revert: remove codebase intelligence system

Rolled back the intel system due to overengineering concerns:
- 1200+ line hook with SQLite graph database
- 21MB sql.js dependency
- Entity generation spawning additional Claude calls
- Complex system with unclear value

Removed:
- /gsd:analyze-codebase command
- /gsd:query-intel command
- gsd-intel-index.js, gsd-intel-session.js, gsd-intel-prune.js hooks
- gsd-entity-generator, gsd-indexer agents
- entity.md template
- sql.js dependency

Preserved:
- Model profiles feature
- Statusline hook
- All other v1.9.x improvements

-3,065 lines removed

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
This commit is contained in:
Lex Christopherson
2026-01-21 10:28:53 -06:00
parent 93b963c3fe
commit d1fda80c7f
20 changed files with 17 additions and 3065 deletions

View File

@@ -324,50 +324,6 @@ Use for: bug fixes, small features, config changes, one-off tasks.
## Why It Works
### Codebase Intelligence
GSD learns your codebase patterns automatically. As Claude writes code, a PostToolUse hook indexes exports and imports, detects naming conventions, and builds a semantic understanding of your codebase.
**How it works:**
1. **Automatic learning** — Every time Claude writes or edits a JS/TS file, the hook extracts exports/imports and updates `.planning/intel/index.json`
2. **Convention detection** — Analyzes exports for naming patterns (camelCase, PascalCase, etc.), identifies directory purposes, detects file suffixes
3. **Graph database** — Stores entity relationships in SQLite for dependency analysis
4. **Context injection** — At session start, injects a summary into Claude's context so it knows your codebase structure and conventions
**For existing codebases:**
```
/gsd:analyze-codebase
```
Performs a bulk scan of your codebase to bootstrap the intelligence layer. Works standalone — no `/gsd:new-project` required. After initial analysis, hooks continue incremental learning.
**Query the graph:**
```
/gsd:query-intel dependents src/lib/db.ts # What depends on this file?
/gsd:query-intel hotspots # Most-depended-on files
```
**Files created:**
| File | Purpose |
|------|---------|
| `.planning/intel/index.json` | File exports and imports index |
| `.planning/intel/conventions.json` | Detected patterns (naming, directories, suffixes) |
| `.planning/intel/graph.db` | SQLite database with entity relationships |
| `.planning/intel/summary.md` | Concise context for session injection |
**Benefits:**
- Claude follows your naming conventions automatically
- New files go in the right directories
- Consistency maintained across sessions
- Query blast radius before refactoring
- Identify high-impact hotspot files
- No manual documentation of patterns needed
### Context Engineering
Claude Code is incredibly powerful *if* you give it the context it needs. Most people don't.
@@ -478,8 +434,6 @@ You're never locked in. The system adapts.
| Command | What it does |
|---------|--------------|
| `/gsd:map-codebase` | Analyze existing codebase before new-project |
| `/gsd:analyze-codebase` | Bootstrap codebase intelligence for existing projects |
| `/gsd:query-intel <type>` | Query dependency graph (dependents, hotspots) |
### Phase Management