- Home
- Skills
- Prompt & Skill Engineering
- Skill Seller Kit: Turn a Workflow into a SKILL.md That Sells
Works with the AI Agents you already use
Skill Seller Kit: Turn a Workflow into a SKILL.md That Sells
Turn a workflow into a sellable agent skill: 8-question interview, scaffold, spec lint, trigger check, price from 356 real Agensi listings, listing copy, cover and a verified upload zip.
$7
Skill Seller Kit: Turn a Workflow into a SKILL.md That Sells
Example session with this skill installed
skillkit new: every release I write the notes by hand (git log between tags, drop release/merge commits, group features/fixes/docs/deps, link PRs, call out breaking changes, thank contributors). Turn that into a skill I can sell on Agensi.
- Read your context and instructions
- Compiled the seller kit
(Real run, 20 Oct 2026. Full files in examples/release-notes-from-git-2026-10-20/.)
- Interview: 8 questions answered -> 02-spec.json (job, buyers, 6 trigger phrasings, 4 near-miss prompts, inputs, 5 steps, boundaries).
- Scaffold: python3 scripts/new_skill.py --spec 02-spec.json --out .
created release-notes-from-git/ (SKILL.md 491-char description, 5 steps, 3 evals, 8 trigger prompts) - Lint 1: Score 75/100 · P1 1. MISSING_FILE: SKILL.md runs scripts/release_notes.py but it is not in the folder.
- Trigger check 1: likely misses 3: "what's new in this version?" (0% overlap), "prepare the release announcement for 1.20" (25%), "list the PRs merged since the previous tag" (33%). Missing words: version, prepar, announcement, merg, prs, previou.
-> description rewritten (571 chars); trigger check 2: likely misses 0 · false triggers 0. - Real run of the new script on axios v1.19.0 -> v1.20.0 (31 commits):
v1.20.0 (2026-08-19)
30 changes since v1.19.0 · full diff
Features
- HttpStatusCode: Add RFC 9110 aliases for 413 and 422 (#11082)
Bug fixes
- Core methodList, HTTP adapter errors, and add tests (#11096)
- xhr: Flush final progress during the live loadend dispatch (#11121)
interceptors: Tolerate nullish handlers in syncHandlerEntries (#11118)
- Harden runtime option handling (#11141)
http: Stop per-socket error listener from retaining the installing request's context (memory leak) (#11091)
- xhr: Reject navigation-canceled requests as ECONNABORTED (#11094)
...
summary: 30 changes, 1 release commit skipped, 13 bot commits, 11 human contributors, 0 unclassified.
- Lint 2: Score 100/100 · P1 0 · P2 0 · P3 0 (official agentskills validate: Valid skill).
- Price: $3. 5 paid comps, median $5 -> ~70% = $3.50 -> $3. Free rival changelog-generator has 75 installs, so the first line names what it adds.
- Listing: 12/12 checks pass (title 55/60, summary 171/200, 6-row sample table, example 3,017 chars).
- Package: release-notes-from-git-v1.0.0.zip · 10 files · 10,110 bytes · one top folder · integrity ok.
Next: you review upload/, then upload it on Agensi yourself (fields in templates/marketplace-fields.md).
Connects securely to your tools. The creator never sees your data.
About this skill
Unlike free skill creators that stop at a SKILL.md, this kit takes your skill to a sale: spec + review lint, trigger check, price from 356 real Agensi listings, listing copy with 12 checks, cover and a verified zip.
Sample output: real worked example, a release-notes skill built with the kit on 20 Oct 2026 (run on the public axios repo)
| Step | Kit tool | Real result |
|---|---|---|
| Scaffold | new_skill.py | SKILL.md + evals + 8 trigger prompts from the interview spec |
| First lint | lint_skill.py | 75/100: P1 MISSING_FILE scripts/release_notes.py |
| Trigger check | trigger_check.py | 3 likely misses ("what's new in this version?") -> 0 after the description rewrite |
| Real run | the new skill's script | axios v1.19.0 -> v1.20.0: 31 commits, 30 changes, 13 bot bumps folded into 1 line, 0 unclassified |
| Final lint | lint_skill.py | 100/100, and passes the official agentskills validate |
| Price | price_comps.py | $3: 5 paid comps (median $5) + free rival changelog-generator (75 installs) |
| Listing | build_listing.py | 12/12 checks pass |
| Package | package_skill.py | 10 files, 10,110 bytes, one top folder, SHA-256 verified |
What you get
- SKILL.md: 11-step workflow, 8 interview questions, rules, 9 commands
- 7 stdlib Python tools: scaffold, lint (P1/P2/P3 + 0-100 score), trigger check, price comps, listing builder, cover renderer, packager
- Agensi snapshot: 356 listings with price, installs and views (20 Oct 2026) + a 'what sells' report
- 9 templates: spec, listing, trigger prompts, description, example session, FAQ bank, cover, launch posts, marketplace fields (Agensi, MCP Market, Gumroad, AgentMart, GitHub)
- The full real worked example (interview to zip) + the linter run on all 19 skills in anthropics/skills
- Self-test: 33 checks (bash tests/run_tests.sh)
Who it's for
Developers and creators who want to sell agent skills on Agensi, MCP Market, Gumroad or AgentMart, or fix one that got rejected or isn't selling.
What it won't do
- Building MCP servers or Apify actors (the listing steps still apply)
- Cloning another seller's paid skill
FAQ
Why pay when the free skill-creator exists? skill-creator helps you write and test a skill. It has nothing on pricing, listings, covers or marketplace review. This kit covers that half: lint against the spec and review problems, comps-based price, listing copy with checks, cover, verified zip, launch checklist.
Does it upload or publish for me? No. It prepares every field and file; the agent asks before any submit, publish or post, and you click.
Which marketplaces? Field guides for Agensi, MCP Market, Gumroad, AgentMart and GitHub. The skill itself follows the open Agent Skills spec, so it works anywhere SKILL.md folders load.
How current is the price data? The bundled snapshot is 356 public Agensi listings from 20 Oct 2026. price_comps.py reads any CSV with the same columns, so you can swap in a fresh export.
What do I need installed? Python 3.8+ (standard library only). Chrome or Chromium is optional, for the cover PNG.
Is the worked example real? Yes: a release-notes skill built step by step on 20 Oct 2026 and run on the public axios repo (v1.19.0 -> v1.20.0). Every numbered output file is in examples/.
Trust Agent Skills format (Claude Code, Cursor, Codex) · v1.0.0 · passes the official agentskills validator · 33-check self-test · stdlib Python, no API keys · never uploads or publishes without asking
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 9 days ago
Requires