suede-creator-skills

Install · The skills · Docs site · Skill catalog · Blog · MCP · @johnnysuede

By Jason Colapietro / Suede Labs AI

Your coding agent is fast, capable, and completely unsupervised. It will happily ship the bug, skip the test, publish the slop, and tell you everything went great.

This pack is the supervision: 73 public skill folders, MIT-licensed and broadly reusable, for Claude Code, OpenAI Codex, and any skills-compatible agent. Outcome-bound orchestration, multi-agent teams, Codex worker fleets, code review with a blunt A–F ship grade, AI evals, design, conversion copy, SEO, 41 marketing lanes, mobile app factories, and creator-rights tooling. Every skill is a plain skills/<name>/SKILL.md file you can read before you trust it. No binaries, no telemetry, no accounts.

Install in 30 seconds

Claude Code — add the marketplace, install the pack:

/plugin marketplace add JasonColapietro/suede-creator-skills
/plugin install suede-skills@suede

suede-skills installs all 73 skills. Want less? Two focused subsets: /plugin install suede-agent-workflows@suede (Full Send, orchestration, workflows, evals) and /plugin install suede-code@suede (review, grade, ship-gate).

Codex — add the Codex-native marketplace, install the complete plugin:

codex plugin marketplace add JasonColapietro/suede-creator-skills --ref main
codex plugin add suede-skills@suede-codex

The Codex plugin loads all 73 skills and registers both read-only MCP discovery profiles. Restart Codex after installing or updating.

Any agent (Cursor, Copilot, Windsurf, Claude Code, Codex) via the skills CLI:

npx skills add JasonColapietro/suede-creator-skills

Prefer a clone? install.sh copies all 73 skills into ~/.claude/skills/ and prints the installed count:

git clone https://github.com/JasonColapietro/suede-creator-skills.git && bash suede-creator-skills/install.sh

You can also copy individual skill folders into .claude/skills/ (project) or ~/.claude/skills/ (user).

Codex — install one skill from GitHub:

python3 ~/.codex/skills/.system/skill-installer/scripts/install-skill-from-github.py \
  --repo JasonColapietro/suede-creator-skills \
  --path skills/suede-agent-teams

Swap skills/<skill> for any skill folder, or pass extra skills/<name> paths to install several at once. Restart Codex after installing.

Claude Code — project-level copy of a single skill:

git clone https://github.com/JasonColapietro/suede-creator-skills.git /tmp/suede-creator-skills
mkdir -p .claude/skills
cp -R /tmp/suede-creator-skills/skills/suede-agent-teams .claude/skills/

For a user-level install, copy into ~/.claude/skills/ instead. Claude.ai and organization-managed Claude Skills may use upload or admin-managed flows instead of filesystem copy. Review skill contents before enabling code execution.

MCP — dependency-free stdio server:

node mcp/suede-skills-mcp.mjs --profile all

Exposes 7 tools (list_suede_skills, get_suede_skill, suede_install_options, suede_copy_seo_audit, suede_visibility_grade, suede_code_grade, suede_qa_checklist), 6 resources, and 5 prompts over JSON-RPC.

Try it right now

Install the pack, then ask for a code review with a ship grade on your current changes:

Use suede-code to review my staged diff and give it an A-F ship grade.

For a broad authorized outcome, invoke Full Send directly:

Use suede-full-send to finish this release across code, docs, CI, and live verification. Preserve unrelated WIP and return proof.

If the pack saves you an hour, star the repo — stars are how other builders find it.

Full Send: broad outcomes, one authority

Tell your agent "max effort, spare no compute, fix everything" and most of the time you get enthusiasm, not engineering. suede-full-send turns that intent into a structure: one authorized controller, every useful non-colliding lane running in parallel, adversarial reconciliation where lanes attack each other's claims, and a close that is either concise proof or one named blocker.

The rest of the orchestration lane backs it up:

• suede-agent-teams wires complex changes into coordinated agent lanes with WIP collision detection, RFC mode, feature-flag strategy, rollback trees, and a handoff checklist that won't close without evidence. Its public-contribution mode adds scored issue queues, atomic leases, isolated worktrees, and authority-gated contribution packets.
• suede-ship runs the canonical orchestration DAG on a repo: scout, multi-lens research, a red-teamed lane plan, disjoint build lanes, dual-lens review, adversarial refutation, an integration gate, and a release verifier with an evidence handoff.
• suede-ship-copy is the same DAG rebuilt for one high-stakes piece of writing: five blind research lenses, a claim audit that closes the set of assertable facts, disjoint section writers, adversarial refutation, a deslop pass, and a publish-readiness gate.
• suede-codex-fleet is the Suede Fable Fleet: a Claude orchestrator decomposes a high-volume job, writes self-contained briefs, spawns parallel OpenAI Codex CLI codex exec workers, and reviews every output against acceptance criteria before anything ships. Claude judges, Codex generates.

The A–F ship grade

suede-code runs a findings pass (TypeScript, React, Next.js, OWASP, and database checklists), then returns a grade card scoring seven lanes: Correctness, Security and permissions, Data and state, Domain truth, UX and release behavior, Tests and verification, and Deploy readiness. The weakest lane caps the grade, and auth and payment surfaces carry grade caps of their own.

If it hits an instant-F trigger — a hardcoded secret, a permission check bypassable via a request param — the grade locks to F with the exact file and line, and no other lane can raise it. A polite review tells you what could be better. A grade tells you whether to ship.

The skills

Every link below goes to the skill's folder in this repo; the skill catalog has the same list with install commands per skill.

Agent orchestration & workflows

Code quality & shipping

| Skill | What i