Search across all past agent sessions to reuse solved problems.
What Is CASS?
CASS (Coding Agent Session Search) indexes all your past agent conversations—Claude Code, Codex, Gemini, Cursor, and more—so you can find solutions to problems you've already solved.
It's like having a searchable memory of everything your agents have ever done across all projects.
Multi-Agent Index
Claude, Codex, Gemini, Cursor, ChatGPT sessions
Full-Text Search
Search across code, prompts, and responses
Cross-Project
Find solutions from any project or machine
Fast Retrieval
Instant results with context snippets
Why Use CASS?
You've likely solved many problems before with agents. Without CASS:
You hit an error you've seen before but can't remember the fix
Search CASS for the error message → find the exact solution
New project needs auth—you've implemented it before
Search 'authentication' → find your past implementation
Different agent solved a similar problem better
Search across all agents → find the best approach
Essential Commands
Important: Never run bare cass—it launches a TUI that may block your session. Always use --robot or --json.
Search Patterns
Basic Search
Find any mention of a term
cass search "database migration" --robot
Filter by Agent
Only search Claude Code sessions
cass search "error handling" --agent claude --robot
Recent Only
Search last 7 days
cass search "docker" --days 7 --robot
Minimal Output
Just essential fields
cass search "auth" --robot --fields minimal --limit 10
Error Messages
Find solutions to specific errors
cass search "Cannot read property of undefined" --robot
The Search Workflow
Search
Find relevant past sessions
Review
Check snippets and scores
Expand
View full context if needed
Apply
Use the solution in your current work
Understanding Output
CASS returns structured results with session info and snippets:
$ cass search "PostgreSQL connection" --robot --limit 2{"hits": [{"source_path": "/home/ubuntu/.claude/projects/.../session.jsonl","line_number": 87,"agent": "claude_code","workspace": "/projects/myapp","snippet": "...fixed the PostgreSQL connection by setting pool_size=20...","score": 0.92},{"source_path": "/home/ubuntu/.codex/sessions/2025-01-12.jsonl","line_number": 45,"agent": "codex","workspace": "/projects/backend","snippet": "...PostgreSQL connection string format: postgres://user:pass...","score": 0.85}],"_meta": { "query": "PostgreSQL connection", "took_ms": 42 }}
cass expand with the source path and line number to see the full conversation context!Best Practices
Use specific search terms
'PostgreSQL timeout error' is better than just 'error'
Filter by agent for focused results
If you remember which agent solved it, use --agent
Check multiple solutions
Different agents may have solved it differently
Use --days for recent context
Older solutions might use outdated patterns
Try It Now
1# Check your indexing status2$ cass health34# Search for a common pattern5$ cass search "import" --robot --limit 367# View full documentation8$ cass robot-docs guide