agit search
Search through your neural commit history to find past reasoning and decisions.
Usage
agit search query <term>
agit search rebuild
Description
AGIT maintains a full-text search index of all reasoning traces. Use this to find:
- Past decisions and their rationale
- Similar problems you've solved before
- Context about why code was written a certain way
Subcommands
query
Search for entries matching a term:
agit search query <term>
rebuild
Manually rebuild the search index:
agit search rebuild
Options
| Option | Default | Description |
|---|---|---|
--limit, -l | 10 | Maximum results to return |
Examples
Search for Past Reasoning
agit search query "authentication"
Output:
Found 3 results:
[1] 2024-01-15 10:42:18 (score: 2.45)
Category: reasoning
Content: Will implement JWT authentication with refresh token rotation...
[2] 2024-01-10 14:33:02 (score: 1.82)
Category: intent
Content: Add OAuth authentication for third-party login...
[3] 2024-01-08 09:15:44 (score: 1.23)
Category: reasoning
Content: Decided against session-based authentication due to scalability...
Limit Results
agit search query "refactor" --limit 5
Rebuild Index
If search results seem stale or incomplete:
agit search rebuild
Output:
Rebuilding search index...
Indexed 247 entries from 52 commits
Search index rebuilt successfully
How It Works
Indexing
The search index is automatically updated:
- On commit: New entries are indexed after
agit commit - After pull: Incrementally indexes new commits from
agit pull
Storage
The search index is stored locally:
.agit/
└── search_index/ # Tantivy search index
This is not synced between machines. Each clone builds its own index.
What's Indexed
All fields from your reasoning traces:
- User intents
- AI reasoning
- Error notes
- Categories and timestamps
Troubleshooting
No Results Found
The index may be empty or outdated:
# Rebuild the index
agit search rebuild
Stale Results
After pulling new commits:
# Pull with auto-index
agit pull
# Or manually rebuild
agit search rebuild
Index Corruption
If you see errors, delete and rebuild:
rm -rf .agit/search_index
agit search rebuild
Use Cases
Finding Past Decisions
"Why did we use Redis instead of Memcached?"
agit search query "redis memcached"
Onboarding
Help new team members understand architectural choices:
agit search query "architecture decision"
Debugging
Find context about when a feature was implemented:
agit search query "caching layer"
See Also
- agit log - View commit history
- agit show - View full commit details
- Architecture - How the search index works