Guide · Agent skill
How to use the Test Reporting agent skill for Hive
Updated · written with AI from the project's own documentation and checked against our catalogue data
The test-reporting skill is an agent skill for the Hive repository. When you ask your agent to run integration tests, it runs the Level 2 dummy-agent test suite and writes a styled HTML report that lists every test with its inputs, expected outcome, actual outcome and any failure traceback. This guide explains what the skill does, how to install it, how to trigger it, and what it needs before you run it.
Integration Test Reportingaden-hive/hive · Apache-2.0 Download the skill Quick start
- 1Download the ZIP. Download the test-reporting skill ZIP from this page.
- 2Install the skill. Unzip it into ~/.claude/skills/ or a project's .claude/skills/, or upload it under Settings → Capabilities → Skills in the Claude apps.
- 3Check the Hive checkout. Confirm you have the Hive repository with the core/ directory, uv, and the pytest and pytest-asyncio dependencies.
- 4Confirm an LLM provider. Make sure an LLM provider is available, and a Claude Code subscription if you plan to run a single file.
- 5Trigger the skill. Ask the agent to run the integration tests, or use /test-reporting with an optional test file name.
- 6Open the report. Find the HTML report at /tmp/hive_integration_test_report.html and review the per-test table and verdict.
What is an agent skill and what does this one do?
An agent skill is a folder containing a SKILL.md file with instructions that an AI agent follows when you invoke it. This skill is defined by a single file, SKILL.md, and tells the agent how to run the Hive integration tests and collect the results.
When invoked, the skill has the agent run either the full Level 2 dummy-agent integration suite or one chosen test file. It then gathers the JUnit XML, console output and summary table, and writes an HTML report to /tmp/hive_integration_test_report.html. The report is aimed at maintainers who want a readable record of a test run rather than raw pytest output.
- Runs the full suite or a single test file
- Collects JUnit XML, stdout/stderr and the summary table
- Writes a styled HTML report with a row for every test, not just failures
- Includes per-test input, expected outcome, actual outcome and failure traceback where relevant
How do you install the test-reporting skill?
Download the ZIP for the skill from this page. Then install it using one of the two options below.
For Claude Code, unzip the file into ~/.claude/skills/ to make it available everywhere, or into a project's .claude/skills/ folder to make it available only in that project. For the Claude apps, upload the ZIP under Settings → Capabilities → Skills.
- Claude Code, personal: unzip into ~/.claude/skills/
- Claude Code, single project: unzip into the project's .claude/skills/
- Claude apps: upload the ZIP under Settings → Capabilities → Skills
What do you need before running it?
The skill only works inside a Hive repository checkout that includes the core/ directory. It uses uv (a Python package runner) and needs the Python test dependencies, including pytest and pytest-asyncio. The tests call a real LLM, so you need an available LLM provider. The single-file path specifically requires a Claude Code subscription, and it raises an error if one is not configured. You also need write access to /tmp.
- Hive repository checked out with the core/ directory
- uv (Python package runner)
- Python test dependencies including pytest and pytest-asyncio
- An available LLM provider; a Claude Code subscription for the single-file path
- Write access to /tmp
How do you trigger the skill with a request?
The skill is triggered when you want to run integration tests and see results. The source defines these trigger forms: /test-reporting for the full suite, /test-reporting with a test file name for one file, and /test-reporting --all for all tests. You can also describe the request in plain language, such as asking the agent to run the integration tests for the queen live component and give you an HTML report.
If you give a specific file or pattern, the agent uses it. Otherwise it runs the full suite.
- /test-reporting runs the full suite
- /test-reporting test_component_queen_live.py runs one file
- /test-reporting --all runs all tests
- Plain request: run the Hive integration tests and write me an HTML report
/test-reporting test_component_queen_live.pyWhat does the HTML report contain?
The header shows the run timestamp in ISO 8601 format, the provider used, total, passed, failed and skipped counts, total wall-clock time, and an overall PASS or FAIL verdict. The per-test table has a row for every test with its component, name, status, duration, a one-line description, setup-action-assertion summary, why it matters, inputs, expected outcome, actual outcome, and full traceback for failures.
The component names map to test files. For example, test_component_queen_live.py is grouped as Queen Phase (Live), and test_component_llm.py as LLM Provider. Failures also get a short analysis covering root cause, impact and suggested fix.
- Header: timestamp, provider, counts, duration, verdict
- Per-test rows for every test, including passes
- Failure rows include traceback and diagnosis
What are the safety notes?
This skill runs scripts and uses the network. It runs Python test scripts in the repository and makes calls to a real LLM, which can be slow and cost money depending on your provider. It also writes files to /tmp. Review the SKILL.md file and the test scripts before running the skill in a repository you do not control.
- Runs scripts from the Hive repository
- Uses the network to call an LLM provider
- Writes the report and JUnit XML to /tmp
- Tests can be slow and costly, so this is not a quick unit test run
What people use it for
- A maintainer producing a shareable HTML summary of a full integration test run
- A developer debugging a failing queen live test who needs per-test inputs and outcomes
- A reviewer checking which tests passed or failed in a single component file
- A team archiving a readable record of a test run instead of raw pytest output
Frequently asked questions
How do I install the test-reporting agent skill?
Download the ZIP from this page and unzip it into ~/.claude/skills/ for all projects, or into a project's .claude/skills/ folder for one project. In the Claude apps, you can upload the ZIP under Settings → Capabilities → Skills.
What do I need to run the test-reporting skill?
You need a Hive repository checkout with the core/ directory, uv, Python test dependencies including pytest and pytest-asyncio, an available LLM provider, and write access to /tmp. The single-file path specifically needs a Claude Code subscription.
Is the test-reporting skill free?
The source does not state pricing or licence terms for the skill beyond its licence, which is Apache-2.0. The tests call a real LLM, which may cost money depending on your provider, so check your provider's pricing.
Why does the single test file run fail with an error?
The single-file path raises an error if no Claude Code subscription is detected among the available providers. Set up a Claude Code subscription or configure the provider, then run it again.
Where is the HTML report saved?
The report is written to /tmp/hive_integration_test_report.html. The JUnit XML is written to /tmp/hive_test_results.xml when the single-file path runs.
Can I use this skill instead of running pytest directly?
Yes, if you want a readable HTML summary with per-test inputs and outcomes. If you only need a quick unit test run, this skill is not the right choice, because these tests call a real LLM and can be slow and costly.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
This skill suits Hive maintainers and developers debugging component tests who want a shareable HTML record of integration runs. It is not suited to quick unit test runs or to anyone without a configured LLM provider, since the tests call a real model and can be slow and expensive.
Download the skill