git-bisect-debugging
Locate regression-introducing commits using git bisect with deterministic classification. Use when a reproducible regression exists but the introducing commit is unknown; do not use for CI workflow design or application behavior implementation.
Works with
--- name: git-bisect-debugging description: Locate regression-introducing commits using git bisect with deterministic classification. Use when a reproducible regression exists but the introducing commit is unknown; do not use for CI workflow design or application behavior implementation. license: Apache-2.0 --- # Git Bisect Debugging ## Overview Use this skill to isolate first-bad commits quickly and reproducibly, so fixes target root cause instead of symptoms. ## Scope Boundaries - Use this skill when the task matches the trigger condition described in `description`. - Do not use this skill when the primary task falls outside this skill's domain. ## Shared References - Determinism guidance: - `references/bisect-determinism-guidance.md` ## Templates And Assets - Session log template: - `assets/bisect-session-log-template.md` - Verification checklist: - `assets/bisect-checklist.md` ## Inputs To Gather - Known good commit and known bad commit. - Deterministic pass/fail command for classification. - Environment requirements for reproducible execution. - Evidence expectations for culprit validation. ## Deliverables - Culprit commit candidate with evidence trail. - Bisect session log with classified steps. - Confidence statement and next action recommendation. ## Workflow 1. Validate good/bad endpoints and deterministic classifier command. 2. Run bisect and classify commits consistently. 3. Capture classifications in `assets/bisect-session-log-template.md`. 4. Re-validate candidate culprit around boundary commits. 5. Close with `assets/bisect-checklist.md` and next-step recommendation. ## Quality Standard - Good/bad endpoints are verified before bisect begins. - Classification is deterministic and repeatable. - Culprit claim is supported by reproducible evidence. - Session records are sufficient for peer audit. ## Failure Conditions - Stop when classification is flaky or environment-dependent. - Stop when no reliable good/bad endpoints can be identified. - Escalate when interacting commits prevent single-commit isolation.
More Debugging skills
diagnosing-bugs
mattpocock/skills
Diagnosis loop for hard bugs and performance regressions. Use when the user says "diagnose"/"debug this", or reports something broken/throwing/failing/slow.
explore-code
lllllllama/rigorpilot-skills
Rigor Improve implementation leaf skill for auditable candidate implementation in deep learning research repositories. Use when the researcher explicitly authorizes exploratory work on an isolated branch or worktree to transplant modules, adapt a backbone, add LoRA or adapter layers, replace a head, or stitch together meaningful low-risk migration ideas with rollback-aware records in `explore_outputs/`. Do not use for end-to-end exploration orchestration on top of `current_research`, trusted baseline reproduction, conservative debugging, environment setup, verified contribution claims, or default repository analysis.
safe-debug
lllllllama/rigorpilot-skills
Rigor Debug / Rigor Audit skill for deep learning research work. Use when the user pastes a traceback, terminal error, CUDA OOM, checkpoint load failure, shape mismatch, NaN loss symptom, or training failure and wants conservative diagnosis before any patching, with debug fixes clearly separated from research contributions. Do not use for broad refactoring, speculative adaptation, automatic exploratory patching, or general repository familiarization.
