paperclip-plugin-codrag

<p align="center"> </p>

> Give every Paperclip agent deep structural codebase knowledge.

CoDRAG indexes your codebase with embeddings, a structural graph, and epistemic analysis. This plugin makes that intelligence available to every agent in your Paperclip instance — no per-agent configuration needed.

AI agents working on code operate blind. They grep files, read READMEs, and hope they stumble onto the right architecture. When they modify a file, they don't know what depends on it. When they search for code, they match keywords instead of meaning. When they finish a task, the next agent starts from scratch.

CoDRAG pre-computes your codebase's structural graph — modules, dependencies, hub files, domain clusters — and serves it on demand. The Paperclip plugin registers 5 tools that any agent can call during their runs:

| Tool | What It Does | |------|-------------| | codrag:context | Structural overview — modules, hub files, atlas. Call at the start of every task. | | codrag:search | Semantic code search. Finds code by meaning, with structural trace expansion. | | codrag:impact | Blast radius analysis. See exactly what depends on a file before changing it. | | codrag:audit | Codebase health findings — tech debt, dead code, architecture issues. | | codrag:observe | Cross-session memory. Agents save observations for the next agent. |

This is epistemic-first agent orchestration — instead of agents discovering codebase structure through trial and error, CoDRAG pre-computes it and serves it on demand:

• Pre-computed structure — Module clusters, import chains, and hub files are already indexed
• Role-scoped context — Backend agents see backend files. Frontend agents see frontend files
• Impact-aware changes — Agents know the blast radius before modifying code
• Persistent memory — Ob