DEDUPEDWho it is forA developer whose repo has grown through many agent edit and fix cycles and who wants to see what could be cut before changing anything.
The problemRedundant tests, producer-verifies-producer checks, and fallback layers pile up, and it is hard to tell which ones still protect real behavior.
The outcomeA repository-wide list of high-confidence removals with evidence chains, plus a list of behavior the Skill chose to preserve, with no files changed.
What you need first
- A local git checkout of the target repository
- Codex or Claude Code installed
- Git, if you install the Skill from a local clone
Steps
- Open Codex and run
$skill-installerwith this URL: https://github.com/MrZoyo/deslop-GPT/tree/v0.3.4/skills/deslop - Start Codex in the target repository root.
- Run
$deslop deepto request a repository-wide read-only audit. - Review the evidence chain behind each HIGH item and each PRESERVE item.
- Do not run
applyyet. Keep the report for review.
Prompt for your agent
Using the deslop Skill from MrZoyo/deslop-GPT, run `$deslop deep` as a read-only audit of <target-repo-path>. Expected output: HIGH items with evidence chains and PRESERVE items with reasons. Verify with `git status` that no files were changed.
Caveat. The README says the Skill favors keeping code when evidence is incomplete, so the audit may list fewer removals than you expect.
Use this in your AI agent
Ask your agent
Using the Deduped MCP, get the setup steps for MrZoyo/deslop-GPT and "Audit an agent-built repo for removable bloat"
Add Deduped to Claude Code once
claude mcp add --transport http deduped https://dedupedai.com/mcp
Every Monday: the 5 hottest repos and exactly how to use them. Free, one email a week, unsubscribe anytime.
Get them freeShare: X LinkedIn Hacker News Email Copy link
Third-party repos we don't own. AI-written use cases can be wrong. Disclaimer