What it does
cachly indexes a project's git history to extract patterns, bug fixes, and architectural decisions, then exposes that knowledge to Claude, Cursor, Copilot, Windsurf, and other editors during coding sessions. It acts as a semantic cache layer that persists context across sessions and team members, reducing the need to re-explain shared knowledge on each request.
Who it's for
Engineers and teams whose AI context resets between sessions, or who repeatedly re-explain the same architectural patterns, deployment gotchas, and past bug fixes to different team members. Particularly valuable in codebases with tribal knowledge about why certain fixes were needed.
Common use cases
- Extract lessons from git history to brief Claude on deployment constraints before a feature task
- Preserve context about recurring bugs so the AI recalls the root cause without re-diagnosis
- Share troubleshooting patterns across team members' editing sessions
- Reduce LLM calls via semantic caching of project-specific knowledge
Setup pitfalls
- Reads entire filesystem and git history; ensure no credentials, API keys, or secrets are in commit history before enabling
- One secret was detected during analysis—review commits for accidentally-committed tokens before setup
- Requires browser-based authentication via device-flow on first run
- Sends project data to their service; verify compliance with your security and privacy requirements