/suede-code-grader · A-F code grading

One letter. No negotiation.

A hardcoded secret, or an auth check bypassable through a request parameter, locks the grade at F with the exact file and line, and no other lane can lift it. Otherwise seven lanes get graded on evidence, from correctness and security through tests and deploy readiness, and the weakest lane sets the letter. Auth, payment, migration, and public-API surfaces carry grade caps of their own. You get the grade, the evidence, and the ranked upgrades that would move it; the findings list lives in suede-code-review.

Install the skill View skill folder

"just give me the letter grade" /suede-code-grader Ask in plain words. The router reads the request and loads this lane; MCP agents find it with list_suede_skills.
$ /suede-code-grader release/v2.4 7 evidence lanes · grade caps armed auth surface detected · capped until the session fix lands SHIP GRADE: B · CAPPED one fix from an A

Public install command

Claude Code, adding the marketplace and the code-quality subset:

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

Codex is the public route. It installs from GitHub as a standard Codex skill folder.

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

Restart Codex after installing the skill.

The seven lanes

  • Correctness: behavior, edge cases, error paths, async, routing, data flow, and regression risk.
  • Security and permissions: auth, secrets, payment, injection, path, SSRF, and data exposure fail closed.
  • Data and state: schemas, migrations, caches, jobs, queues, webhooks, retries, and idempotency stay consistent.
  • Domain truth: published statements, rights, provenance, royalty routing, and product claims match the implementation.
  • UX and release behavior: loading, empty, error, success, mobile, screenshot, and route states hold together.
  • Tests and verification: changed behavior has meaningful tests, builds, runs, readbacks, or named caveats.
  • Deploy readiness: env vars, flags, configs, migrations, rollback notes, and release sequencing are clear.

Where it fits

Run it as the gate between "done" and "shipped." It reads the actual source and the evidence that exercises the change, not the PR description, then returns one overall grade with the single biggest reason and the upgrade that would move it.

Pair it with the full code review when you want findings and fix briefs too. The grader answers one question fast: does this ship, and if not, what is the smallest thing that changes the answer.

Instant-F triggers

Any single match is an automatic F. The grade stops there and names the file and line. No other lane can raise it.

Grade caps by surface

Certain surfaces cannot earn an A or B on a passing build alone. The cap is stated in the output when it applies.

  • Auth changes need the bypass and escalation path tested, not just the happy path, or the grade caps at C.
  • Payment and wallet flows need error paths tested and server-side amount and recipient validation, or they cap at C.
  • Data migrations need a documented, tested rollback; no rollback plan caps at D.
  • Breaking public API changes need verified backward compatibility or a versioned migration path, or they cap at C.

Output format

It opens with a plain-language summary a non-coder can follow, then the seven lane grades, the overall grade, any cap, the evidence-backed why, and the ranked required upgrades.

Correctness: A-F
Security and permissions: A-F
Data and state: A-F
Domain truth: A-F
UX and release behavior: A-F
Tests and verification: A-F
Deploy readiness: A-F
Overall: A-F

Best prompts

Use $suede-code-grader to give this PR a blunt A-F ship verdict with evidence and the one required upgrade.
Use $suede-code-grader on this auth change. Apply the surface grade cap and tell me what evidence would lift it.
Use $suede-code-grader on this release build. I want the ship decision, not a full review.

Safety boundary

The grade is a working review, not an audited guarantee. The skill does not invent tests, screenshots, live checks, or deploy status, and it does not raise a grade because the work was hard or because CI passed without exercising the change.

It never ships a C, D, or F without naming the upgrade that would move it.

Cue Suede

Feedback can happen mid-workflow or at the end. Say Cue Suede to ask for choices: change something, preserve what worked so the agent can mimic it later, or keep as-is by saying nothing.

Install this skill

It ships inside the Suede Creator Skills pack, so one install brings the whole pack and this skill with it. Claude Code takes two commands the first time and one after that; Codex installs the same pack natively.

terminal · install the pack
Claude Code
$ /plugin marketplace add JasonColapietro/suede-creator-skills
$ /plugin install suede-skills@suede
Codex
$ codex plugin add suede-skills@suede-codex

Cursor, Copilot, and Windsurf install through the skills CLI. Every path is on the install page, and every folder is readable first on GitHub.

Proof is part of the release.

First install takes two commands. After the marketplace is added, install is one. Every skill is plain Markdown.

$ /plugin marketplace add JasonColapietro/suede-creator-skills → /plugin install suede-skills@suede All install paths