Source: www.producthunt.com/products/cleanslop
CleanSlop maps a codebase's call paths, module dependencies, background jobs, data stores and external services. It then flags issues that work in a demo but fail with real users, such as unbounded work, slow operations blocking requests and jobs that are silently dropped. The team says it has retrained its scanning engine to produce a shorter list of risks and cut scan times by up to 99 percent.
The new version runs from a single line pasted into Claude Code, Codex, Cursor, Grok Bot or Muse. The agent signs in using Link's attestation that a real person stands behind it, with no sign-up form or password, then scans the current folder, checks findings against the code, applies fixes and rescans. When scans run out, the agent offers a credit pack or monthly plan, which the user approves in Link, and the agent never sees the card.
Support is limited to TypeScript and Node.js apps, with the best results on Next.js and Supabase. Link identity is available in the US and Canada; elsewhere users can sign in with GitHub, which the team says can upload local code without a repository. CleanSlop also offers a CLI and an MCP server.
Why it matters if you build agents
Agents that generate or modify services can now prove that a human is accountable for them without a form or password, which is a pattern worth copying for other agent-facing sign-ins. Link identity currently works only in the US and Canada, so agents serving other regions will need the GitHub fallback. Scanning is limited to TypeScript and Node.js today.
Get stories like this every weekday
One short email with the day's agent tooling news.