agentsclimarketplace
Sign in

Basic skill

Skill wind-will-not-stop/skillViewer/sample-skills/basic-skill

Demonstrate a flow graph with serial steps, branching, parallel review, tool calls, approval gates, and retry loops.From its SKILL.md

Install
npx -y skills add wind-will-not-stop/skillViewer --skill basic-skill

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
  • runs commandsInstructs the agent to run 2 commands, including `scripts/compose_summary.py` and 1 more.

SKILL.md

1.7 KB, 335 tokens by cl100k_base, as published. Nobody here has run it

Basic Skill

Overview

This skill demonstrates a workflow-oriented graph. It validates a request, fans out into parallel review steps, joins the results, calls a policy helper, runs synthesis, validates the draft, and either delivers the result or routes the work to manual review.

Required Steps

  1. Read references/summary.md to understand the task context.
  2. In parallel, review references/style-guide.md, references/policy-rules.md, and agents/openai.yaml.
  3. Join the review results into a single working context.
  4. Call a policy helper to normalize policy and compliance hints.
  5. Run scripts/compose_summary.py to prepare a draft answer.
  6. If the draft is incomplete, loop back and run scripts/compose_summary.py again with a tighter summary.
  7. Run scripts/validate_summary.py to verify compliance before delivery.
  8. If compliance still fails, route the draft to manual review; otherwise return the final answer to the user.

Decision Logic

  • If the request is missing required context, stop early and ask for clarification.
  • Otherwise continue into the review flow.
  • If the draft misses required points, retry the synthesis step once before finalizing.
  • If the final validation fails, request manual review instead of delivering the draft directly.

Outputs

  • output/final-summary.md
  • output/manual-review.md

Guardrails

  • Do not edit source files.
  • Keep the final answer under five sentences.
  • Keep the retry loop bounded to one additional pass.
  • Only mark the flow as complete after validation succeeds or manual review is requested.

What ships with it: 9 files

50.6 KB alongside SKILL.md, 2 of them executable

agents/

output/

references/

scripts/

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.