OpenHands / OpenHands/software-agent-sdk
[Harness Watch] P1 — diagnose confirmed trajectory differences
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 1.1k
- Forks
- 539
- Avg merge
- 1d 19h
- Merged PRs (30d)
- 137
Description
Parent epic: #4627
Goal
After the deterministic comparison framework is stable, explain selected confirmed differences and propose one human-reviewable experiment. This does not block P0.
Reuse or refactor the existing propose-harness action instead of building a second diagnosis platform.
P1
- Consume the paired report and only the leading outcome-discordant or resource-divergent trajectories.
- Run read-only with no repository, branch, push, or issue-write capability.
- Produce a structured hypothesis, alternatives, evidence and counterevidence with instance/event links, predicted observable change, and one single-variable experiment.
- Ground quantitative statements in the deterministic report or tool output.
- Stop at the proposal; a human decides whether to implement or run it.
The agent interprets evidence. It does not calculate statistics, decide significance, or modify repositories.
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start with parent epic #4627 and the existing propose-harness action, then review how the paired report exposes outcome-discordant or resource-divergent trajectories. Done means a read-only proposal containing hypotheses, alternatives, evidence and counterevidence with instance/event links, predicted observable changes, and one single-variable experiment, without calculating statistics or modifying repositories.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- ai, testing-qa
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Active
- Clarity
- Mostly clear
- Newbie friendliness
- 38/100