Now liveThe Skillselion MCP - thousands of ranked skills, loaded into your agent mid-task. No install.Get it →
schpet avatar

Speccer

  • 1 installs
  • 5 repo stars
  • Updated June 1, 2026
  • schpet/toolbox

Distills rough notes into a single structured spec document with feature domains, open questions, and actionable issues.

About

Transforms unstructured project ideas into one docs/spec.md with overview, domain analysis, and issues with acceptance criteria. A developer uses it to turn rough notes into a well-defined specification.

  • Produces a single docs/spec.md, not scattered files
  • Identifies ambiguities as open questions plus actionable issues

Speccer by the numbers

  • 1 all-time installs (skills.sh)
  • Ranked #2,479 of 3,282 Productivity & Planning skills by installs in the Skillselion catalog
  • Data as of Aug 5, 2026 (Skillselion catalog sync)
npx skills add https://github.com/schpet/toolbox --skill speccer

Add your badge

Show developers this skill is listed on Skillselion. Paste this into your README.

Listed on Skillselion
Installs1
repo stars5
Last updatedJune 1, 2026
Repositoryschpet/toolbox

What it does

Distills rough notes into a single structured spec document with feature domains, open questions, and actionable issues.

Files

SKILL.mdMarkdownGitHub ↗

Speccer

Transform rough ideas into a structured project specification with actionable issues.

Overview

This skill orchestrates a multi-phase process to distill unstructured input into a single spec document (docs/spec.md) containing: 1. Project overview and tech stack 2. Feature/domain analysis sections 3. Open questions for the user 4. Actionable issues with acceptance criteria

Everything goes in one file. Do not create multiple files or a directory of spec documents.

Output

The entire spec is written to docs/spec.md. This is the only file speccer creates or modifies.

Workflow

Phase 0: Project Foundation

Before analyzing features, establish the project's technical foundation.

1. Identify the tech stack by asking the user:

  • What language/framework will this project use? (e.g., Rust, Rails, Node.js, Python)
  • Is this a new project or adding to an existing codebase?

2. For new projects, determine prerequisites and scaffolding:

StackPrerequisitesScaffolding Command
Rustrustup, cargocargo new or cargo init
RailsRuby, bundler, rails gemrails new + database choice
Node.jsNode, npm/pnpm/yarnnpm init or framework CLI
PythonPython, pip/uv, venvuv init or framework setup
GoGo toolchaingo mod init
ElixirErlang, Elixir, mixmix new or mix phx.new

3. Ask clarifying questions about project setup:

  • Database requirements? (PostgreSQL, SQLite, MySQL, none)
  • Package manager preferences? (npm vs pnpm, pip vs uv)
  • Any specific framework version requirements?
  • CI/CD requirements? (GitHub Actions, etc.)
  • Deployment target? (affects scaffolding choices)

Phase 1: Decomposition

When invoked with rough input:

1. Read the input and identify distinct feature/domain areas 2. Create the initial docs/spec.md with the project overview, tech stack, and list of identified feature areas 3. Output a brief summary of identified features before proceeding

Phase 2: Deep Analysis

For each identified feature/domain area, analyze:

1. What's well-defined vs ambiguous 2. Implementation concerns (without designing solutions) 3. Questions that need user clarification 4. Draft acceptance criteria for potential issues

Write each feature's analysis as a section within docs/spec.md. Use sub-agents (Task tool with subagent_type="general-purpose") to analyze features in parallel if needed, but have them return their analysis as text — do not have sub-agents write separate files. Incorporate their output into the single spec document yourself.

Phase 3: User Clarification

Present questions to the user using AskUserQuestion tool:

  • Group related questions where possible
  • Provide context for each question
  • Offer reasonable default options when applicable
  • Update docs/spec.md with answers as they come in

For complex clarifications, ask in batches of 3-4 questions max per interaction.

Phase 4: Refinement

After receiving user answers:

1. Update relevant sections of docs/spec.md with clarifications 2. Mark answered questions as resolved 3. If new questions arise from answers, repeat Phase 3

Phase 5: Issue Generation

When all clarifications are complete:

1. Compile issues starting with setup, then feature issues 2. Add an Issues section to docs/spec.md with:

  • Setup issues first: toolchain installation, project scaffolding, initial configuration
  • Feature issues grouped by area
  • Each issue has: title, description, acceptance criteria
  • Issues are ordered by dependency (setup → foundation → features)

3. If user wants beads integration, for each issue:

   Use beads:create skill to create the issue with:
   - Title from spec
   - Description including acceptance criteria
   - Labels for feature area

4. Update the Status section: "Specification complete"

Invocation

The skill can be invoked:

  • /speccer - Start fresh with new input
  • /speccer refine - Continue refining existing spec (re-run Phase 3-5)
  • /speccer issues - Skip to issue generation from existing spec
  • /speccer issues --beads - Generate issues and create beads

Maintaining Context

When resuming work:

1. Always read docs/spec.md first 2. Check the Status section to determine which phase to continue 3. Look for the Open Questions section to see pending clarifications

Tips

  • Prefer more granular features over fewer large ones
  • Questions should be concrete and actionable, not abstract
  • Acceptance criteria should be testable/verifiable
  • When uncertain about scope, err toward asking the user

Related skills

This week in AI coding

Five minutes, every Monday - the tools, releases and tactics for developers.

unsubscribe anytime.