opsmill/infrahub / .agents/skills/speckit-opsmill-auto
speckit-opsmill-auto
VerifiedRun the full speckit workflow end-to-end — specify, plan, critique, tasks, implement, review — making all decisions autonomously. Stops before extract; run that manually.
User Input
$ARGUMENTS
You MUST consider the user input before proceeding (if not empty).
Outline
You are running the full speckit pipeline end-to-end. The user's input (above) is the feature description that will seed the specification phase.
This skill is a thin orchestrator over two sibling skills:
speckit-opsmill-prep— Specify → Plan → Critique → Tasks (+ spec/ask alignment check)speckit-opsmill-implement— Implement → Review → Final report
Run them back-to-back, making all decisions autonomously. Do not stop to ask the user for input between phases — if a sub-phase requires choices (clarification questions in specify, research decisions in plan, drift remediation in alignment), use your best judgment and proceed. The user expects a hands-off, one-shot execution. Extraction (speckit-opsmill-extract) is not run by this command — the user invokes it manually after reviewing the implementation report.
Each phase below is executed by invoking the named skill (e.g. via the agent's Skill tool). Skills are agent-agnostic, so this workflow runs identically across any harness that supports skill discovery — not only those exposing speckit slash commands.
Phase A — Preparation (delegated to speckit-opsmill-prep)
Invoke the speckit-opsmill-prep skill with the user's feature description ($ARGUMENTS) verbatim.
- This covers Specify → Plan → Critique → Tasks and the trailing spec/ask alignment check.
- The sub-skill commits its own artifacts after each phase via
speckit-checkpoint-commit. - If the alignment check inside prep loops (re-running plan/critique/tasks to fix spec drift), let it run to completion. Do not interfere.
Read prep's status line. speckit-opsmill-prep ends its output with a literal final line:
STATUS: <READY|BLOCKED> | SPEC_DIR: <absolute spec-dir path> | CRITIQUE: <extension|fallback|none|n/a> | REASON: <...>
Parse that line — do not infer success from the prose summary:
- Take
SPEC_DIRfrom this line as the spec directory to pass to Phase B. This is the deterministic hand-off; do not scrape the path out of free text. - If
STATUS: BLOCKED(e.g., alignment never converged within its retry budget, or a phase could not complete), surface the reason to the user and stop. Do not proceed to implementation on a misaligned or incomplete spec. - Take
CRITIQUEas the critique coverage of the run. Anything other thanextensionorn/ameans the critique extension was unavailable and prep either ran a reduced fallback critique or none at all.n/ais not a coverage claim — it means prep aborted before Phase 3 ever resolved a provider, so read it together withSTATUS. This does not block Phase B — carry it into the final summary. - Only proceed to Phase B when
STATUS: READY.
Phase B — Implementation tail (delegated to speckit-opsmill-implement)
Invoke the speckit-opsmill-implement skill with the SPEC_DIR path from Phase A as its argument. Because you pass an explicit spec-dir path, the sub-skill runs in autonomous-parent mode: its Phase 0 stop-conditions abort with a STATUS: BLOCKED line instead of pausing for a user.
- This covers Phase 0 (Preflight) → Phase 5 (Implement loop in clean-context subagents) → Phase 6 (Review) → Phase 7 (Final report).
- The sub-skill writes
<spec-dir>/opsmill-implement-report.mdand commits it.
Read implement's status line. speckit-opsmill-implement ends its output with a literal final line:
STATUS: <DONE|INCOMPLETE|BLOCKED> | SPEC_DIR: <...> | REVIEW: <plugin|command|fallback|none|n/a> | REASON: <...>
STATUS: BLOCKED— a Phase 0 stop-condition aborted before any implementation (no report written). Surface the reason in the final summary; do not retry from this orchestrator.STATUS: INCOMPLETE— blocked tasks, missing local-pass evidence, orREVIEW: none; capture that for the final summary. The user decides whether to re-run.STATUS: DONE— all chunks completed with evidence.- Take
REVIEWas the review coverage of the run. A missing reviewer does not stop the implement phase:REVIEW: pluginandREVIEW: commandare full reviews,REVIEW: fallbackmeans the sub-skill ran a reduced built-in review,REVIEW: nonemeans no review happened at all and forcesSTATUS: INCOMPLETE.REVIEW: n/ais not a coverage claim either — it means implement aborted before Phase 0 resolved a provider, and it pairs withSTATUS: BLOCKED. Never treat a completed implement phase as a reviewed change set without checking this field.
In every case, do not retry from this orchestrator.
Completion
After both phases are complete, provide a brief summary:
- Feature name and spec directory
- Whether the alignment check inside auto-prep required retries (and how many)
- Number of tasks completed (from the auto-implement report)
- Any review findings that were fixed inline vs. deferred
- Coverage of the quality gates (REQUIRED). State the
CRITIQUEvalue from Phase A and theREVIEWvalue from Phase B explicitly. IfCRITIQUEis notextension, orREVIEWis notpluginorcommand, lead the summary with it, name what is missing, and give the install command (specify extension add critique//plugin install pr-review-toolkit@claude-plugins-official). An autonomous run that quietly skipped a quality gate is the one outcome the user must never have to discover by opening a report file - Any notable decisions you or the sub-skills made autonomously
Do not open a PR, push, run extraction, or start a new feature. The user takes it from there — extraction (/speckit.opsmill.extract) is intentionally left as a manual follow-up so the user can review the implementation report first.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| planning-and-task-breakdownaddyosmani/agent-skillsBreaks work into ordered tasks. Use when you have a spec or clear requirements and need to break work into implementable tasks. Use when a task feels too large to start, when you need to estimate scope, or when parallel work is possible. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |
| screenpipethedotmack/claude-memSearch screen and audio history recorded by Screenpipe and optionally save selected context into claude-mem. Use when the user asks to recall desktop activity, a meeting, or something seen in another app through Screenpipe. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |
| contact-widgetnexu-io/open-designSelf-contained floating chat widget with welcome screen, social links, meeting button, and message input. Single HTML file, zero dependencies. | Stars: 108k | Updated | Claude CodeCodex | Verified | Install |
| meeting-notesnexu-io/open-designMeeting notes page — title bar with attendees, agenda checklist, decisions block, action items table with owners + dates, and a "next meeting" footer. Use when the brief mentions "meeting notes", "minutes", "1:1 notes", "all-hands recap", or "会议纪要". | Stars: 108k | Updated | Claude CodeCodex | Verified | Install |