|
| 1 | +--- |
| 2 | +description: "Run the full speckit workflow end-to-end (specify β clarify β plan β tasks β analyze β implement) with automatic progression between steps." |
| 3 | +--- |
| 4 | + |
| 5 | +## User Input |
| 6 | + |
| 7 | +```text |
| 8 | +$ARGUMENTS |
| 9 | +``` |
| 10 | + |
| 11 | +You **MUST** consider the user input before proceeding (if not empty). |
| 12 | + |
| 13 | +## Pipeline Overview |
| 14 | + |
| 15 | +This command orchestrates the full Spec-Driven Development workflow automatically. You MUST execute each step in sequence by invoking the corresponding `/speckit.*` command. After each step completes, proceed immediately to the next one WITHOUT returning control to the user, UNLESS a step explicitly requires user input (e.g., clarification questions). |
| 16 | + |
| 17 | +**Pipeline Steps:** |
| 18 | + |
| 19 | +1. `/speckit.specify` β Create feature specification |
| 20 | +2. `/speckit.clarify` β Resolve ambiguities (interactive: user answers questions) |
| 21 | +3. `/speckit.plan` β Generate technical plan |
| 22 | +4. `/speckit.tasks` β Break plan into actionable tasks |
| 23 | +5. `/speckit.analyze` β Cross-artifact consistency analysis |
| 24 | +6. `/speckit.implement` β Execute all tasks |
| 25 | + |
| 26 | +## Execution Rules |
| 27 | + |
| 28 | +### Auto-Progression (CRITICAL) |
| 29 | + |
| 30 | +- After each step completes successfully, **immediately invoke the next step**. Do NOT ask the user "shall I proceed?" or "ready for the next step?". Just do it. |
| 31 | +- The ONLY reasons to pause are: |
| 32 | + - A step requires user input (e.g., `/speckit.clarify` asking clarification questions, or `/speckit.implement` asking about incomplete checklists) |
| 33 | + - `/speckit.analyze` reports CRITICAL issues (ask user whether to proceed or fix first) |
| 34 | + - A step fails with an error |
| 35 | +- When user input is received and the step completes, resume auto-progression to the next step. |
| 36 | + |
| 37 | +### Step Invocation |
| 38 | + |
| 39 | +- Use the agent's native mechanism to invoke each slash command (e.g., Claude Code's Skill tool, Copilot's chat commands). |
| 40 | +- Step 1 (`/speckit.specify`): Pass the full `$ARGUMENTS` as the feature description. |
| 41 | +- Steps 2-6: Invoke with no arguments (they auto-detect the feature context from the current branch and feature directory). |
| 42 | + |
| 43 | +### Error Handling |
| 44 | + |
| 45 | +- If any step fails, report the error and suggest the corrective action. Do NOT auto-proceed past a failed step. |
| 46 | +- If `/speckit.analyze` reports CRITICAL issues, pause and present the user with two options: |
| 47 | + 1. Fix the issues before proceeding to implementation |
| 48 | + 2. Proceed to implementation despite the issues |
| 49 | + |
| 50 | +### Progress Reporting |
| 51 | + |
| 52 | +- Before each step, output a single line: `## Pipeline Step N/6: [step name]` |
| 53 | +- After each step completes, output: `Step N/6 complete. Proceeding to next step...` |
| 54 | +- Keep progress reporting minimal β the individual commands already produce their own output. |
| 55 | + |
| 56 | +## Start |
| 57 | + |
| 58 | +Begin by invoking `/speckit.specify` with the feature description from `$ARGUMENTS`. If `$ARGUMENTS` is empty, ask the user for a feature description before starting. |
0 commit comments