modakbul-gongbang/harness/specify
Claude Code and agent skill by modakbul-gongbang/harness
- community
Checking that the SKILL.md file is still in the repository.
Turn a goal into an implementation plan (spec.md). Simplified layer chain: L0:Goal → L1:Context → L2:Decisions → L3:Requirements → L4:Tasks. Evidence-based clarity scoring at L2. User approves at L2, L3, L4. Output is a single spec.md file written with the Write tool. Use when: "/specify", "specify", "plan this"
Install
npx vskill@latest install modakbul-gongbang/harness/specifySKILL.md
/specify — Spec Generator
Generate a spec.md through a structured derivation chain. Each layer builds on the previous — no skipping, no out-of-order writes.
Core Rules
- Write tool is the writer — All spec output goes to
{specDir}/spec.mdvia the Write tool. No CLI, no JSON. - Append, don't overwrite — Read existing spec.md before writing. Append new sections or update specific sections in place using Edit.
- Reference before writing — Read the reference file for the current layer (
references/L*) before constructing content. - Validate at layer transitions — After writing each layer, read spec.md and verify required sections exist.
- One layer at a time — Complete and validate each layer before advancing.
Spec Directory
The spec is written to a project-relative directory:
{name} = kebab-case derived from the goal.
Create the directory at Session Init:
Layer Flow
Execute layers sequentially. Read each reference file just-in-time.
| Layer | Read Reference | What | Gate | |
|---|---|---|---|---|
| L0 | references/L0-L1-context.md | Mirror → Goal, Non-goals, Confirmed Goal | User confirms mirror | |
| L1 | (same file) | Codebase research → Research section | Auto-advance | |
| L2 | references/L2-decisions.md | Interview → Decisions + Constraints | Self-validate + L2-reviewer + User approval | |
| L3 | references/L3-requirements.md | Derive requirements + sub from decisions | Self-validate + User approval | |
| L4 | references/L4-tasks.md | Derive tasks + external deps, Plan Summary | Self-validate + User approval |
Session Init (before L0)
Then create spec.md with initial content via Write tool.
User Approval Protocol
Three approval gates (L2, L3, L4). Each uses the same pattern:
- Approve → advance to next layer
- Revise → user provides corrections, update spec.md sections, re-present (loop until approved)
- Abort → stop
Checklist Before Stopping
- [ ] spec.md at
specs/{name}/spec.md - [ ]
## Goalsection populated - [ ]
## Confirmed Goalsection populated - [ ]
## Non-goalssection populated (or "(none)") - [ ]
## Researchsection populated - [ ]
## Decisionssection populated with at least one decision - [ ]
## Constraintssection populated (or "(none)") - [ ]
## Requirementssection with every requirement having at least 1 sub-requirement with GWT - [ ]
## Taskssection with every task havingFulfillslinking to requirements - [ ] Plan Summary presented to user
- [ ]
Approved byandApproved atwritten to Meta section after final approval
Security
- Security scanPassed, 100 of 100Pattern checks on the skill's files.
- Model reviewNot runA model review is scheduled for the next rescan.
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/specify/SKILL.md
Popularity
- 7-day trend score
- 0
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/modakbul-gongbang/harness/specify)