<< All versions

Skill v1.0.0

currentAutomated scan100/100
suriyel/longtaskforagent/long-task-finalize
──Details
PublishedSeptember 29, 2026 at 06:33 PM
Content Hashsha256:bef5d09ad954e1b3...
Git SHA
──Files
Files (1 file, 4.4 KB)
SKILL.md4.4 KBactive
SKILL.md · 124 lines · 4.4 KB

version: "1.0.0" name: long-task-finalize description: "Use after ST Go verdict — generate usage examples and finalize release documentation via SubAgent"


Finalize — Post-ST Documentation & Examples

Generate scenario-based usage examples and finalize release documentation after System Testing passes with a Go/Conditional-Go verdict.

Announce at start: "I'm using the long-task-finalize skill. ST passed — generating examples and finalizing documentation."

Idempotent: Safe to re-invoke after ST defect-fix loops. Overwrites examples/ content cleanly on each run.

<HARD-GATE> Do NOT invoke this skill unless the ST verdict is Go or Conditional-Go. If the verdict is No-Go, loop back to Worker for fixes instead. </HARD-GATE>

Checklist

You MUST create a TodoWrite task for each step and complete them in order:

1. Gather Context

  • Read feature-list.json — all passing non-deprecated features, tech_stack, quality_gates
  • Read SRS document (docs/plans/*-srs.md) — requirement descriptions, user personas
  • Read Design document (docs/plans/*-design.md) — architecture, public API surface
  • Read UCD document (docs/plans/*-ucd.md) — only if UI features exist
  • Read task-progress.md — session history for ST summary entry
  • Read RELEASE_NOTES.md — current state for version entry
  • Note paths for SubAgent dispatch

2. Generate Examples (SubAgent)

Dispatch the example-generator SubAgent to produce scenario-based usage examples.

  1. Construct SubAgent prompt:

``` You are an Example Generator SubAgent.

## Your Task

  1. Read the agent definition: Read <skills_root>/../agents/example-generator.md
  2. Follow the process to generate scenario-based usage examples
  3. Return your result using the Structured Return Contract

## Input Parameters

  • feature-list.json: <path>
  • SRS: <srs_path>
  • Design: <design_path>
  • UCD: <ucd_path> (or "none")
  • tech_stack: <tech_stack_json>
  • Working directory: <project_root>

```

  1. Dispatch:

`` Agent( description = "Generate usage examples for all features", prompt = [constructed prompt] ) ``

  1. Parse return contract:
  • PASS: All planned scenarios generated and verified
  • PARTIAL: Some examples generated; log warnings for gaps
  • FAIL: Log error; proceed anyway — examples are non-blocking

Record in task-progress.md:

- Examples: <verdict> — N scenarios, N examples generated, N features covered

3. Update RELEASE_NOTES.md

Add ST completion and version entry (moved from ST Persist):

  • Add entry under [Unreleased] or create versioned section if appropriate
  • Include: ST verdict, date, test summary (categories run, defects found/fixed)
  • Reference the ST report document path

4. Update task-progress.md

Add ST session summary entry (moved from ST Persist):

  • ST categories executed, pass/fail counts
  • Defects found and fixed (with severity)
  • Full mutation score
  • Final quality metrics
  • Example generation results (from Step 2)

5. Persist

  • Git commit all documentation artifacts:

`` git add examples/ RELEASE_NOTES.md task-progress.md git commit -m "docs: finalize release — examples, release notes, progress update" ``

  • Validate:

``bash python scripts/validate_features.py feature-list.json ``

6. Summary

Output completion summary:

Finalize — DONE
Examples: N scenarios generated (N features covered, N skipped)
RELEASE_NOTES.md: Updated with ST completion
task-progress.md: Updated with ST session summary

Critical Rules

  • Non-blocking — example generation failure does NOT retroactively change the Go verdict
  • Idempotent — safe to re-run; overwrites examples/ cleanly
  • SubAgent for examples only — RELEASE_NOTES and task-progress are updated directly by this skill (not by the SubAgent)
  • No new features — do not add, modify, or test any features; documentation only
  • Follow project conventions — examples match the project's language, style, and patterns

Integration

Called by: long-task-st (Step 13, after Go/Conditional-Go verdict) Reads: feature-list.json, docs/plans/*-srs.md, docs/plans/*-design.md, docs/plans/*-ucd.md (if UI), task-progress.md, RELEASE_NOTES.md, implementation code Produces: examples/ (usage examples + README.md), updated RELEASE_NOTES.md, updated task-progress.md Agent: agents/example-generator.md

All versions