AgentStack
Browse Sign in
Browse Why AgentStack Sell Docs
Sign in
SKILL verified MIT Self-run

Prd Executor

skill-sixz-ai-prd-workflow-prd-executor · by Sixz-AI

|

No reviews yet
0 installs
7 views
0.0% view→install

Install

$ agentstack add skill-sixz-ai-prd-workflow-prd-executor

✓ scanned · ✓ verified, works with Claude Code, Cursor, and more.

Security review

✓ Passed

No issues found. Passed automated security review. · v0.1.0 How review works →

  • Prompt-injection patterns
  • Secret / credential exfiltration
  • Dangerous shell & filesystem operations
  • Untrusted network calls
  • Known-malicious package signatures

What it can access

  • Network access No
  • Filesystem access No
  • Shell / process execution No
  • Environment & secrets No
  • Dynamic code execution No

From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.

View the full security report →

Verified badge

Passed review? Show it. Paste this badge into your README, it links to the public security report.

AgentStack Verified badge Links to your public security report.
[![AgentStack Verified](https://agentstack.voostack.com/badges/verified.svg)](https://agentstack.voostack.com/security/report/skill-sixz-ai-prd-workflow-prd-executor)

Reliability & compatibility

Security review passed
0 installs to date
no reviews yet
3mo ago

Declared compatibility

Claude CodeClaude Desktop

Compatibility is declared by the source manifest. End-to-end runtime verification is coming, see below.

Preview Execution monitoring

We're building live execution health for every listing: tool-call success rate, median latency, uptime, and last-checked timestamps, measured, not self-reported. It isn't live yet, so we don't show numbers we can't stand behind.

How agent discovery & health will work →
Are you the author of Prd Executor? Claim this listing to set pricing, connect Stripe payouts, and keep 70% of every sale.
Sign up to claim

About

PRD Executor Skill

Overview

This skill guides a Dev Agent through executing a prd_decomposer breakdown located at docs/[Feature_Name]_breakdown/. It enforces three hard gates:

  1. Requirements confirmation (brainstorming all 10 closure dimensions) before any coding
  2. TDD workflow during implementation
  3. Git commit per completed REQ, push per completed Phase

When to Use

  • A breakdown directory exists at docs/[Feature_Name]_breakdown/
  • User says "start executing", "start development", "implement this PRD", "execute breakdown"
  • Dev Agent is handed a breakdown directory and needs to know how to proceed

Workflow

Step 1: Initialize

Read docs/[Feature_Name]_breakdown/outline.md. Extract:

  • Total Phases and REQs
  • Dependency graph between REQs
  • Current status of each REQ (look at the Status column)

Announce to the user: which REQs are pending, which (if any) are already in progress or done.

Step 2: Select Next Phase

Find the first Phase in outline.md whose status is ⬜ (not started) and whose REQ dependencies (if any cross-Phase) are all ✅ or ⚠️.

If no such Phase exists:

  • If all Phases are ✅: announce completion, done.
  • If a Phase is 💬 or 🔍 or 🚧 or 🧪: resume that Phase from the appropriate stage below.

Stage A: Discussion — Discuss All REQs in the Phase [HARD GATE 1]

  1. Update Phase status in outline.md to 💬 (Discussing)
  2. For each REQ in this Phase, in order:

a. Open the REQ file. Check whether 🔍 详细需求确认 still contains ⚠️ [待填充]. b. If YES (placeholder present):

  • Update REQ status in REQ file and outline.md to 💬
  • Invoke superpowers:brainstorming with the REQ content as context
  • Ensure ALL 10 closure dimensions are addressed:
  • API closure
  • Business logic closure
  • UI/UX closure
  • Error handling closure
  • Data consistency closure
  • Permission/security closure
  • Performance boundary closure
  • Testing closure (unit test scenarios + manual UAT checkpoints)
  • Analytics/tracking closure
  • Compatibility closure
  • Replace the entire ⚠️ [待填充] placeholder block with the brainstorming conclusions

c. If NO (placeholder already replaced): skip brainstorming for this REQ.

  1. YOU ARE FORBIDDEN FROM WRITING ANY IMPLEMENTATION CODE DURING STAGE A

Stage B: Summary Confirmation

  1. Update Phase status in outline.md to 🔍 (Pending Confirmation)
  2. Present a summary of all REQ discussion conclusions to the user:

``` Phase X Discussion Summary:

  • REQ-X.1 [Title]: [2-3 key conclusions]
  • REQ-X.2 [Title]: [2-3 key conclusions]
  • REQ-X.3 [Title]: [2-3 key conclusions]

```

  1. Ask the user: "Does the above look correct? Ready to start implementation?"
  2. Wait for the user's response:
  • Confirmed → proceed to Stage C
  • Revision requested for a specific REQ → re-run brainstorming for that REQ only, update its conclusions in the REQ file, re-present the full summary, then repeat step 3 (wait for confirmation again)

Stage C: Implementation — Implement All REQs Autonomously

  1. Update Phase status in outline.md to 🚧 (Implementing)
  2. For each REQ in this Phase, in order:

a. Update REQ status in REQ file and outline.md to 🚧 (In Progress) b. Invoke superpowers:test-driven-development with the REQ's Acceptance Criteria as test targets c. When Acceptance Criteria pass internally: keep REQ status as 🚧 — it will be updated to ✅ in Stage G after the user passes UAT d. Commit immediately after each REQ: ``bash git add -p # stage only implementation files, NOT debug/mock artifacts git commit -m "feat(REQ-X.X): [short description of what was built]" ``

  1. DO NOT interrupt the user at any point during Stage C
  2. Do NOT commit during: debug sessions, mock data setup, adding/removing console.log, WIP that doesn't pass AC

Stage D: UAT Notification

  1. Update Phase status in outline.md to 🧪 (Pending UAT)
  2. Generate a UAT checklist from each REQ's Acceptance Criteria:

``` Phase X UAT Checklist:

REQ-X.1 [Title] Please verify:

  • [ ] [AC item 1]
  • [ ] [AC item 2]

REQ-X.2 [Title] Please verify:

  • [ ] [AC item 1]
  • [ ] [AC item 2]

```

  1. Notify the user: "All REQs in Phase X are implemented. Please verify each item in the checklist above and report which REQs pass or fail (with reason for failures)."

Stage E: User UAT

Wait for the user to report results. While waiting, keep Phase status as 🧪 (Pending UAT) and do not modify any code. Expected format (user can use any natural language):

  • "REQ-X.1 pass, REQ-X.2 fail — [reason]"

Stage F: Fix Loop (failing REQs only)

For each REQ the user reports as failing:

  1. Do NOT re-run brainstorming — go straight to fixing based on user feedback
  2. Fix the implementation
  3. Re-generate the UAT checklist for that REQ only:

``` REQ-X.2 [Title] — Fixed. Please re-verify:

  • [ ] [AC item 1]
  • [ ] [AC item 2]

```

  1. Ask the user to re-verify only this REQ
  2. Repeat until this REQ passes
  3. REQs that already passed are not touched
  4. When all previously-failing REQs have now passed, proceed to Stage G

Stage G: Phase Completion

When all REQs in the Phase have passed UAT:

  1. Update each REQ file status:
  • ✅ Done — all Acceptance Criteria verified by user UAT
  • ⚠️ Partial — main feature works, some AC items shelved (write reason in 备注 field)
  1. Update the matching rows in outline.md
  2. Update the 最后更新 field in each REQ file to today's date
  3. Update Phase status in outline.md to ✅ (Complete)
  4. Push:

``bash git push ``

  1. Return to Step 2 for the next Phase

Status Reference

| Status | Emoji | Meaning | | ------ | ----- | ------- | | Not Started | ⬜ | Default at decomposition time | | Discussing | 💬 | Brainstorming in progress | | Changed | 🔄 | Requirements modified after confirmation (update 备注 with what changed and why) | | In Progress | 🚧 | Implementation underway | | Done | ✅ | All Acceptance Criteria verified | | Partial | ⚠️ | Main feature done, some AC items shelved (reason in 备注) | | Shelved | 🚫 | Blocked, see 备注 for reason |

Phase Status Reference

| Status | Emoji | Meaning | | ------ | ----- | ------- | | Not Started | ⬜ | Default | | Discussing | 💬 | Brainstorming REQs in progress | | Pending Confirmation | 🔍 | All REQs discussed, waiting for user sign-off | | Implementing | 🚧 | AI implementing all REQs autonomously | | Pending UAT | 🧪 | Waiting for user verification | | Complete | ✅ | All REQs passed UAT, pushed |

Handling Requirements Changes (🔄)

If requirements change mid-execution on a REQ that is 🚧 In Progress:

  1. Set status to 🔄 Changed
  2. In the 备注 field, record: what changed, why, and the date
  3. Update the 🎯 Requirement and ✅ Acceptance Criteria sections of the REQ file
  4. Re-run brainstorming on the changed dimensions only
  5. Resume implementation with the updated requirements

Gotchas

  1. Never skip the placeholder check — brainstorming surfaces edge cases the PRD missed; skipping it leads to rework
  2. Status must be updated in two places — REQ file AND outline.md; both must stay in sync
  3. Commit after each REQ, push after each Phase — do not batch commits across multiple REQs
  4. ⚠️ Partial ≠ done without documentation — a Partial REQ must have a reason in 备注 before it counts toward Phase completion
  5. 🔄 Changed invalidates in-progress work — when requirements change, re-confirm before continuing implementation

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

Install and usage instructions live in the source repository linked above.

Reviews

No reviews yet, be the first.

Versions

  • v0.1.0 Imported from the upstream source.