Skip to content
Back to skills

Pm Spec

ASecurity

Kick off the Define stage. Spawns spec-writer (PRD) and spec-reviewer (mechanical clarity check). Outputs a PRD ready for gate:spec.

  • 3 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added September 22, 2026
ai-agentsgobash

Security analysis

A100/100

Pro scans all 2 files and shows the line behind each finding

Scanned September 22, 2026

npx -y skills add VandanaAjayDubey111/great-pm --skill pm-spec --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Pm Spec?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Pm Spec
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/vandanaajaydubey111-pm-spec/badge)](https://www.skillsdirectory.com/skills/vandanaajaydubey111-pm-spec)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: pm-spec
description: "Kick off the Define stage. Spawns spec-writer (PRD) and spec-reviewer (mechanical clarity check). Outputs a PRD ready for gate:spec."
---

## Codex host binding

- Treat references to Claude slash workflows as the equivalently named Codex skill.
- Before delegating to any specialist, read the `great-pm-runtime` skill and the selected packaged role file.
- Treat "invoke", "assign", "delegate", "spawn", and source Agent-tool instructions as a required Codex `spawn_agent` call with that role and a bounded assignment.
- For the spawn tool's `task_name` or `agent_name` identifier, replace canonical role-name hyphens with underscores (`query-refiner-pm` → `query_refiner_pm`). Keep canonical names unchanged in prompts, filenames, verdict logs, and user-facing output; this is an internal identifier mapping, not a role rename.
- Store every returned agent identifier. Never call a wait tool until a spawn has returned an identifier, and wait only on identifiers returned by successful spawns.
- If `spawn_agent` is unavailable or a spawn fails, report BLOCKED; do not impersonate the specialist or wait on an empty agent set.
- Resolve bundled paths from the installed GreatPM plugin root.
- Ignore Claude-only model aliases, colors, turn limits, and tool allowlists.
- Preserve GreatPM human gates, governance, state, and reporting contracts.


You are the great-pm `$pm-spec` command. Stage 4 of the 6-stage loop.

## Parse arguments

- **`<initiative-slug-or-scope-item>`** (required) → either the whole
  initiative or a single top-of-backlog item.
- **`--from-backlog`** → read the prioritized backlog and write a PRD
  for the top-ranked item only.

## Pre-flight

```bash
SCOPE="${1:?usage: $pm-spec <slug-or-scope-item> [--from-backlog]}"
ls .great-pm/PROJECT.md 2>/dev/null && echo "PROJECT_OK" || echo "NO_PROJECT"
mkdir -p .great-pm/drafts
```

Verify the initiative is past gate:strategy:
```bash
bd list --label "gate:strategy" --status closed | grep -q "$SCOPE" && echo "STRATEGY_PASSED" || echo "NO_STRATEGY_GATE"
```

If `NO_STRATEGY_GATE` → BLOCKED. Tell the user gate:strategy must close
before specs are authored. (This is contractual — strategy gate first,
then specs.)

## Operating procedure

0. **Step 0 — refine the user's query** (transparent Mode B). Invoke
   `query-refiner-pm` with `$ARGUMENTS`. The refiner returns:
   ```
   You typed:  <original>

   Refined to: <refined brief>

   What changed: <one line>

   Proceeding with refined. Reply "use original" to override.
   ```
   Use the **refined** version as the brief for subsequent steps UNLESS
   the user replies "use original". Log the refinement to
   `.great-pm/refinements/$(date +%Y-%m-%d).log`. This wiring is
   universal across great-pm commands per the gate-policy: explicit
   discipline — you make the user's leverage visible while preserving
   their ability to override.

1. **spec-writer**
   - Brief: read strategy doc (and backlog if `--from-backlog`). Author a
     PRD covering all 10 sections of the template:
     Problem, Why Now, Users + JTBD, Solution Overview, User Stories with
     Acceptance Criteria, Scope, Non-goals, Edge Cases, Required Success
     Metrics, Definition of Done.
   - Output: `.great-pm/drafts/prd-<slug>.md` using
     `templates/PRD-template.md`.

2. **spec-reviewer** (mandatory mechanical check)
   - Brief: review the PRD against clarity / completeness / testability
     rubric. Surface any vague language, missing AC, missing edge cases,
     untestable metrics.
   - Verdict: PASS / NEEDS-WORK / FAIL.
   - The verdict travels **unedited** to the user.

3. If verdict is **PASS** → suggest:
   ```
   $pm-promote .great-pm/drafts/prd-<slug>.md --gate=spec
   ```

   If verdict is **NEEDS-WORK** or **FAIL** → print verbatim, tell user
   to fix in `.great-pm/drafts/prd-<slug>.md` and re-run.

4. Update PROJECT.md: stage `define` → complete (only if PASS).

## Reporting

- **DONE (PASS)**: `DONE: $pm-spec — prd-<slug>.md PASS. Ready for $pm-promote --gate=spec.`
- **HELD (NEEDS-WORK|FAIL)**: `HELD: $pm-spec — spec-reviewer <verdict>. Reasons: <verbatim>.`
- **BLOCKED**: missing strategy gate / missing PROJECT.md. tried + failed_because + need.

## Notes

- Spec-reviewer is **lightweight and mandatory** — clarity / completeness
  / testability only. It is NOT pm-reviewer (which does strategic
  critique).
- A PRD without a Definition of Done line is not done. The reviewer will
  catch this.

Files in this skill

  • SKILL.md4.4 KB
  • agents/openai.yaml220 B

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.