L
lenamonj/jeffy-loop
A Claude Code loop that finds bugs in your repo, fixes them, and checks every fix before it commits.
★ 13·updated 2026-11-19
View on GitHub ↗Direct link · no redirect page
After install: how to tell it works
- What you see after install — The command exits 0 and the terminal shows no "command not found" or "permission denied".
- One-line self-check — Run the smallest usage example from that repository's README. If the output matches what the README describes, it is working.
- If that does not work — This project has no troubleshooting notes. Search its issues for "install".
The install command comes from the repository; these checks describe observable results only and do not replace the project's own docs.
- Star count
- 13 ★ · lenamonj/jeffy-loop · read from the GitHub search API · (checked 2026-11-19 20:00:48)
- Source
- read directly from its GitHub page · type: open-source project
- Harvest history
- first seen 2026-11-20 · last seen 2026-11-20 · seen 1× · listed by 1 sources
- Score
- score 103 · 13 ★ · 1 sources · repo
- Install
- see Install & config above
FAQ (answers come from our records)
Who is it for?
We filed it under these tags: Open Source. The tags come from the entry itself, not a taxonomy we imposed.
Who maintains it?
repository
lenamonj/jeffy-loop;first catalogued 2026-11-20;last seen 2026-11-20。There is no official certification or quality rating here — we record sources and dates only.README excerpt (English, untranslated) · Read the full README on GitHub →
%20behavioural%20checks&replace=%241%2B%20per%20host&label=behavioural%20checks&
Excerpted from the repository README (abridged)
Quick Install · Usage · How it works · RSI · The receipts · Headless · White paper · STE100 description
## A Claude Code loop that finds bugs in your repo, fixes them, and checks every fix before it commits. Maintainers have merged 79 of its patches.
The model writes the patch. The harness around it decides whether that patch can be trusted. Jeffy Loop is a verification harness for Claude Code: type /jeffy and it audits your codebase, finds a problem, fixes it, checks the fix, and commits the result locally, one task at a time. It keeps going until the codebase passes a fresh audit.
The harness is deliberately hard to fool. If a fix breaks your tests, the loop is told to revert it. The loop is told never to push, and the hook never pushes; keep push off its allowlist. A standard run counts as complete only when a fresh audit is clean, your tests pass again, and an adversarial review by a second AI agent with no part in the run signs off. The goal is fixes that survive scrutiny.
There is a real-world test for that. Anthropic fixed a security issue this loop found in claude-code-action, and closed the report as Resolved after we retested the fix. Maintainers with no stake in this project have merged 79 of its pull requests, each filed from a local clone, into projects run by NVIDIA, Arm, Meta, Tesla, Google, Apple, Microsoft, Netflix, Apache, Palo Alto Networks, Oracle, IBM, HashiCorp, Cisco, Uber, Square, Cloudflare, Dropbox, Tencent, and more. See them all.
Sending a fix upstream? If you use Jeffy Loop to open pull requests on repositories you do not own, we strongly recommend running housebroken (pip install housebroken-cli) alongside it. Jeffy proves the fix; housebroken gets it through the maintainer's door. It reads the project's AI policy first, checks that the finding is not already fixed or ruled intended, holds the pull request to one finding and the smallest diff, has a second model attack it, and refuses to file until every gate has passed. Every pull request this project has filed since housebroken shipped went through it.
## The record
Jeffy was run against 132 open-source projects with no connection to this repository, each judged by its own test suite. Every one of these runs is published, failures included.
| Projects tested | Converged | Failed | PRs merged | PRs open | Issues filed |
| 132 | 103 | 28 | 79 | 10 | 3 |
Converged: the closing audit came back clean and the loop's adversarial evaluator, a fresh-context sub-agent, countersigned it, a standard this repository set and checks itself. That happened in 103 projects across 13 languages with no language-specific analyzer; all but five were countersigned, four predate the gate and one ran without it, and six carry a finding blocked at the close, named on its row. Failed: the run spent the budget declared before it started without converging, was stopped short of it by decision (thor), or, for libuv, was abandoned before it had one. That leaves PapaParse, an audit held to the same method rather than a loop run, which the receipts page counts as Fixed alongside the 103.
## A Claude Code loop that finds bugs in your repo, fixes them, and checks every fix before it commits. Maintainers have merged 79 of its patches.
The model writes the patch. The harness around it decides whether that patch can be trusted. Jeffy Loop is a verification harness for Claude Code: type /jeffy and it audits your codebase, finds a problem, fixes it, checks the fix, and commits the result locally, one task at a time. It keeps going until the codebase passes a fresh audit.
The harness is deliberately hard to fool. If a fix breaks your tests, the loop is told to revert it. The loop is told never to push, and the hook never pushes; keep push off its allowlist. A standard run counts as complete only when a fresh audit is clean, your tests pass again, and an adversarial review by a second AI agent with no part in the run signs off. The goal is fixes that survive scrutiny.
There is a real-world test for that. Anthropic fixed a security issue this loop found in claude-code-action, and closed the report as Resolved after we retested the fix. Maintainers with no stake in this project have merged 79 of its pull requests, each filed from a local clone, into projects run by NVIDIA, Arm, Meta, Tesla, Google, Apple, Microsoft, Netflix, Apache, Palo Alto Networks, Oracle, IBM, HashiCorp, Cisco, Uber, Square, Cloudflare, Dropbox, Tencent, and more. See them all.
Sending a fix upstream? If you use Jeffy Loop to open pull requests on repositories you do not own, we strongly recommend running housebroken (pip install housebroken-cli) alongside it. Jeffy proves the fix; housebroken gets it through the maintainer's door. It reads the project's AI policy first, checks that the finding is not already fixed or ruled intended, holds the pull request to one finding and the smallest diff, has a second model attack it, and refuses to file until every gate has passed. Every pull request this project has filed since housebroken shipped went through it.
## The record
Jeffy was run against 132 open-source projects with no connection to this repository, each judged by its own test suite. Every one of these runs is published, failures included.
| Projects tested | Converged | Failed | PRs merged | PRs open | Issues filed |
| 132 | 103 | 28 | 79 | 10 | 3 |
Converged: the closing audit came back clean and the loop's adversarial evaluator, a fresh-context sub-agent, countersigned it, a standard this repository set and checks itself. That happened in 103 projects across 13 languages with no language-specific analyzer; all but five were countersigned, four predate the gate and one ran without it, and six carry a finding blocked at the close, named on its row. Failed: the run spent the budget declared before it started without converging, was stopped short of it by decision (thor), or, for libuv, was abandoned before it had one. That leaves PapaParse, an audit held to the same method rather than a loop run, which the receipts page counts as Fixed alongside the 103.
Listed badge (optional)
Copy this line into the README if you want the listing mark. Optional, and it does not affect being listed.
[](https://anclum.com/en/s/lenamonj-jeffy-loop/)
- Source
- github · https://github.com/lenamonj/jeffy-loop
- Browse by field
- Open source
- Traction
- 13 ★
- added
- today
More entries
Filed under
Every tag opens the full list for that filter. Counts are the live catalogue figures.