DEDUPED Hot AI repos, turned into use cases Get 5 a day

How to score one agent patch on a SWE-sweep task with Swe Sweep

Run a unified diff against one benchmark task and get graded Harbor results.

Who it is forAn ML engineer or agent builder who has a model's patch for a bug-fixing task and needs a graded score.

The problemChecking whether a model's patch fixes the hidden bugs in a real repository requires running the right task with the right harness.

The outcomeA graded Harbor trial for the task, with results written under the jobs/ directory.

What you need first

  • Python with uv (or pip) installed
  • Harbor, as pinned by the repo
  • A unified diff file (model.patch) made against a task's base commit
  • Docker or the environment Harbor needs for the verifier, if Harbor requires it (inferred)

Steps

  1. git clone https://github.com/facebookresearch/swe-sweep.git
  2. cd swe-sweep
  3. uv sync
  4. sweep infra doctor
  5. uv run harbor run --path tasks/pandora-bench__dateutil --agent swesweep.patch_agent:PatchAgent --agent-kwarg "patch_path=$(realpath /path/to/model.patch)" --jobs-dir jobs --n-concurrent 1 --yes (inferred, check the README)

Prompt for your agent

Using the repo at facebookresearch/swe-sweep, run the Harbor command for the task tasks/<task-name> with the patch at <path-to-model.patch> as the patch_path, writing output to jobs/. Report the directory where results were written and whether the trial completed without errors.

Caveat. Harbor 0.23 does not support the separate verifier environments these tasks use, so the project pins a Harbor source revision until Harbor 0.24 is released.

Share: X LinkedIn Hacker News Email Copy link

Disclaimer

Deduped lists open-source repositories that belong to other people. We do not own, maintain, audit, or vouch for any of them, and a repo appearing here is not an endorsement or a safety check. The use cases, steps, and prompts are written by AI and may be incomplete, outdated, or just wrong. Where a use case is marked "Seen in the wild", we link to public posts where someone described something similar, but we have not verified those posts and the use case itself is still our AI's own write-up. Use all of this at your own risk. Only run code, commands, or agents on systems and accounts you are authorized to use, and read the code and every command before you run it, especially anything that touches credentials, files, or the network. Nothing here is professional, security, legal, or financial advice. All trademarks and project names belong to their respective owners, and their use here does not imply affiliation. Content is provided as is, without warranties of any kind. Found a mistake, a safety problem, or something you want removed? Email [email protected] and we will look at it promptly. Full disclaimer page