Skip to content
Agentmart
Submit a tool

spec

Stage 2 (Design) of the AI-native SDLC.

Skill by bibutikoley

IndexedMIT★ 3
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/bibutikoley-spec

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Stage 2 (Design) of the AI-native SDLC. Read an accepted intent.md and produce a requirements-and-design spec.md constrained by the organisation's policy skills (security, compliance, brand, UX), with flagged concerns first. Use when an intent is accepted, when asked for a spec/requirements/design from an intent, or on /sdlc:spec. Also used non-interactively by CI.

Use cases: design, devops, security

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/spec

What's inside

Skill name
spec
Description
Stage 2 (Design) of the AI-native SDLC. Read an accepted intent.md and produce a requirements-and-design spec.md constrained by the organisation's policy skills (security, compliance, brand, UX), with flagged concerns first. Use when an intent is accepted, when asked for a spec/requirements/design from an intent, or on /sdlc:spec. Also used non-interactively by CI.
License
MIT
Instructions
54 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# /sdlc:spec — requirements and design in one pass

Read `sdlc-loop`. The product owner reviews this spec; they do not write it.

## Preconditions

1. Resolve the change: `$ARGUMENTS` is a change id, a path to `intent.md`, or empty (then list changes whose
   `intent.md` is `status: accepted` and has no `spec.md`, and ask which one).
2. `intent.md` must be `status: accepted`. If it is `draft`, stop and say the product owner must accept it
   first (merge). If the user explicitly overrides, proceed and write `override: intent not accepted` in the
   spec frontmatter.
3. Load every policy skill available: `security-policy`, `compliance-policy`, `brand-guidelines`,
   `ux-standards`, `secure-api-review`, plus any under the repo's `.claude/skills/`. Record which ones you
   applied in `skills_applied` (name@last-reviewed date from the skill header).
4. If a Figma URL is given or referenced in the intent and the Figma connector is available, read the frames
   with `get_design_context` / `get_screenshot` and treat them as the approved mock.
5. If the tracker holds the record (`sourceOfTruth: tracker|linked`), read the ticket first (`tracker-sync`).

## Produce `spec.md`

Use `sdlc/changes/_template/spec.md`. The prompt you are executing is, in the course's words:

> Read the intent.md and produce a requirements and design spec for integrating it into our existing codebase.
> Apply the skills available to you so the spec conforms to our brand guidelines, security policies and UX
> standards. Document the spec fully as spec.md, ready to hand to the engineering team. Describe clearly any
> areas of concern, especially where you cannot satisfy contradicting policies.

Rules:
- **Flagged concerns come first.** Every place a policy conflicts with the intent, with another policy, or
  with the codebase gets a row: concern, policy/skill, owner who must decide, decision (blank until made).
  These are the points an analyst would have escalated.
- Read the codebase (read-only) so the design names real modules, endpoints and conventions. Cite paths.
- Requirements are numbered and testable and each traces to a line in `intent.md`.
- Acceptance criteria are observable checks an engineer can turn into tests.
- Carry every open question from `intent.md` forward as answered (where) or still open (who answers).
- Set `intent_commit` to the SHA of the accepted intent.md. Inherit `risk`; raise it if the design reveals
  money movement, auth changes, PII, regulated data or infra.

## Commit and hand over

- Branch `sdlc/spec/<change-id>`, commit `spec: <title>` with trailer `SDLC-Change: <id>`, PR titled
  `Spec: <title>` assigned to the product owner (from `owner`). In CI, label the PR `sdlc:spec`.
- If the tracker is authoritative, write a summary + SHA back to the ticket.
- Report: the path, the list of flagged concerns with owners, and the gate: **the product owner resolves each
  flagged concern with its policy owner, then accepts by merging with `status: accepted`. A tech lead must also
  approve if `risk: high`. Acceptance starts plan mode (`/sdlc:plan`).**

## Do not

- Do not write code or an implementation plan (files, order of work). That is Stage 3.
- Do not silently resolve a policy conflict. Flag it.
- Do not mark `status: accepted`.

From bibutikoley/ai-sdlc-claude, MIT.

More from bibutikoley/ai-sdlc-claude

Signals

Stars
3
Forks
0
Open issues
0
Last commit (this folder)
Sep 5, 2026 (1 month ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from bibutikoley

Cite this page

spec skill. Agent Mart. https://agentmart.dev/skills/bibutikoley-spec (last updated 2026-10-24)