Bootstrap Knowledge - Build Project Knowledge
Scans project, generates persona-specific knowledge, creates vector index for code search.
Inputs
| Input |
Type |
Default |
Purpose |
project |
string |
auto |
Project from config.json (auto from cwd) |
personas |
string |
"developer,devops,tester,release" |
Comma-separated |
deep_scan |
bool |
false |
Deeper analysis (slower) |
create_vector_index |
bool |
true |
Create semantic code index |
start_watcher |
bool |
true |
Auto-update index on file changes |
Workflow
1. Detect Project
- If not provided: detect from cwd via config.json repositories
- Validate project exists in config and path exists
2. Check Known Issues
check_known_issues("code_index", "") - vector indexing issues
3. Knowledge Scan per Persona
- For each in personas_list:
knowledge_scan(project, persona, force=false)
- Personas: developer, devops, tester, release
4. Create Vector Index
code_index(project) - semantic embeddings for code search- Parse result: files indexed, chunks created
5. Start File Watcher
code_watch(project, action="start") - auto-reindex on changes- Only if vector index created successfully
6. Deep Scan (if enabled)
code_search(query="class definition pattern implementation", project, limit=10)code_search(query="exception handling try except raise error", project, limit=5)- Read README.md (first 5000 chars)
- Scan tests/ for test_*.py, *.test.js, conftest.py
7. Build Summary
- List generated files:
memory/knowledge/personas/<persona>/<project>.yaml
- Vector index stats: files, chunks, watcher status
- Deep scan: README length, test counts, patterns, error handlers
8. Learn from Failures
- If "permission denied":
learn_tool_fix("code_index", "permission denied", "Cannot write to vector cache", "Check ~/.cache/aa-workflow/vectors/")
- If "out of memory": learn fix for memory
- If "already running": note watcher is active
9. Log and Track
memory_session_log("Bootstrapped knowledge for <project>", "Personas: X, Vector: Y")- Update state/knowledge with project, last_bootstrap, vector_indexed
10. Output Format
## 📚 Knowledge Bootstrap Complete: <project>
**Project Path:** `path`
**Personas:** developer, devops, tester, release
### Generated Knowledge Files
- memory/knowledge/personas/developer/<project>.yaml
- ...
### 🔍 Vector Index Created
- Files indexed: N
- Chunks created: N
- File watcher: ✅ Running
### Next Steps
1. knowledge_query(project='...', persona='developer')
2. code_search(query='...', project='...')
3. knowledge_update() for gotchas
Key Details
- Chains to:
knowledge_refresh, learn_architecture
- Provides context for:
gather_context, review_pr, start_work
- Project path from config.json repositories
1---2name: bootstrap-knowledge3description: Scan a project and generate comprehensive knowledge for all personas. Creates semantic vector index, deep analysis, file watcher. Use when user says "bootstrap knowledge", "learn architecture", or "scan project structure".4---56# Bootstrap Knowledge - Build Project Knowledge78Scans project, generates persona-specific knowledge, creates vector index for code search.910## Inputs1112| Input | Type | Default | Purpose |13|-------|------|---------|---------|14| `project` | string | auto | Project from config.json (auto from cwd) |15| `personas` | string | "developer,devops,tester,release" | Comma-separated |16| `deep_scan` | bool | false | Deeper analysis (slower) |17| `create_vector_index` | bool | true | Create semantic code index |18| `start_watcher` | bool | true | Auto-update index on file changes |1920## Workflow2122### 1. Detect Project23- If not provided: detect from cwd via config.json repositories24- Validate project exists in config and path exists2526### 2. Check Known Issues27- `check_known_issues("code_index", "")` - vector indexing issues2829### 3. Knowledge Scan per Persona30- For each in personas_list: `knowledge_scan(project, persona, force=false)`31- Personas: developer, devops, tester, release3233### 4. Create Vector Index34- `code_index(project)` - semantic embeddings for code search35- Parse result: files indexed, chunks created3637### 5. Start File Watcher38- `code_watch(project, action="start")` - auto-reindex on changes39- Only if vector index created successfully4041### 6. Deep Scan (if enabled)42- `code_search(query="class definition pattern implementation", project, limit=10)`43- `code_search(query="exception handling try except raise error", project, limit=5)`44- Read README.md (first 5000 chars)45- Scan tests/ for test_*.py, *.test.js, conftest.py4647### 7. Build Summary48- List generated files: `memory/knowledge/personas/<persona>/<project>.yaml`49- Vector index stats: files, chunks, watcher status50- Deep scan: README length, test counts, patterns, error handlers5152### 8. Learn from Failures53- If "permission denied": `learn_tool_fix("code_index", "permission denied", "Cannot write to vector cache", "Check ~/.cache/aa-workflow/vectors/")`54- If "out of memory": learn fix for memory55- If "already running": note watcher is active5657### 9. Log and Track58- `memory_session_log("Bootstrapped knowledge for <project>", "Personas: X, Vector: Y")`59- Update state/knowledge with project, last_bootstrap, vector_indexed6061### 10. Output Format6263```markdown64## 📚 Knowledge Bootstrap Complete: <project>6566**Project Path:** `path`67**Personas:** developer, devops, tester, release6869### Generated Knowledge Files70- memory/knowledge/personas/developer/<project>.yaml71- ...7273### 🔍 Vector Index Created74- Files indexed: N75- Chunks created: N76- File watcher: ✅ Running7778### Next Steps791. knowledge_query(project='...', persona='developer')802. code_search(query='...', project='...')813. knowledge_update() for gotchas82```8384## Key Details8586- **Chains to:** `knowledge_refresh`, `learn_architecture`87- **Provides context for:** `gather_context`, `review_pr`, `start_work`88- Project path from config.json repositories
dmzoneill