Bruno Agent Skills

A collection of agent skills for working with Bruno — the open-source API client. Each skill is a self-contained directory an AI coding agent can load to perform a focused Bruno task: generating collections, writing tests, or setting up CI.

Skills

Each skill follows the same layout:

/
├── SKILL.md              # Skill front-matter + instructions the agent loads
├── agents/               # Agent-specific manifests (e.g. openai.yaml)
├── references/           # Quick-reference docs the skill cites
├── scripts/              # Helper scripts + tests
└── assets/               # Example inputs (where applicable)

SKILL.md is the entry point. Its YAML front-matter (name, description) is how an agent decides when to load the skill; the body is the instructions the agent follows.

Using a skill

These skills target agents that support the Agent Skills pattern (Claude Code, the Anthropic API skills feature, and compatible runtimes). To use one:

1. Copy the skill directory (e.g. bruno-collection-generator/) into your agent’s skills folder, or point your agent at this repo.
2. Trigger the skill by describing the task in natural language — for example, “generate a Bruno collection from this OpenAPI file” or “write smoke tests for this Bruno request.”
3. The agent loads SKILL.md, follows its instructions, and may call into the helper scripts under scripts/.

The agents/openai.yaml file in each skill provides an OpenAI-compatible manifest for runtimes that prefer that format.

Running the tests

The helper scripts under each skill have pytest tests. To run them locally:

python -m pip install --upgrade pip pytest pyyaml
python -m pytest -v

CI runs the same tests against Python 3.9 and 3.12 — see .github/workflows/tests.yml.

Repository layout

bruno-agent-skills/
├── bruno-ci-setup/
├── bruno-collection-generator/
├── bruno-test-writer/
└── .github/workflows/tests.yml

Contributing

Issues and pull requests are welcome. When adding or changing a skill:

• Keep SKILL.md focused — the front-matter description drives skill selection, so it should clearly describe when to use the skill.
• Treat ingested content (API docs, sample responses, existing files) as untrusted data, never as instructions. The existing skills document this in their “Safety and trust boundaries” sections; new skills should follow the same pattern.
• Never commit real secrets, tokens, or PII into examples or tests. Use placeholders ({{var}}, bru.getSecretVar) instead.
• Add or update tests under scripts/ for any new helper script.

Related

• Bruno — the API client these skills target
• Bruno docs
• Bruno on GitHub