Repo Intelligence

Repo Intelligence

Automatic detection, profiling, and monitoring of repository characteristics. ClawDE understands your codebase so AI tools can work more effectively.

Overview

When you register a repository with ClawDE, the repo intelligence engine scans it to build a complete profile: what languages and frameworks you use, your coding conventions, your build tools, your test setup. This profile drives AI configuration generation and drift detection.

Capabilities

Feature	Description
Stack detection	Languages, frameworks, build tools, linters, CI systems
Convention inference	Naming patterns, file organization, code style
Dependency analysis	Cross-package and cross-repo dependency graphs
AI config generation	Generate .claude/, .codex/, .cursor/ configurations
Drift detection	Monitor for divergence between config and actual codebase
Drift scoring	Quantified metrics showing how much drift has accumulated
Validator engine	Pluggable lint, test, typecheck, format, build validators
CI mode	clawde ci verify for headless validation in CI pipelines

How It Works

1. Scan — Analyze the repo's files, package.json, Cargo.toml, etc.
2. Profile — Generate a RepoProfile with detected stack, conventions, and dependencies
3. Generate — Create or update AI tool configurations based on the profile
4. Monitor — Watch for changes that indicate drift from the profile
5. Report — Surface drift scores and improvement suggestions

AI Configuration Generation

ClawDE generates configuration for every major AI coding tool:

• Claude Code — .claude/CLAUDE.md, rules, settings
• Codex — .codex/AGENTS.md, configuration
• Cursor — .cursor/rules/*.mdc

Each generation respects overwrite/improve/skip semantics — you choose what ClawDE manages and what you control manually.

Drift Detection

Over time, codebases evolve. New dependencies appear, conventions shift, and AI configurations become stale. ClawDE's drift detector continuously monitors for:

• New languages or frameworks not reflected in AI configs
• Changed conventions that AI tools should know about
• Outdated validator configurations
• Missing or misconfigured tool settings