Create a New Skill from Topic

Generate a production-ready SKILL.md + tiny examples, evals, and an index entry aligned to Anthropic Skills.

Best use case

Create a New Skill from Topic is best used when you need a repeatable AI agent workflow instead of a one-off prompt.

Generate a production-ready SKILL.md + tiny examples, evals, and an index entry aligned to Anthropic Skills.

Teams using Create a New Skill from Topic should expect a more consistent output, faster repeated execution, less prompt rewriting.

When to use this skill

  • You want a reusable workflow that can be run more than once with consistent structure.

When not to use this skill

  • You only need a quick one-off answer and do not need a reusable workflow.
  • You cannot install or maintain the underlying files, dependencies, or repository context.

Installation

Claude Code / Cursor / Codex

$curl -o ~/.claude/skills/core-skill-authoring/SKILL.md --create-dirs "https://raw.githubusercontent.com/williamzujkowski/cognitive-toolworks/main/skills/core-skill-authoring/SKILL.md"

Manual Installation

  1. Download SKILL.md from GitHub
  2. Place it in .claude/skills/core-skill-authoring/SKILL.md inside your project
  3. Restart your AI agent — it will auto-discover the skill

How Create a New Skill from Topic Compares

Feature / AgentCreate a New Skill from TopicStandard Approach
Platform SupportNot specifiedLimited / Varies
Context Awareness High Baseline
Installation ComplexityUnknownN/A

Frequently Asked Questions

What does this skill do?

Generate a production-ready SKILL.md + tiny examples, evals, and an index entry aligned to Anthropic Skills.

Where can I find the source code?

You can find the source code on GitHub using the link provided at the top of the page.

SKILL.md Source

## Purpose & When-To-Use
Use when a developer specifies a new Skill topic and needs a compliant, token-efficient `SKILL.md` with a tiny example, tests, and an index entry.

## Pre-Checks
- **Authoritative time**: Set `NOW_ET` to ISO 8601 in `America/New_York` using NIST/time.gov semantics. Include `NOW_ET` in citations’ access dates.
- **Scope**: Extract intended users, inputs/outputs, success criteria, and token budgets.
- **Sources**: Identify primary standards/docs. Prefer authoritative sources; record title, URL, and access date (`NOW_ET`).
- **Footprint**: Default to Tier 1 unless ambiguity or regulation requires expansion.

## Procedure
### Tier 1 — Draft Minimal Skill (T1≤2000 tokens)
1. Write **METADATA**: concise `name`, `slug`, ≤160-char `description`, `capabilities`, `inputs`, `outputs`, `keywords`, `version`, `owner`, `license`, `security`, `links`.
2. Author compact sections: `Purpose & When-To-Use`, `Pre-Checks`, `Procedure (Tiered)`, `Decision Rules`, `Output Contract`, `Quality Gates`, `Resources`.
3. Add **one** short runnable example (≤30 lines) or precise pseudo-code.
4. Produce `index_entry` JSON for `index/skills-index.json`.

### Tier 2 — Source-Backed Enrichment (T2≤6000 tokens)
1. Gather 2–4 primary sources; add citations with titles, URLs, and `NOW_ET` access dates.
2. Add explicit **Decision Rules** (ambiguity thresholds, abort conditions).
3. Link small `resources/` artifacts (schemas/templates) without pasting large texts.

### Tier 3 — Deep Validation (T3≤12000 tokens)
1. Create `tests/evals_<slug>.yaml` with 3–5 scenarios (happy path, missing input, edge case).
2. Add security notes (secrets handling, PII posture, auditability).
3. Emit `CHANGELOG.md` (v1.0.0) and finalize `index_entry`.

## Decision Rules
- If regulated domains (NIST/OMB/FedRAMP) are involved → require Tier 2.
- If schemas/versions are unknown → emit TODO and **stop** with `needs-input`.
- If example exceeds 30 lines → **truncate** and link to `resources/`.

## Output Contract
- Create `skills/<slug>/` containing:
  - `SKILL.md` (this format)
  - `examples/<slug>-example.txt` (≤30 lines)
  - `resources/` (optional, linked)
  - `CHANGELOG.md`
- Emit `index_entry` JSON with: `slug`, `name`, `summary` (≤160 chars), `keywords`, `owner`, `version`, `entry`.

## Examples
**Minimal example (pseudo-code, ≤30 lines):**
```text
INPUT:
  topic: "OSCAL SSP validator for FedRAMP Moderate"
  constraints: { "strict": true }

OUTPUT (files):
  skills/compliance-oscal-validator/SKILL.md
  skills/compliance-oscal-validator/examples/compliance-oscal-validator-example.txt
  tests/evals_oscal-ssp-validate.yaml
  index_entry (JSON fragment)

INDEX_ENTRY:
  {
    "slug": "compliance-oscal-validator",
    "name": "OSCAL SSP Validator for FedRAMP Moderate",
    "summary": "Validates OSCAL SSP against FedRAMP Moderate baseline requirements.",
    "keywords": ["oscal", "ssp", "fedramp", "validation"],
    "owner": "cognitive-toolworks",
    "version": "1.0.0",
    "entry": "skills/compliance-oscal-validator/SKILL.md"
  }
```

## Quality Gates
- **Token budgets**: T1≤2k, T2≤6k, T3≤12k — enforce via `validate_skill.py`.
- **Example size**: ≤30 lines runnable/pseudo-code; longer samples → `resources/`.
- **Metadata**: `description` ≤160 chars; all required front-matter keys present.
- **Sources**: T2+ requires 2–4 authoritative links with titles/URLs and access dates (`NOW_ET`).
- **Safety**: No secrets, PII, or long pasted texts; `validate_skill.py` checks for secret patterns.
- **Determinism**: Index entries must be unique slugs; `build_index.py` enforces no duplicates.
- **Evals**: 3–5 scenarios in `tests/evals_<slug>.yaml`; must pass basic sanity checks.

## Resources
- **Anthropic Skills Overview** (accessed 2025-10-25): https://docs.claude.com/en/docs/agents-and-tools/agent-skills/overview
- **Anthropic Skills Best Practices** (accessed 2025-10-25): https://docs.claude.com/en/docs/agents-and-tools/agent-skills/best-practices
- **Claude Code Skills Documentation** (accessed 2025-10-25): https://docs.claude.com/en/docs/claude-code/skills
- **NIST Time Services** (accessed 2025-10-25): https://www.nist.gov/pml/time-and-frequency-division/time-services
- **ISO 8601 Date/Time Format** (accessed 2025-10-25): https://www.iso.org/iso-8601-date-and-time-format.html

Related Skills

UX Wireframe Designer

5
from williamzujkowski/cognitive-toolworks

Design user experience wireframes, user flows, and interactive mockups for web and mobile applications using industry-standard notation

TypeScript Tooling Specialist

5
from williamzujkowski/cognitive-toolworks

Generate TypeScript/JavaScript project scaffolding with npm/pnpm/yarn, Jest/Vitest, ESLint/Prettier, and bundling (Vite/Rollup/esbuild).

Python Tooling Specialist

5
from williamzujkowski/cognitive-toolworks

Generate Python project scaffolding with Poetry/pipenv, pytest configuration, type hints (mypy), linting (ruff/black), and packaging (setuptools/flit).

Java Tooling Specialist

5
from williamzujkowski/cognitive-toolworks

Generate Java project scaffolding with Maven/Gradle, JUnit 5, Mockito, Checkstyle/SpotBugs, and packaging (JAR/WAR/native-image).

C# .NET Tooling Specialist

5
from williamzujkowski/cognitive-toolworks

Generate C# .NET project scaffolding with dotnet CLI, xUnit/NUnit, StyleCop analyzers, and packaging (NuGet/Docker).

Unit Testing Framework Generator

5
from williamzujkowski/cognitive-toolworks

Generate unit test scaffolding and test suites for Jest, PyTest, Go testing, JUnit, RSpec with mocking, assertions, and coverage configuration

Testing Strategy Composer

5
from williamzujkowski/cognitive-toolworks

Compose comprehensive testing strategies spanning unit, integration, e2e, and performance tests with optimal coverage.

Load Testing Scenario Designer

5
from williamzujkowski/cognitive-toolworks

Design load testing scenarios using k6, JMeter, Gatling, or Locust with ramp-up patterns, think time modeling, and performance SLI validation.

Integration Testing Designer

5
from williamzujkowski/cognitive-toolworks

Design integration test scenarios with database fixtures, external service mocks, contract testing, and test environment setup for microservices and APIs.

Chaos Engineering Experiment Designer

5
from williamzujkowski/cognitive-toolworks

Design chaos engineering experiments to test system resilience with controlled failure injection, hypothesis formulation, and blast radius control.

Terraform Module Best Practices

5
from williamzujkowski/cognitive-toolworks

Design reusable Terraform modules with variable validation, output schemas, module composition, and testing (Terratest).

Service Level Objective Validator

5
from williamzujkowski/cognitive-toolworks

Validate SLO definitions against actual metrics, generate alerting rules, and design error budget policies with burn rate calculations.