AI coding agents lose context the moment a task spans more than one repository. Polygraph is a new tool that gives agents cross-repo visibility and persistent session memory — two capabilities most IDE-integrated AI tools still lack.
The core pitch solves a specific pain point: when an agent working in Repo A needs to understand types, APIs, or conventions defined in Repo B, it currently can’t. Polygraph bridges that gap with two main features:
- Cross-repo awareness — agents can query and reference code across multiple repositories in a single session, rather than being siloed to whatever project is open
- Session memory — context from previous interactions persists, so agents don’t start from zero every time you open a new chat or restart a task
Both features target teams working with microservices, monorepo-adjacent setups, or any architecture where business logic is split across multiple codebases. If your agent can only see one repo at a time, it’s working with partial information — and partial information produces partial solutions.
NOTE
Polygraph surfaced on Hacker News as a Show HN with limited public detail on pricing, supported agents, or integration specifics. The landing page at trypolygraph.com is the primary source — expect the feature set to evolve quickly at this stage.
The timing is relevant. Cursor, Claude Code, and Copilot Workspace all handle single-repo context reasonably well, but multi-repo orchestration remains a gap. Tools like Polygraph are betting that as agents take on larger tasks — refactoring shared libraries, updating API contracts across services — cross-repo context becomes a prerequisite, not a nice-to-have.
Polygraph is early-stage and worth watching if you run multi-repo architectures where AI agents keep hallucinating imports or missing shared types. Check trypolygraph.com for access.