Agent skill
dev-workflow-full
Run the full ticket lifecycle end-to-end (Jira or GitHub issue). Chains initialize, start-work, prepare-commit, create-pr, and review-pr. Supports autopilot or pair mode. Use when the user says "full workflow", "end to end", or wants to go from ticket/issue to reviewed PR in one go.
Install this agent skill to your Project
npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/other/other/dev-workflow-full
SKILL.md
Full Dev Workflow
Run the complete ticket lifecycle back-to-back: create branch, implement, commit, open PR, and review.
Pipeline
dev-workflow-initialize
↓
dev-workflow-start-work (includes dev-workflow-prepare-commit when atomic)
↓
[final commit if needed] (dev-workflow-prepare-commit)
↓
dev-workflow-create-pr
↓
dev-workflow-review-pr
Step 0: Choose Mode and Collect Info
ALWAYS use the AskQuestion tool:
- Title: "Workflow Mode"
- Questions:
- id: "mode", prompt: "How would you like to run the workflow?", options:
- id: "autopilot", label: "Autopilot - collect info upfront, then run everything without prompting"
- id: "pair", label: "Pair - walk through each step together, I'll confirm at each decision point"
- id: "mode", prompt: "How would you like to run the workflow?", options:
If Autopilot: Collect Everything Upfront
Gather all required inputs in a single AskQuestion call before doing any work:
- id: "source", prompt: "What are you working from?", options:
- id: "jira", label: "Jira ticket (e.g. RETIRE-123)"
- id: "github", label: "GitHub issue (URL or owner/repo#N)"
- id: "unticketed", label: "Unticketed work (use RETIRE-1908)"
- id: "open_cursor", prompt: "Open the new worktree in Cursor?", options:
- id: "yes", label: "Yes, open in Cursor"
- id: "no", label: "No, stay here"
If "jira" or "github", ask conversationally for the ticket number or issue URL/ref. Then run initialize with that source (see dev-workflow-initialize for Jira vs GitHub).
Autopilot defaults (applied automatically, no prompts):
| Decision Point | Default |
|---|---|
| Branch name | Auto-generated from ticket title |
| Branch/worktree conflict | Create with numeric suffix |
| Commit strategy | Atomic commits |
| Unstaged changes | Stage all changes |
| Minor code issues | Fix automatically |
| Lint errors | Fix automatically (--fix) |
| Test failures | STOP -- always stop on test failures regardless of mode |
| Commit message | Auto-generated conventional commit |
| Push after commit | Yes |
| Critical issues (secrets, etc.) | STOP -- always stop on critical issues regardless of mode |
If Pair: Proceed Normally
Each sub-skill runs with its full interactive prompts. The user confirms at every decision point. This is the standard behavior of each individual skill.
Step 1: Initialize
Use the dev-workflow-initialize skill to create the branch and worktree.
- Autopilot: Use the ticket or issue collected in Step 0 (Jira key, GitHub issue URL/ref, or unticketed). Skip the "Which Ticket or Issue?" prompt. Auto-generate branch name per source type. Skip "Open in Cursor?" (use collected answer). Skip "Start Implementation?" (always yes).
- Pair: All prompts go to the user as normal.
Step 2: Start Work
Use the dev-workflow-start-work skill to implement the ticket.
- Autopilot: Present the implementation plan for review (always show the plan -- this is the one pause in autopilot). After user confirms, use atomic commit strategy automatically. Skip the commit strategy prompt. During implementation, use the
dev-workflow-prepare-commitskill after each logical unit with all defaults (stage all, auto-fix minor issues, auto-fix lint, auto-generate message, auto-push). - Pair: All prompts go to the user as normal (plan confirmation, commit strategy, per-commit decisions).
Step 2.5: Final Commit (single-commit strategy only, pair mode)
Only applies in pair mode when the user chose "single commit at end". Use the dev-workflow-prepare-commit skill to review, lint, test, and commit. Auto-push when done.
This step never runs in autopilot (autopilot always uses atomic commits).
Step 3: Create PR
Use the dev-workflow-create-pr skill to push and open the pull request.
- Autopilot: Push if needed. Generate PR description and create the PR. Skip "Start PR Review?" (always yes). Proceed directly to Step 4.
- Pair: All prompts go to the user as normal.
Step 4: Review PR
Use the dev-workflow-review-pr skill to review the pull request.
- Autopilot: Review the current branch's PR automatically. Skip the "Which PR?" prompt.
- Pair: All prompts go to the user as normal.
Step 5: Done
Report the final status:
Full workflow complete!
Mode: [Autopilot / Pair]
Ticket/Issue: [TICKET-ID or issue ref]
Branch: [BRANCH-NAME]
Worktree: [PATH]
Commits: [N] atomic commits
PR: [PR-URL]
Review: Posted inline comments
Skipping Steps
If the user is partway through the workflow (e.g., already on a branch with work in progress), detect the current state and skip completed steps:
- Already on a feature branch with a worktree: Skip Step 1, start at Step 2
- Implementation done, changes uncommitted: Start at Step 2.5
- Changes committed and pushed: Start at Step 3
- PR already exists: Start at Step 4
Use git status, git log, and gh pr view to detect the current state.
Hard Stops (Both Modes)
These always halt the pipeline regardless of mode:
- Test failures: Stop and report. Never skip failing tests.
- Critical security issues: Hardcoded secrets, API keys, credentials. Stop and report.
- Git/gh errors: Push failures, auth issues. Stop and report.
The user can re-run the workflow after fixing the issue and it will pick up from where it left off (see "Skipping Steps").
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
Didn't find tool you were looking for?