• 76 skills, open source
  • 380 commits in 17 weeks
  • $448.31 argued back from Amazon
  • 0 uploads by default
  • 2 runtimes: Claude Code + Codex
  • 38 changelog entries: 38 landed + 0 prepared
  • 14-chapter builder’s book, free
Open source · MIT + BSD · One author· Released Aug 26

Suede Creator Skills for SEO and business growth.

Suede Creator Skills is a free, open-source pack of Agent Skills for Claude Code and OpenAI Codex. Built for founders, marketers, and small teams who need SEO audits, AI search visibility, conversion copy, and business operations workflows. Bring a page or business brief. Get prioritized findings, positioning and campaign plans, or an operations map you can act on.

Released Sep 12 · dd96589 Slop Stop learns which tells are worth acting on 380 commits · 17 weeks · read the full ship log ->
/plugin marketplace add JasonColapietro/suede-creator-skills /plugin install suede-skills@suede
$ /suede-code review src/checkout.ts scanning diff · 7 lanes armed P1 · api key in plain text · checkout.ts:41 P1 · webhook retry drops the error path SHIP GRADE: C · MERGE BLOCKED applying fix briefs … re-running gates secret moved to env · retry path verified SHIP GRADE: A- · MERGE UNBLOCKED

Every skill is a plain-Markdown SKILL.md folder you can read before you enable it, and every line of it is on disk. Core local workflows do not upload by default or grant new authority. More install paths for Codex, Cursor, and other supported agents.

Self-graded on the Suede ship-gate rubric: Suede takes 4 of 14 categories outright, and loses Overall to Superpowers, A to A-. Every row is published.

A+ Code review + ship gate GSD A- · Superpowers A A+ Breadth of coverage GSD A · Superpowers B A+ Evidence integrity GSD B · Superpowers B A+ Domain expertise GSD B · Superpowers C

One rubric, three packs, read from their published files on 2026-07-05: see every row.

Changelog · landed + prepared

A review pack had better survive its own review.

So the work ships in public. Released entries cite the commit that landed on main. A newest prepared entry is explicitly unreleased and cites only the current branch's merge base until review and merge provide a landing hash. No "various improvements."

Version 0.19.0 released Sep 12, 2026

380 commits · 17 weeks · newest on the right

  1. 2026-09-12 Rework dd96589

    Slop Stop learns which tells are worth acting on

    Version 0.19.0 reworks suede-deslop, which weighted every pattern the same and so edited hard against prose that was fine. Its rules now carry two confidence tiers. Chat residue, formulaic structure, throat-clearing and vague declaratives justify an edit on one sighting; passive voice, stacked qualifiers, hyphenated pairs, curly quotes, a lone em dash and a single triad are marked weak alone and need company in the same passage. A new When not to act section protects quotations, proper names, salutations, technical senses of watched words, and the specifics that carry a writer's voice. Two rules cover ground the pass used to walk past: rule 9 strips chat and draft residue, from assistant wrappers and knowledge-cutoff disclaimers to a gap patched with a plausible guess, and rule 10 removes formatting applied by rule. Both do most of their damage in READMEs, documentation and agent instruction files, which is where the pass runs most. The tiering and the two new rules are adapted from Humanizer (MIT), which draws its pattern set from Wikipedia's WikiProject AI Cleanup guide to signs of AI writing; both are credited in the skill and its kill list.

  2. 2026-09-06 New skills 317f6d5

    Google Play delivery and cross-surface parity join the pack

    Version 0.18.0 adds two skills. suede-play-release takes an Android build from credential preflight through AAB upload, track promotion, and a staged rollout, then reads back from the Play Developer API what production is actually serving. suede-parity-contract generates a machine-readable contract from one surface's live constants and makes every other surface assert against it, so a threshold that drifts between web, iOS, and Android fails a test instead of reaching a user. Both ship in the full plugin; the pack now counts 76 skills.

  3. 2026-09-05 Fix d8ba14d

    The marketplace loads again, and the 74 skills pass an audit

    Version 0.17.0 fixes the plugin manifests so the four marketplace entries load in Claude Code 2.1: the full pack is a strict manifest, the three subsets root at skills/, and the Graph Flo XR agent profiles ship inside their skill folder. A read-through of the 74 skills repaired 19 defects, among them two truncated Codex descriptions, two stale Suede Ship Gate names, a dead slash-tool reference, an advisory-gate contradiction, two NOT FOR clauses the validator could not see, and nine catalog descriptions resynced to their skills. DISTRIBUTION.md records where the pack is listed and where it is not, the homepage proof tape says 74 and is now guarded, and a vendorable frontend-design-review copy gives single-skill directories a stable link.

  4. 2026-08-26 Design ae08836

    Both design skills get a vetted component source

    Version 0.16.0 gives suede-design and johnny-suede-design one vetted UI component-source reference each, listing only sources that cleared a live load, a license read, and a threat-reputation scan, each pinned to its source repository. A seven-check adoption checklist governs every import, one evaluated source is held as not-yet-vetted with its promotion conditions stated, and the reference loads only during component imports.

  5. 2026-08-21 Orchestration d8a6ae8

    Graph search replaces the fixed shipping DAG

    Version 0.15.0 replaces suede-graph-flo-xr's fixed single-plan DAG with a bounded Graph-of-Thoughts search. It generates competing plans, scores and prunes them, adversarially refutes and improves survivors, aggregates compatible lanes, then builds, reviews, and gates only the selected plan. Light, standard, and deep stop at 55, 110, and 200 total agent calls. Production access remains read-only; the skill never deploys.

  6. 2026-08-16 MCP 1e2d70e

    Six specialities, and an MCP that navigates them

    Version 0.14.0 splits the lopsided specialty and gives the MCP a way through the whole pack. The five-way cut shipped hours earlier left Convert & earn at 23 while every other bucket sat between 9 and 15, so it splits on its real seam: Demand channels 8 and Revenue & retention 15, alongside Ship 15, Craft 9, Get found 12, and Strategy & rights 14. Largest bucket 23 to 15. The MCP could previously only be pointed at the three bundle areas; it now ships list_suede_specialties and a suede://specialties resource so a client picks a group before asking for skills, accepts --profile for any one specialty, and advertises only the specialities a given profile can serve. Closes a real hole on the way: amazon-returns-recovery and subscription-recovery sat in the catalog and on the site behind no registered server, unreachable by any MCP client. consumer was a valid profile that .mcp.json did not register.

  7. 2026-08-16 Rework 67c4c7f

    The pack gets five specialities

    Version 0.13.0 re-cuts how the pack is browsed. The 73 skills were grouped four ways by area (marketing 41, workflow 25, creator 5, consumer 2) and separately eight ways on the catalog page, with two more inconsistent cuts on the homepage. Marketing was a 41-name wall in every one of them. There are now five specialities, each with named sub-lanes: Ship 15, Craft 9, Get found 12, Convert & earn 23, Strategy & rights 14. The largest bucket goes 41 to 23. area is deliberately unchanged. It still decides which plugin bundle and MCP profile ships a skill, so nothing moved between install surfaces. list_suede_skills and search_suede_skills take a specialty filter, and the per-specialty counts on all three site surfaces are generated guards, not hand-typed numbers.

  8. 2026-08-16 Site 687d7b9

    Every page in the estate shares one mobile nav

    The last page still carrying its own navigation markup moves onto the shared no-JS-safe <details> disclosure, and a drift guard now covers all 101 pages, so a page added later cannot ship a nav that behaves differently from the rest of the estate.

  9. 2026-08-16 Perf 51bcfd0

    Interior pages stop re-downloading their stylesheet

    The interior-page CSS was inlined into every page, so a reader moving between skill pages paid for the same bytes on every navigation. It is now one extracted stylesheet the browser caches once and reuses across the estate.

  10. 2026-08-16 Site f7ca90b

    All 73 skill pages get the same install block

    Install instructions differed page to page in wording and placement. Every skill page now carries an identical install block with a copy button, so the path from reading a skill to running it is the same everywhere.

  11. 2026-08-16 Site 86f7c99

    The estate gets a display face and a truth sweep

    Fraunces lands as the display face, the hero gains contrast, and every stated pack figure is re-measured against the real count after the pack moved 71 to 73 — the growth had left stale numbers on pages the existing guards did not cover. Ships with a set of mobile P0 fixes found in the same pass.

  12. 2026-08-15 Review ebb5aff

    suede-code-review adopts upstream review techniques

    suede-code-review folds in review techniques from Matt Pocock’s public agent skills: sharper diff-reading order, tighter evidence requirements on each finding, and clearer separation between what the reviewer observed and what it inferred.

  13. 2026-08-15 Rework 1163506

    The entire skill pack gets its first major redesign

    Version 0.12.0 is the first estate-wide redesign of all 73 skills. Each one was audited and reshaped for clearer routing, lower instruction overhead, and stronger cross-agent portability. Hot-path SKILL.md files shed 1,149 lines overall. Long workflows moved into progressive-disclosure references, trigger descriptions now say when to use and avoid each skill, and orchestrators gained explicit model and fan-out bounds. The portable SKILL.md format gives Grok and other skills-compatible agents a leaner signal while preserving the Claude Code, Codex, and MCP install surfaces.

  14. 2026-08-10 Routing 672740e

    suede-clip-to-guide gets its boundaries back

    suede-clip-to-guide turns a clip, talk moment, or transcript into a package that carries short-form attention to a long-form asset: moment scoring, rights routing, exact post copy, and an approval bundle. It shipped with a catalog description condensed to a single sentence, and the condensing dropped the NOT FOR: block that keeps it clear of suede-video and suede-social requests. The catalog entry is now copied verbatim from the skill's own frontmatter, so the two cannot disagree again, and a clip-social-video routing group proves the separation holds in both directions.

  15. 2026-08-10 MCP a49a998

    Skill search stops needing your exact wording

    Version 0.11.1 fixes search_suede_skills matching tokens exactly, so "my landing page is not converting" never reached suede-site-alchemy. The catalog says "conversion". A shared four-character stem now counts for half an exact match, and a stopword list replaces the three-character floor that was dropping ai, qa, ux, and cro. A 15-case routing eval now runs in CI: top-3 accuracy went 13/15 to 15/15.

  16. 2026-08-09 MCP 15de94d

    41 skills come out from behind the MCP

    Version 0.11.0 gives the marketing and consumer lanes a profile that can reach them. Before this, 41 of 71 skills were in the catalog but behind no server anyone shipped. Adds a suede_marketing_mcp registration, a search_suede_skills tool that ranks skills against a task description, and includeBody on get_suede_skill so a client can read the instructions, not just the summary. The competitive-claim guard, which two copy edits had quietly downgraded to a skipped warning, fails hard again.

  17. 2026-08-04 New skill a086a81

    Instagram growth gets an evidence system

    Version 0.10.0 adds suede-instagram-growth: authorized 30-post account audits, attention-versus-conversion mapping, scored ideas, Reels, carousels, Stories, repurposing, experiment cards, and an approval-gated daily operating loop. The pack moves 68 → 70 and the marketing lane moves 38 → 39.

  18. 2026-08-04 Route fix 831214b

    Shipping and CI stop sharing one name

    suede-graph-flo-xr-gate becomes suede-ci-gate, separating the CI and branch-protection owner from the end-to-end suede-graph-flo-xr DAG. Catalog, install, routing, documentation, and sitemap references now resolve to the same canonical name.

  19. 2026-08-03 New skill e51c635

    The canonical shipping DAG joins the pack

    suede-graph-flo-xr runs nontrivial changes through scout, multi-lens research, gap criticism, collision-free lane planning, parallel build, dual review, adversarial refutation, integration gates, and release verification. The pack moves 67 → 68.

  20. 2026-08-01 Fix 67a2b51

    Reserved MCP metadata no longer breaks startup

    Version 0.9.2 accepts the protocol-reserved _meta object sent by current Codex clients while keeping closed parameter validation for every unrelated key. Both Suede MCP profiles now complete discovery instead of failing at tools/list.

  21. 2026-08-01 Fix f7cd12e

    The thirty-eight marketing skills stop being invisible

    The marketing and growth lane opened inside the gold install band instead of the catalog, so every skill name in it rendered gold on gold at a contrast ratio of 1.0. The same thirty-eight were absent from llms.txt, which listed 29 of 67, and their plugin was advertised on no page. Catalog titles, share-card alt text and lane badges still read 28. Each fix ships with a validator guard, so the gap cannot reopen silently.

  22. 2026-08-01 Fix 8e30eca

    Seventy-two pages become keyboard passable

    Every page carries the same nav, and 72 of them had no way past it. All 67 skill pages plus the catalog, install, copy bank and both blog pages now ship a skip link targeting a focusable <main>, per WCAG 2.4.1. Four homepage section labels also stopped using the fill-only red as text, moving from 2.03:1 to 4.56:1 and above once every ancestor background is composited, per WCAG 1.4.3.

  23. 2026-07-28 Rework 189ae51

    The marketing stack becomes Suede-native

    Version 0.9.1 rewrites thirty-eight marketing and growth skills around Suede contracts, evidence and authority boundaries, interruption-safe state, deterministic routing, and aligned metadata, evals, references, and docs. Licensed source acknowledgements remain in the repository-level legal notices.

  24. 2026-07-27 New skills f4e0b20

    The demand side arrives

    Version 0.9.0 adds thirty-eight marketing and growth skills: paid acquisition and outbound, monetisation, lifecycle and retention, and the measurement and operations layer, plus ASO, video, image, competitor work and customer research. Licensed source acknowledgements remain in the repository-level legal notices. The pack goes 29 → 68.

  25. 2026-07-27 New skill f0c3751

    The anti-slop pass joins the pack

    Version 0.8.0 adds additional rules for slop. suede-deslop strips the AI tells from finished prose against a merged kill list, scores five dimensions out of 50, and never touches a fact. The pack goes 28 → 29, with the docs synchronized to match.

  26. 2026-07-27 Fix 6cc4e3f

    Fleet runs survive a sleeping Mac

    suede-codex-fleet now spawns every worker under caffeinate -i: the machine holds awake for exactly as long as workers run, releases the moment the fleet drains, and the slept-mid-run failure signature gets its own troubleshooting entry.

  27. 2026-07-25 New skill 7cc263a

    Full Send joins the public pack

    Version 0.7.0 adds the outcome router, one honest Codex marketplace entry, Codex-native plugin metadata, workflow MCP discovery, deterministic triggers, and synchronized full-pack docs.

  28. 2026-07-20 Site cf0e5a0

    Every public surface moves onto one design language

    Skills catalog v2 with its own written-out ship log, guide v2 (1,013 lines of old design system deleted), copy bank with a Copy button on all 17 blocks, and an install page where every command is one click. Same tokens everywhere.

  29. 2026-07-20 Policy 33ebb14

    Twenty-two skills make their gates advisory

    Checks and ship verdicts now return recommendations instead of treating a gate as permission to stop. High-risk findings pause for the user to choose the next step.

  30. 2026-07-20 Fix aeaa416

    Reserved internal names can never reach public copy

    The validator now scans every public text file against hashed sequences, so a slip fails the build instead of shipping.

  31. 2026-07-19 New skill 6acc19f

    Android lane opens: android-app-factory + suede-recommend-next-action

    One prompt to a Play-Store-ready Kotlin/Compose build, plus a skill that reads repo state and names the single next move. The pack goes 25 → 27.

  32. 2026-07-18 Fix df4c925

    MCP server finds its skills from any install path

    suede_creator_mcp now resolves its catalog through CLAUDE_PLUGIN_ROOT instead of assuming a fixed layout, so marketplace-cache installs stop coming up empty.

  33. 2026-07-13 Site 7cdb9df

    Homepage v2: the hero terminal rotates all six lanes

    Ten commits in one sitting: a live sample session that cycles disciplines, a lane index, and copy with an actual hook.

  34. 2026-07-12 New skill a2db2a8

    Consumer-recovery lane: the negotiator points at your bills

    amazon-returns-recovery runs fee discovery and live-chat disputes; subscription-recovery generalizes it to any recurring charge.

  35. 2026-07-09 Perf 778946a

    LCP: above-the-fold previews load first

    Skill preview cards above the fold get fetchpriority="high" with async decoding; everything below the fold stays lazy.

  36. 2026-07-08 Fix 5046f9c

    Artist profile no longer returns zero skills

    The MCP artist filter matched a skill.area no catalog entry has; it now maps to the creator-tagged skills. classify_file also stops reading ancestor directory names outside the project.

  37. 2026-07-05 New skill 6c38b9c

    site-to-ios-app: live URL to App Store archive

    Wrap an existing site into a submittable iOS shell. The mobile lane opens.

  38. 2026-07-05 Fix e1fc5b5

    Skill-count drift guard in the validator

    The build now fails whenever public copy disagrees with the number of skills actually in the pack. This page can't quietly lie.

Full history on GitHub →

Six specialities, one install

Find the skill for your next business task.

Choose a workflow for search visibility, content, acquisition, conversion, or operations. Each skill defines the inputs it needs and the work it returns. Code review, app delivery, and creator-rights tools support the work that follows.

Ship Orchestration & routing, Code review & CI, Apps: iOS & Android 17 skills Craft Design & copy, Creative assets 9 skills Get found Search & AI visibility, Content & audience 13 skills Demand channels Paid & outbound, Lifecycle messaging, Acquisition assets 8 skills Revenue & retention Money & funnel, Measure & operate, Consumer recovery 15 skills Strategy & rights Strategy & research, Creator toolkit 14 skills

The Ship DAG

Max effort in. Proof or one blocker out.

Tell most agents "max effort, fix everything" and you get vibes. /suede-graph-flo-xr turns that intent into a structure: one authorized controller, every useful non-colliding lane in parallel, adversarial reconciliation where lanes attack each other's claims, a call budget the file caps at 55, 110, or 200, and a close that is either concise proof or one named blocker. It reads production and never deploys.

The ship pipeline A broad authorized outcome flows through one controller into parallel code, docs, CI, and live-verification lanes, then adversarial reconciliation, then proof. Broad authorized outcome One controller single authority Code lane Docs lane CI & merge-gate lane Live verification lane Adversarial reconciliation lanes attack each other's claims Proof

House line: "Never end your allocation above zero." Dry joke, not a literal token promise. The Suede Thought Graph repo search ships as /suede-graph-flo-xr, and the separate copy-only DAG as /suede-ship-copy.

Benchmarks: self-graded, losses shown

Suede against GSD and Superpowers, losses included.

We ran the Suede ship-gate rubric on Suede Creator Skills, GSD, and Superpowers the same way, reading each pack's published skill files, on 2026-07-05. We wrote the rubric, so read this as our reading of three packs, not a neutral benchmark. Superpowers scores higher than Suede on Overall. The full category breakdown is below, including the rows where GSD or Superpowers beats Suede.

Use caseWinnerGrade
Suede product / company work Suede A+
Public copy, design, SEO, rights, launch Suede A+
Long project lifecycle GSD A+
Strict engineering discipline Superpowers A+
Multi-agent orchestration Suede / GSDtie A
Skill ecosystem maintainability Superpowers A+
Ship-risk grading Suede A+

Suede wins outright on 3 of 7 use cases, plus a tie with GSD on multi-agent orchestration. GSD takes long project lifecycles. Superpowers takes strict engineering discipline and skill ecosystem maintainability. Install the pack that matches this week's work.

See the full scorecard
CategorySuede SkillsGSDSuperpowers
Overall A- A- A
Breadth / Coverage A+ A B
Planning / Specs A- A+ A
Execution Workflow A- A A
Debugging Method A- A A+
Testing Discipline B+ A A+
Verification Discipline A A A+
Git / Worktree Hygiene B+ A A+
Durable State / Resume A- A+ B+
Code Review / Ship Gate A+ A- A
Skill Authoring / Governance B+ B+ A+
Evidence / Legal Boundaries A+ B B
Domain Expertise A+ B C
Install / Portability A- B+ A
Estate Cleanliness B+ A- A

Work on search visibility, conversion, and delivery.

01 Code review → ship verdict

Review that ends in
a blunt A-F grade.

Deep findings across TypeScript, React, Next.js, Swift/iOS, OWASP, and database checklists, then a letter verdict across 7 evidence-backed lanes. A hardcoded secret is an instant F with the exact file and line. No other lane can raise it. iOS gets its own traps list: force unwraps, main-thread violations, and native contract drift that crashes shipped apps.

02

Cheap to carry

An installed skill costs your context its name and its description. The body loads only when a request matches it, so you carry the whole folder for about the price of its table of contents

03

Agent teams

The handoff will not close without evidence. Scout, build, review, verifier, and release lanes share one context instead of acting like siloed subagents, with WIP collision detection and rollback trees underneath

04

AI evals

A release that fails its own eval gate does not ship, and the spec names the case that killed it. LLM, RAG, classifier, and agent surfaces each become a measurable AI-SPEC with failure-mode rubrics and eval cases

05

CI ship gate

One required check that cannot deadlock, which is the failure mode every branch-protection setup hits. It reads your stack and your lockfile, writes the CI, and wires the protection rule to that single check

06

Suedify

Point it at any URL. It reverse-engineers the layout logic, type rhythm, and color strategy that make that site work, then rebuilds your page in that language, with your own copy and your own assets. Nothing traced, nothing copied

07

Conversion copy

Twelve headline formulas, and the framework that fits the reader's state: AIDA, PAS, StoryBrand, or JTBD. Every draft is scored out of 70 across seven dimensions, and below 58 it goes back before a word ships, below 62 for a launch surface

08

SEO, AEO & AI EO

Nine audit lanes in one pass: technical, intent, metadata, structure, schema, AI citation readiness, copy quality, E-E-A-T, and topic clusters. It hands back the exact rewrites, not a list of complaints

09

Visibility grade

Tells you whether a page is worth the traffic you are about to buy, A to F, and names the screen that lost the grade: first-screen clarity, CTA pull, proof, AI readability, schema depth, search result fit

10

Site to iOS app

Takes a live URL through the App Store 4.2 wrapper-risk gate, a native or Capacitor shell, and the store screenshots. The same terminal that runs this pack has archived and uploaded 8 iOS apps

11

Design, from first principles

Never a template with your logo dropped in. It commits to a named aesthetic lane and writes that choice into the file before it picks a single color, then builds color, type, and motion up from it

12

Consumer recovery

The same negotiation discipline pointed at your own accounts: one skill argued Amazon into refunding $448.31, the other audits Netflix, gyms, and every App Store subscription you forgot you had. One of those refunds had already been denied after the return window closed

13

Taste memory

Say "Cue Suede" mid-run and choose: change something, preserve what worked, or accept as-is. The choice carries into every pass after it, so the pack accumulates judgment: what to do, what not to do, and when to stop

14

Android app factory

One prompt to a Play Store-ready native app: Kotlin + Jetpack Compose scaffolding, Play Billing, ASO listing copy sized to Play's character limits, and a signed release

15

Recommend next action

Ask "what's next" and get one move, written as a short prompt you can paste and run. It is scored against the goal, the clock, and the rework it prevents, and it opens into a full operator brief only when you ask for one

The router

Don't memorize a command list.

Ask in plain words. The umbrella skill, /suede-workflow-skills, reads the request and loads the right lane: one install, every skill reachable. Every description also carries a NOT FOR line naming the sibling that should win instead, which is why breadth does not turn into an argument inside one prompt. Agents that speak MCP get the same routing as structured tools: list_suede_skills finds the skill, get_suede_skill loads it.

The optional MCP server ships in the repo and runs local with no dependencies: 9 tools, 7 resources, and 5 prompts over stdio. Wire-up lives on the install page.

"grade this diff before I merge" /suede-code "why isn't this page cited by AI answers" /suede-seo-audit "make this landing page much less boring" /johnny-suede-design "turn this site into an iOS app" /site-to-ios-app "plan the rollout for this single" /suede-campaign-in-a-box "is Amazon still holding money on me" /amazon-returns-recovery
76
Public skills
2
runtimes: Claude Code + Codex
$448.31
clawed back from Amazon by one skill, one sitting
0
uploads by default

Install the open-source pack in two commands.

Two lines in Claude Code, about thirty seconds, and the whole pack is on disk. Prefer less? suede-marketing@suede (42 growth skills), suede-agent-workflows@suede, and suede-code@suede install focused subsets, and install.sh covers the clone route. Bring your own company, repo, page, or product.

/plugin marketplace add JasonColapietro/suede-creator-skills /plugin install suede-skills@suede
Supported agents: Cursor, Copilot, Windsurf, and more npx skills add JasonColapietro/suede-creator-skills
Prefer a clone: it copies the pack into ~/.claude/skills/ git clone https://github.com/JasonColapietro/suede-creator-skills.git && bash suede-creator-skills/install.sh
Browse all skills Star on GitHub GitHub repo Codex + MCP install docs Read the full guide

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

The full pack

Read the source of any skill before you trust it.

Every skill is a plain-Markdown SKILL.md folder, with its references, scripts, and templates beside it. Click any row for its docs, or browse the full catalog with install paths.

Ship

17 skills

Get the change built, reviewed, and merged.

Orchestration & routing

suede-agent-teams Coordinate agent lanes with WIP collision detection, RFC mode, rollback trees, and a signed handoff -> suede-recommend-next-action Scores 2-4 candidate moves on goal fit, urgency, and leverage, then hands back one recommendation as a short runnable prompt -> suede-graph-flo-xr Suede Thought Graph shipping search: competing plans are scored, pruned, refuted, improved, and aggregated before only the selected plan mutates; 55/110/200-call caps; production reads only -> suede-ship-copy A separate copy-only DAG for one high-stakes piece of writing: blind research, claim audit, section writers, refutation, publish gate -> suede-workflow-skills Umbrella skill that loads the full pack as one workflow ->

Code review & CI

suede-ai-eval AI-SPEC artifacts, failure-mode rubrics, eval cases, and acceptance gates for AI surfaces -> suede-code Deep review plus the A-F ship grade in one pass, from web stacks to Swift/iOS -> suede-code-grader A-F ship verdict only: 7 evidence lanes, instant-F triggers, grade caps for auth and payment surfaces -> suede-code-review Findings-only review with TypeScript, React, Next.js, Swift/iOS, OWASP, accessibility, and database checklists -> suede-mcp-qa QA pass for the bundled Suede Skills MCP server before it ships -> suede-ci-gate CI that gates the merge: stack and lockfile detection, one required check, branch protection ->

Apps: iOS & Android

android-app-factory One prompt to a Play Store-ready native app: Kotlin + Jetpack Compose scaffolding, Play Billing, ASO listing, signed release -> site-to-ios-app Convert any website into an App Store-ready iOS app: URL audit, 4.2 wrapper-risk gate, Capacitor or native shell, release gate ->

Demand channels

8 skills

Reach people who have not heard of you yet.

Paid & outbound

suede-ads suede-cold-email suede-prospecting

Lifecycle messaging

suede-emails suede-sms

Acquisition assets

suede-free-tools suede-lead-magnets suede-referrals

Field notes

Why memory belongs in a skill, not a prompt.

The design decisions behind the pack, written up as essays: how 76 skills stay cheap to carry, how they route without colliding, and where judgment should live when a prompt is the wrong place for it.

Sep 2026 Two skills, and the bugs that wrote them Google Play from the terminal, and a contract that keeps four surfaces honest Aug 2026 71 skills installed. Your agent reads almost none of them. Progressive disclosure, Suede Thought Graph shipping search, and the MCP layer Aug 2026 Why breadth is free now, and what that changes The economics of a big pack Aug 2026 NOT FOR: the two words that make a big skill pack work How skills route without colliding Aug 2026 Your memory file is a tax you pay on every prompt Why memory belongs in skills Jul 2026 Two new lanes: a next-action recommender and a one-prompt Android app factory What shipped and why it exists

All essays ->

Two commands from here

Put the skills to work on your business.

Start with an SEO audit, a positioning brief, or an operations assessment. Read the workflow, install the pack, and bring your own business context.

Browse all skills
/plugin marketplace add JasonColapietro/suede-creator-skills → /plugin install suede-skills@suede

What it costs you, and what it will not do.

What are Agent Skills?

Agent Skills are SKILL.md folders that Claude Code and OpenAI Codex load on demand to follow a documented workflow. An installed skill costs your context its name and its description, and the body loads only when a request matches it, so the pack sits cheap and runs narrow. Original work in this pack is MIT licensed, and adapted components retain their upstream notices, including BSD-3-Clause terms for Graph of Thoughts. Every folder is plain Markdown you can read before you install it.

How do I install the pack?

In Claude Code, run /plugin marketplace add JasonColapietro/suede-creator-skills, then /plugin install suede-skills@suede. In Codex, run codex plugin marketplace add JasonColapietro/suede-creator-skills --ref main, then codex plugin add suede-skills@suede-codex. The direct Codex skill installer remains available for one selected skill. Full paths live on the install page. Start with suede-seo-audit to review a page and get prioritized fixes, or suede-product-marketing to document your audience and positioning.

Does it work with both Claude Code and Codex?

Yes. Each skill is a standard SKILL.md folder. Claude Code and Codex both install the full pack through their native plugin marketplaces. Cursor, Copilot, Windsurf, and other agents that read skills take the same pack through one npx skills add command; direct folder and single-skill installs remain available.

Is it safe to run?

The skills inspect local files and pages and write local reports. They do not upload files by default, write to any registry, request secrets, or claim legal clearance. Original work is MIT licensed; adapted components retain their upstream notices. The pack is inspectable before you enable anything.