2.6k
GitHub Stars
3
Bundled Files
2 months ago
Catalog Refreshed
3 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill openclaw/skills --skill github-issue-resolver- _meta.json301 B
- guardrails.json5.1 KB
- SKILL.md6.2 KB
Overview
This skill is an autonomous GitHub Issue Resolver agent that discovers, analyzes, and fixes open issues in repositories while enforcing strict safety guardrails. It supports the full workflow from issue discovery through drafting a pull request, with mandatory checkpoints that require user approval for risky operations. The agent logs every action to an audit trail and prevents scope creep, unauthorized access, and dangerous changes.
How this skill works
On a repository input, the skill runs a recommendation engine to fetch and score open issues, then presents prioritized options. When an issue is selected, it locks that issue, clones the repo into a sandbox, creates a safe branch, plans a fix and requests approval before applying changes. Tests run in the sandbox and failing tests can trigger an automatic rollback; committing, pushing, and PR creation require explicit user approval and guarded commands.
When to use it
- You want an agent to find and prioritize open issues across a GitHub repo.
- You want a safe, auditable workflow to implement fixes without manual orchestration.
- You need code changes, tests, and a draft PR but want explicit approval before pushes.
- You want to avoid touching protected branches, secrets, or critical files.
- You need a one-issue-at-a-time disciplined approach to avoid parallel scope creep.
Best practices
- Provide the repo in owner/repo form and confirm scope before discovery.
- Review the recommendation output and pick one issue to lock and work on.
- Approve the proposed fix plan before any code changes are applied.
- Never allow the agent to run with elevated permissions; keep credentials out of the sandbox.
- Require manual approval for commits, pushes, and PR creation to maintain control.
- Inspect the audit logs after each session to verify actions and outcomes.
Example use cases
- Scan a repository for high-impact bugs and get a recommended issue to fix.
- Have the agent propose a concrete fix plan including files and estimated diff size.
- Implement a small bug fix, run tests in a sandbox, and draft a PR for review.
- Generate basic tests for a fix when the project lacks coverage.
- Use the audit trail to demonstrate changes and approvals for compliance.
FAQ
No. Pushes and PR creation always require explicit user approval per the guardrails.
Can it modify protected branches or secrets?
No. Protected branches and sensitive files are blocked by non-negotiable guardrail rules.