferred)
|
|
- However, a later filing revised this to $8.5M ([[sec-filing-q1-2026]], extracted)
|
|
- The actual growth rate remains disputed ([[annual-report-2025]], [[sec-filing-q1-2026]], disputed)
|
|
```
|
|
|
|
| Tag | Meaning | When to use |
|
|
|-----|---------|-------------|
|
|
| `([[source]], extracted)` | Directly stated in the source | Verbatim facts, statistics, dates, names |
|
|
| `([[source-a]], [[source-b]], extracted)` | Corroborated across sources | Same fact confirmed independently |
|
|
| `(inferred)` | Derived by the LLM | Connections, implications, patterns not explicitly stated |
|
|
| `([[source-a]], [[source-b]], disputed)` | Sources contradict | Present both claims, let the user judge |
|
|
|
|
### Page-Level Confidence
|
|
|
|
Set in frontmatter `confidence` field:
|
|
|
|
| Level | Rule |
|
|
|-------|------|
|
|
| `high` | All key claims are `extracted` from 2+ corroborating sources |
|
|
| `medium` | Mix of extracted and inferred, OR all extracted from a single source |
|
|
| `low` | Primarily inferred, OR based on one unverified source |
|
|
| `disputed` | Contains at least one claim tagged `disputed` |
|
|
|
|
**Propagation in syntheses:** A synthesis page's confidence = the MINIMUM confidence of its consulted pages. If any consulted page is `disputed`, the synthesis must flag this.
|
|
|
|
### Confidence Update Triggers
|
|
- New source corroborates an existing claim → consider upgrading to `high`
|
|
- New source contradicts an existing claim → downgrade to `disputed`
|
|
- Source retracted or superseded → re-assess claims dependent on it
|
|
|
|
---
|
|
|
|
## 7. Synthesis Methodology
|
|
|
|
### Combining Multiple Sources
|
|
|
|
1. **Identify overlapping claims** — same fact from different sources
|
|
2. **Note corroboration** — claims supported by 2+ sources get higher confidence
|
|
3. **Surface contradictions** — present both sides with citations, do not silently pick one
|
|
4. **Fill complementary gaps** — where sources complement each other, weave together
|
|
5. **Maintain provenance chain** — every claim in the synthesis traces back to source pages
|
|
|
|
### Handling Contradictions
|
|
|
|
```markdown
|
|
|
|
## 2. Page Templates
|
|
|
|
## 12. Worked Examples
|