om-auto-fix-issue

Automates end-to-end tracker issue resolution from triage through PR review and QA.

1|Updated Apr 23, 2026
One-click install
npx skills add https://github.com/tkogut/agents-os-core --skill om-auto-fix-issue-tkogut
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: om-auto-fix-issue
Source: https://github.com/tkogut/agents-os-core/tree/main/global_skills/om-auto-fix-issue
Command: npx skills add https://github.com/tkogut/agents-os-core --skill om-auto-fix-issue-tkogut

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Manually driving a tracker issue from report to merged-quality PR involves many repetitive steps: verifying the bug, isolating a worktree, finding the root cause, implementing the fix, opening a PR, running review loops, and checking UI changes. This Skill orchestrates that entire chain autonomously from a single command, with claim locking so concurrent agents never collide. ## Core Features & Use Cases - Dual routing: Classifies each issue as a bug or feature request, then drives either the bug autofix chain (verify, root-cause, fix, open PR, review, UI QA) or the feature route (spec authoring then implementation). - Brief mode: Accepts a plain problem description, files a well-formed issue first via om-prepare-issue, then continues on it. - Concurrency-safe claims: Three-signal in-progress checks, stale-lock recovery, and a continuous lock handed off from issue to PR so work is never duplicated or left unclaimed. - Isolated worktrees: All changes happen in a temporary git worktree off the base branch, never disturbing the user's active checkout. - Use Case: Run it with "fix issue 1234" and it verifies the defect, implements a minimal fix with regression tests, opens a labeled PR, runs an automated review loop, and reports the PR link. ## Quick Start Ask the agent to fix issue 1234 end to end, or paste a plain problem description and let it file and resolve the issue automatically.

Frequently Asked Questions about om-auto-fix-issue

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automatically fix a GitHub issue end to end?▼

Invoke the skill with an issue number, URL, or plain problem description. It verifies the bug, creates an isolated worktree, finds the root cause, implements the fix with regression tests, opens a PR, and runs an automated review loop before reporting the PR link.

Can I fix an issue from a plain problem description without filing it first?▼

Yes, brief mode accepts a free-form problem description and files a well-formed issue via the om-prepare-issue skill before continuing. A numeric id that does not resolve to a real issue stops the run instead of filing a new one.

What happens if another agent or person is already working on the issue?▼

A three-signal concurrency check (assignee, in-progress label, recent claim comment) detects existing ownership and stops with a prompt. You can override with --force, which posts an explicit override comment naming the previous owner.

Does it handle feature requests as well as bugs?▼

Yes, it classifies issues label-first and routes feature requests through a spec-then-build path using om-auto-write-spec and om-auto-implement-spec. Mixed issues containing both a defect and a new capability are flagged for the user to split.

Will it merge the PR or approve QA automatically?▼

No, the skill never merges PRs or adds the qa-approved label. It leaves the PR in the review pipeline state, runs automated review and UI evidence collection, and reserves merge and QA sign-off for the pipeline's human gates.

What happens to the worktree and locks if the run fails midway?▼

A failure path releases whichever lock is held (issue or PR) with an explanatory abort comment, and a trap/finally block removes any worktree the run created. The assignee is kept so a human can see who last worked on it.