Generates Python implementations and tests from spec stubs and contracts.

Install

mkdir -p .claude/skills/jaunt && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/17291" && unzip -o skill.zip -d .claude/skills/jaunt && rm skill.zip

Installs to .claude/skills/jaunt

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

Use when working with the Jaunt spec-driven code generation framework for Python. Trigger for requests mentioning Jaunt, @jaunt.magic, @jaunt.test, spec stubs, jaunt build, jaunt test, jaunt.toml, __generated__ directories, or writing specs/tests that Jaunt will generate implementations for. Also use when the user wants to set up a new Jaunt project, configure LLM providers, debug build failures, or understand the spec-driven development workflow.
451 chars✓ has a “when” triggerlonger than Claude Code's old 250-char listing cap (fine on current versions)
Intermediate

Key capabilities

  • Generate Python implementations from `@jaunt.magic` stubs
  • Generate pytest tests from `@jaunt.test` stubs
  • Provide project-aware instructions for Jaunt workflows
  • Support spec-driven development
  • Configure LLM providers for code generation

How it works

Jaunt processes spec stubs marked with `@jaunt.magic` or `@jaunt.test` to generate corresponding Python code and pytest suites. It uses the OpenAI Codex CLI for this generation.

Inputs & outputs

You give it
Python spec stubs with `@jaunt.magic` or `@jaunt.test` tags
You get back
Generated Python implementations and pytest tests in `__generated__/` directories

When to use jaunt

  • Generating code from specs
  • Running Jaunt build loops
  • Validating contract implementations

About this skill

Jaunt (spec-driven code generation)

Jaunt is a spec-driven code generation framework for Python: you write intent as @jaunt.magic / @jaunt.test / @jaunt.contract stubs and Jaunt generates real implementations and pytest tests into __generated__/ using the OpenAI Codex CLI.

Get the current workflow primer

Jaunt ships its own always-current, project-aware agent briefing. Run it and load the output into context before working:

jaunt instructions          # markdown primer + a live snapshot of THIS project
jaunt instructions --json   # {command, ok, text, project} for tooling/MCP

jaunt instructions is the single source of truth for how to author specs, run the build/test loop, read build status, and use contract mode. This file is kept intentionally thin so it cannot drift from the installed CLI.

Hard rules (if you cannot run the CLI)

  • Never edit files under __generated__/ — they are overwritten on every build.
  • Never hand-write the body of a @jaunt.magic symbol. The docstring is the contract and Jaunt fills in the code; keep the body a raise RuntimeError(...) stub.
  • Pair every @jaunt.magic spec with @jaunt.test coverage.

When not to use it

  • When working with files under `__generated__/` as they are overwritten
  • When hand-writing the body of a `@jaunt.magic` symbol
  • When not using the Jaunt framework for Python code generation

Limitations

  • Files under `__generated__/` are overwritten on every build
  • Cannot hand-write the body of a `@jaunt.magic` symbol
  • Requires pairing every `@jaunt.magic` spec with `@jaunt.test` coverage

How it compares

This skill automates the creation of Python implementations and tests from high-level specifications, contrasting with manual coding and test writing for each feature.

Compared to similar skills

jaunt side by side with the closest alternatives in the catalog.

SkillInstallsUpdatedSafetyDifficulty
jaunt (this skill)027dReviewIntermediate
adk-engineer310dReviewAdvanced
unit-testing-test-generate23moReviewIntermediate
playwright-roll21moReviewAdvanced

Try saying

Example prompts that trigger this skill in your AI assistant.

Search skills

Search the agent skills registry