# Pi-Wiki Feature Roadmap ## Current State (v1.2.0) ✅ **Working:** - `/wiki-status` - wiki health check - `/wiki-ingest ` - ingest sources (MD, HTML, JSON) - `/wiki-query ` - query wiki with keyword search - `/wiki-list` - list all pages - `/wiki-read ` - read specific page - `/wiki-lint` - check for orphans - Manifest with content hashes - Config resolution - Content trust boundary ❌ **Missing** (for obsidian-wiki parity): --- ## Priority 1: Core Maintenance (Low Effort) ### 1. Cross-linker **File:** `/wiki-crosslink` **Purpose:** Auto-discover unlinked mentions, insert `[[wikilinks]]` ``` Algorithm: 1. Load all page titles 2. For each page, scan body for title mentions without wikilinks 3. If mention found but no wikilink → suggest adding [[title]] 4. Update page with new wikilinks ``` ### 2. Tag Taxonomy **File:** `/wiki-taxonomy` **Purpose:** Canonical tags, audit and normalize across vault ``` Files: - _meta/taxonomy.md - canonical tag list with definitions - Tag audit: scan all pages, compare to taxonomy - Auto-fix: rename/merge tags to match taxonomy ``` ### 3. Append-and-Review **File:** `/wiki-append <note>` **Purpose:** Quick notes that get reviewed during lint ``` Flow: 1. /wiki-append "quick thought about X" 2. Stored in _drafts/append-notes.md 3. /wiki-lint flags drafts for review 4. User confirms → promoted to proper page ``` --- ## Priority 2: Obsidian Integration (Medium Effort) ### 4. Obsidian Vault Link **Purpose:** Use existing Obsidian vault OR pi-wiki's own vault ``` Config options: { "vault_path": "~/pi-wiki", // pi-wiki's own vault "obsidian_vault": "~/Obsidian/vault" // existing Obsidian vault } ``` **Features:** - Read from Obsidian vault (if configured) - Write to Obsidian vault (bi-directional sync) - Respect Obsidian's folder structure - Use Obsidian's image handling - Link with Obsidian's graph view ### 5. Obsidian CLI Integration **Purpose:** Use `obsidian` CLI for native operations ``` Commands to support: - obsidian read <file> - read note via CLI - obsidian search <query> - search via CLI - obsidian daily - today's note - obsidian properties - frontmatter ops ``` ### 6. Obsidian Web Clipper Support **Purpose:** Handle clips from browser extension ``` Save location: sources/web-clips/ Format: .md from Obsidian Web Clipper Auto-ingest on next /wiki-lint ``` --- ## Priority 3: Enhanced Search (Medium Effort) ### 7. QMD Semantic Search **Purpose:** Concept-level matches beyond keyword search ``` Setup: 1. npm install -g qmd 2. qmd index --name wiki ~/pi-wiki/wiki 3. qmd index --name sources ~/pi-wiki/sources Usage: - /wiki-query runs QMD pass before Grep - Falls back to keyword search if QMD unavailable ``` ### 8. Knowledge Graph Export **Purpose:** Visualize wiki structure ``` Formats: - graph.json (Obsidian graph view) - graph.html (interactive browser viz) - graph.graphml (Gephi) - cypher.txt (Neo4j) ``` --- ## Priority 4: History Mining (Medium Effort) ### 9. Pi Session History **Purpose:** Mine pi conversations into wiki ``` Source: ~/.pi/sessions/ Formats to process: - session JSON files - prompt/response pairs - Tool calls and outputs Skill: Extract key decisions, learnings, code patterns ``` ### 10. Daily Update **Purpose:** Automated maintenance cycle ``` Checks: 1. Freshness - pages needing review 2. Index update - reflect current state 3. Hot cache - frequently accessed pages 4. Draft promotion - _raw/ and append-notes ``` --- ## Priority 5: Advanced Features (High Effort) ### 11. Image Description Generation **Purpose:** Multi-modal source support ``` Flow: 1. Ingest detects image (png, jpg, webp) 2. LLM generates detailed description 3. Description embedded in source content 4. Page references both text + image ``` ### 12. Project `.brain/` Sync **Purpose:** Synced project context ``` Structure: projects/<name>/.brain/ ├── index.md → current state, priorities ├── architecture.md → stack, patterns ├── decisions.md → ADRs with rationale ├── changelog.md → diffs since last sync └── deployment.md → env, secrets, URLs Sync: /wiki-sync-project to update ``` ### 13. Research Skill **Purpose:** Autonomous web research, self-filed ``` Flow: 1. /wiki-research <topic> 2. Agent searches web, reads sources 3. Synthesizes findings 4. Files into wiki as new pages 5. Logs research session ``` --- ## Priority 6: Future (High Effort) ### 14. Team MCP Server **Purpose:** Shared wiki for teams ``` Architecture: - pi-wiki as MCP server - Team connects via MCP client - Shared knowledge base - Access control per namespace ``` ### 15. Model Fallback **Purpose:** Handle LLM failures gracefully ``` Flow: 1. Primary LLM (Claude/GPT-4) for compilation 2. If timeout/error → fallback to Haiku/Mini 3. Retry primary with smaller batch 4. Log failures for review ``` --- ## Feature Comparison Matrix | Feature | Current | Planned | obsidian-wiki | Effort | |---------|---------|---------|---------------|--------| | Basic ingest | ✅ | - | ✅ | - | | Manifest hashes | ✅ | - | ✅ | - | | Content trust | ✅ | - | ✅ | - | | Cross-linker | ❌ | P1 | ✅ | Low | | Tag taxonomy | ❌ | P1 | ✅ | Low | | Append-and-review | ❌ | P1 | ❌ | Low | | Obsidian vault link | ❌ | P2 | ✅ | Medium | | Obsidian CLI | ❌ | P2 | ✅ | Medium | | QMD search | ❌ | P3 | ✅ | Medium | | Graph export | ❌ | P3 | ✅ | Medium | | History mining | ❌ | P4 | ✅ | Medium | | Daily update | ❌ | P4 | ✅ | Medium | | Image descriptions | ❌ | P5 | ❌ | Medium | | Project brain sync | ❌ | P5 | ❌ | Medium | | Research skill | ❌ | P5 | ✅ | High | | Team MCP | ❌ | P6 | ❌ | High | | Model fallback | ❌ | P6 | ❌ | High | --- ## Quick Wins (Next Sprint) 1. **Cross-linker** - find unlinked mentions, add wikilinks 2. **Tag taxonomy** - canonical tags + audit 3. **Append-and-review** - quick notes pipeline 4. **Obsidian vault config** - point to existing vault 5. **Graph export** - JSON + HTML visualization --- ## Obsidian Integration Details ### Bi-directional Sync ``` pi-wiki ←→ Obsidian pi-wiki write → Obsidian vault: - Pages save to obsidian_vault/wiki/ - Images save to obsidian_vault/assets/ - Frontmatter matches Obsidian schema Obsidian write → pi-wiki: - Read via obsidian CLI - Monitor for changes (future: webhook) - Import into pi-wiki manifest ``` ### Obsidian Schema Compliance ```yaml # Frontmatter matching Obsidian --- title: Page Title type: entity|concept|summary|synthesis tags: [tag1, tag2] sources: [source-slug] created: 2026-01-15 updated: 2026-01-20 aliases: [Alternate Title] cssclasses: [callout] --- # Page Title ![[image.png]] > [!note] > Callout content ## See Also - [[Related Page]] - [[Another Page]] ``` ### Graph Color-Coding (Obsidian) ``` Color schemes: - by-tag: top 10 tags - by-category: folder structure - by-visibility: pii/internal - combined: visibility + tags Output: vault/.obsidian/graph.json ``` --- ## Implementation Order ``` Phase 1: Core (1-2 hours) ├─ Cross-linker ├─ Tag taxonomy └─ Append-and-review Phase 2: Obsidian Integration (2-3 hours) ├─ Vault link config ├─ Bi-directional sync └─ Graph export Phase 3: Enhanced Search (1-2 hours) ├─ QMD integration └─ Improved query ranking Phase 4: History (2-3 hours) ├─ Pi session mining └─ Daily update skill Phase 5: Advanced (4+ hours) ├─ Image descriptions ├─ Project brain ├─ Research └─ Team MCP ``` --- ## Test Plan ```bash # Core /wiki-ingest https://example.com/article.md /wiki-list /wiki-query "what did I read about X" /wiki-lint /wiki-crosslink /wiki-taxonomy # Obsidian /wiki-status # shows vault config /wiki-read "Page Name" # works with Obsidian vault # Search /wiki-query "concepts around X" # uses QMD if available # Notes /wiki-append "quick thought" /wiki-lint # flags append notes for review ``` --- ## Success Metrics - [ ] Cross-linker adds wikilinks to orphan pages - [ ] Tag taxonomy normalizes inconsistent tags - [ ] Append notes promote to proper wiki pages - [ ] Obsidian vault syncs bidirectionally - [ ] QMD semantic search finds concept matches - [ ] Pi history mines useful patterns - [ ] Daily update keeps wiki fresh