Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Unified developer workflow for fixing bugs. Analyzes issue-tracker context, cross-checks docs/code, proposes a solution, implements the fix, verifies locally, and delivers a PR/MR.
.claude/skills/hoangnguyen0403-dev-fix/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 151% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 140% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 6% | 0% |
| case-07 | ✗→✓ | ▲ Improved | 15% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 124% | 0% |
> !IMPORTANT] > Unified developer workflow for fixing bugs. Analyzes issue-tracker context, cross-checks docs/code, proposes a solution, implements the fix, verifies locally, and delivers a PR/MR.
Optional args: slug=<feature>, ticket=<id/url>, mode=interactive|autonomous|channel, channel=<id>, auto_continue=true|false, profile=business|hybrid|technical.
When the user asks to perform this workflow, execute the following steps:
Goal: Take a bug ticket from root-cause analysis through a locally-verified PR/MR, enforcing a strict Propose -> Approve -> Verify cycle.
/dev-fix <issue-url-or-key>
// turbo
git pull origin main in the standard repo.> !TIP] > Sub-Agent Delegation: If your platform supports sub-agents, delegate ticket extraction to specialist-jira-analyst and context lookup to specialist-codebase-scout. If sub-agents are NOT supported, execute these steps yourself.
Reproduce steps, Expected Result, and Actual Result.docs/prd/prd-plan-[slug].md.common-task-complexity-routing; record snc_tier/model_tier in the plan.snc_tier=high; medium requires self-review first; low may skip approval in autonomous mode only.implementation-readiness; code only after READY or approved PARTIAL.> !TIP] > Sub-Agent Delegation: For the actual fix, delegate the TDD loop to specialist-tdd-implementer. If sub-agents are NOT supported, execute the TDD loop yourself using the TDD skill matched in AGENTS.md.
git worktree add ../<ticket-key> -b fix/<ticket-key> and cd into it.docs/srs/srs-task-list.md.common-tdd or the @specialist-tdd-implementer sub-agent. Follow common-best-practices and service-specific AGENTS.md rules.common-tdd: strict RED first for new behavior; for legacy fixes, characterize only when needed and reproduce the intended change as RED without deleting unrelated implementation.Do NOT rely on "it builds" — verify the fix against the issue reproduction steps.
quality-engineering-playwright-cli. Run the reproduction steps. Capture "After" snapshots.quality-engineering-appium-mcp. Run the reproduction steps on an emulator.Expected Result. If any sub-3px regressions exist, fix them now.docs/srs/srs-walkthrough.md.caveman-commit.docs/srs/srs-walkthrough.md with evidence of the local verification.test-loop REAL_BUG_DO_NOT_HEAL handoff as the ticket input: the healer's TEST: and EVIDENCE: lines plus the failing AC are the reproduce steps.slug, operator_profile (carried, not re-inferred), snc_tier, model_tier, implementation plan path, task list path, walkthrough path, PR/MR link, outcome report, next workflow.md# Implementation Plan: [Name] ## Goal ## Proposed Changes ## Task Slices | Slice | Scope | Verification | | ------- | ------- | -------------- | | [slice] | [scope] | [verification] | ## Risks ## Verification Plan ## Next Workflow implementation-readiness
md# Task: [Name] ## Scope ## Checklist - [ ] [task] ## Decisions ## Evidence ## Next Workflow verify-work
md# Walkthrough: [Name] ## Scope ## Acceptance Criteria ## Evidence | Check | Result | Evidence | | ------- | ------------------- | ---------- | | [check] | [PASS/FAIL/BLOCKED] | [evidence] | ## Risks ## Outcome Report feature_status: implemented | partially_implemented | blocked requirement_trace: BRD-OBJ-* -> REQ-* -> AC-* -> SRS-* -> evidence completed_evidence: []; missing_evidence: []; decision_needed: []; recommended_next_workflow: verify-bug | verify-work ## Next Workflow verify-bug | verify-work
Call get_session_cost(workflow="dev-fix") before final handoff.
close browser/appium sessions used during verification.docs/srs/srs-walkthrough.md.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 2,888 | 4,096 | +42% | 1 | 1 | 0% | 388 | 1,938 | +399% | 0 | 0 | — |
case-02 | fail→fail | 6,375 | 2,911 | -54% | 1 | 1 | 0% | 361 | 1,971 | +446% | 0 | 0 | — |
case-03 | fail→fail | 2,597 | 3,076 | +18% | 1 | 1 | 0% | 234 | 1,986 | +749% | 0 | 0 | — |
case-04 | fail→pass | 5,160 | 2,685 | -48% | 1 | 1 | 0% | 807 | 2,027 | +151% | 0 | 0 | — |
case-05 | fail→pass | 4,325 | 2,291 | -47% | 1 | 1 | 0% | 799 | 1,918 | +140% | 0 | 0 | — |
case-06 | fail→pass | 10,823 | 1,766 | -84% | 1 | 1 | 0% | 1,677 | 1,773 | +6% | 0 | 0 | — |
case-07 | fail→pass | 10,960 | 1,983 | -82% | 1 | 1 | 0% | 1,645 | 1,888 | +15% | 0 | 0 | — |
case-08 | pass→pass | 9,448 | 2,227 | -76% | 1 | 1 | 0% | 1,505 | 1,913 | +27% | 0 | 0 | — |
case-09 | fail→pass | 5,628 | 2,323 | -59% | 1 | 1 | 0% | 843 | 1,892 | +124% | 0 | 0 | — |
case-10 | fail→pass | 9,247 | 1,262 | -86% | 1 | 1 | 0% | 1,494 | 1,723 | +15% | 0 | 0 | — |
case-11 | fail→pass | 7,311 | 1,874 | -74% | 1 | 1 | 0% | 1,286 | 1,853 | +44% | 0 | 0 | — |
case-12 | pass→pass | 7,118 | 1,353 | -81% | 1 | 1 | 0% | 1,029 | 1,764 | +71% | 0 | 0 | — |
case-13 | fail→pass | 18,156 | 1,825 | -90% | 1 | 1 | 0% | 932 | 1,899 | +104% | 0 | 0 | — |
case-14 | fail→fail | 9,775 | 1,744 | -82% | 1 | 1 | 0% | 1,456 | 1,829 | +26% | 0 | 0 | — |
case-15 | fail→pass | 10,266 | 5,144 | -50% | 1 | 1 | 0% | 1,524 | 2,352 | +54% | 0 | 0 | — |
case-16 | pass→pass | 8,899 | 1,222 | -86% | 1 | 1 | 0% | 1,406 | 1,720 | +22% | 0 | 0 | — |
case-17 | fail→pass | 9,539 | 1,391 | -85% | 1 | 1 | 0% | 1,451 | 1,763 | +22% | 0 | 0 | — |
case-18 | pass→fail | 8,423 | 1,960 | -77% | 1 | 1 | 0% | 1,314 | 1,847 | +41% | 0 | 0 | — |
case-19 | fail→pass | 1,980 | 1,607 | -19% | 1 | 1 | 0% | 230 | 1,834 | +697% | 0 | 0 | — |
case-20 | pass→pass | 29,409 | 15,902 | -46% | 1 | 1 | 0% | 6,178 | 4,354 | -30% | 0 | 0 | — |
case-21 | pass→pass | 11,879 | 11,393 | -4% | 1 | 1 | 0% | 2,426 | 3,590 | +48% | 0 | 0 | — |
case-22 | pass→fail | 14,923 | 5,615 | -62% | 1 | 1 | 0% | 2,818 | 1,868 | -34% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 19 counted toward the lift figure. The other 3 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +41 percentage points is the difference between those two pass rates over the 19 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
The publisher has shipped newer versions since this run, so these numbers describe v2, not the version currently listed.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/3/2026 | +45% |
Other measured skills in the registry, with their headline benchmark lift.