Antithesis Skills
antithesis-skills
Official Antithesis skills: the agent analyzes your system, picks reliability properties, builds a harness, launches tests and triages the results
- 299 ★ on the repository
- Last commit
- Apache-2.0
Broad access
Why this level
- Launches runs in Antithesis under your API key, and mutation testing spends the run budget
- Creates an antithesis/ directory in the project and adds checks to the code under test itself
- Builds images with docker or podman and starts the harness containers
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/antithesis-skillsDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- Launches runs in Antithesis under your API key, and mutation testing spends the run budget
- Creates an antithesis/ directory in the project and adds checks to the code under test itself
- Builds images with docker or podman and starts the harness containers
This is third-party code. Review the repository files before installing.
What it does
Antithesis Skills is a set of twelve skills from Antithesis itself that take an agent from a first look at a project to triaging runs on the Antithesis testing platform. The core chain of three skills runs in order, each in a fresh context, with a person reviewing the result between steps. `antithesis-research` reads source code, comments, docs, commit history and issues, launches subagents and writes into `antithesis/scratchbook/` an analysis of the system, a catalog of testable reliability properties and a container topology proposal. `antithesis-setup` prepares the harness, the containers in which the system runs inside Antithesis: it creates an `antithesis/` directory with a Dockerfile, `config/docker-compose.yaml` and a `setup-complete.sh` script and verifies the build locally, and `antithesis-setup-k8s` covers Kubernetes. `antithesis-workload` writes test commands into `antithesis/test/` and places checks at chosen spots in the code under test. `antithesis-launch` builds the images, runs `snouty validate` and only after success submits `snouty launch` for the requested number of minutes. `antithesis-triage` analyzes run reports through `snouty` and `jq`, and `antithesis-debug` executes scripts inside a run with `snouty runs exec`. The remaining skills check properties with mutation testing, review input generation, build a workload for one feature, answer from the docs and help you report problems in the skills themselves.
Who it is for. For developers and testers who connect their system to the Antithesis platform and want an agent to handle reconnaissance, harness setup, workload writing and first-pass triage of findings.
Good fit when
- You are connecting a project to Antithesis for the first time and need a catalog of reliability properties worth testing
- You need to build a Dockerfile and docker-compose for the harness and confirm it builds locally
- Kubernetes manifests must be adapted and minimized so they run inside Antithesis
- A run ended with failed properties and you need to find events across timelines and download logs
- The baseline run is green and you want to know whether your properties can fail on a real bug at all
Not a fit when
- You have no access to Antithesis: the skills launch and analyze runs through `snouty` and the platform API key
- You want an ordinary unit test suite: the skills build workloads and properties for Antithesis, not coverage in the usual sense
- You work on Windows: the README declares compatibility only with macOS and Linux
- You have no Docker Compose v2 and Docker or Podman: building images is part of the core skills
- You do not want to spend time and tokens: by the authors' own warning `antithesis-research` runs over 30 minutes on most codebases
Example request
/antithesis-research Research my codebase and prepare a plan to test it with AntithesisLimitations
- You need the `snouty` CLI version 0.8.0 or newer, Docker Compose v2, Docker or Podman and npm. Platforms: macOS and Debian-based Linux, with installation steps in PREREQUISITES.md.
- The skills were tested in Claude Code and Codex, designed and tested with Claude Opus 4.6. They work best with agents that can spawn subagents for self-review.
- The skills are under active development, and the authors warn that model non-determinism may change results. `antithesis-setup-k8s` offers fewer tools than `antithesis-setup`, notably around instrumentation and the SDK.
- Launching requires Antithesis credentials: the `ANTITHESIS_API_KEY` key, a tenant name and the `ANTITHESIS_REPOSITORY` variable, and `antithesis-launch` sends nothing without a number of minutes.
- The plans and reports from `antithesis-research` need manual review, and the authors suggest fixing them before the next skill.
- Apache-2.0 license.
How to disable. Run `npx skills remove` and select the skills with the antithesis prefix. The `antithesis/` directories generated in your projects and the checks added to your code stay, so remove them yourself. Runs already submitted to Antithesis remain in your tenant.
FAQ
In what order do I run the skills?
First `antithesis-research`, `antithesis-setup` (`antithesis-setup-k8s` for Kubernetes) and `antithesis-workload`, each in a fresh context, with the changes reviewed in between. Then `antithesis-launch`. Once the baseline run is green, `antithesis-mutation-testing` joins in. The authors advise starting with short 15 to 30 minute runs as smoke tests.
What files does antithesis-research create?
In `antithesis/scratchbook/`: sut-analysis.md with architecture and failure-prone areas, existing-assertions.md, property-catalog.md with priorities, deployment-topology.md, per-property evidence files in properties/, property-relationships.md and evaluation reports in evaluation/.
What does mutation testing spend?
The skill builds an image and launches an Antithesis run per property, plus a baseline run and a full re-sweep after every round of fixes. Before starting it asks for a ceiling on the number of runs and concurrent launches. Results go to `antithesis/scratchbook/mutation-testing/`: a patch per mutant, evidence and a report.md report.
What can antithesis-debug do?
It runs a bash script inside a run at a chosen moment with `snouty runs exec` through the Antithesis multiverse debugger. The references cover scripted forensics, process and container inspection, on-disk evidence, gdb attach and the limits of virtual time.
Which external programs do the skills call?
`docker compose` or `docker-compose`, `docker` or `podman`, `snouty`, `jq`, and for mutation testing also `git` and `rsync`. The skills do not set up permissions themselves, the agent asks for confirmation, and a README table lists what each skill needs so you can pre-approve it.
Discussion
No questions or reviews yet.
Related
SkillsAn adversarial code-audit skill: Hunter finds bugs, Skeptic challenges them, Referee rules, and no source edit happens without your permission
A CLI to evaluate and evolve agent skills: declarative YAML cases, runs across engines and structured reports
Unit tests for skills: checks that an agent finds and uses your skill. Node 20 and Docker are needed, plus a model key
QA agents, skills and an MCP server: test generation, coverage gaps, flaky tests, security scans and chaos testing
A skill that lets an agent write and run Playwright scripts: page tests, forms, screenshots and link checks
Tester
Тестер
A scenario testing system for 1C:Enterprise with a set of agent skills that let an AI agent operate the application like a regular user
README in short
The description starts with an overview of the skills and a warning about the long run of `antithesis-research`, then gives the recommended order with a diagram, starter prompts for each skill and the list of files created. It continues with platform and agent requirements, model choice, a table of tools per skill and a step-by-step install wizard. The repository also holds each skill's reference files, a beginner guide, a CHANGELOG and contributor instructions.
README badge
Original SKILL.md
--- name: antithesis-launch description: > Launch an Antithesis run with snouty by discovering the harness layout, building the right Docker Compose config, running `snouty validate`, bailing on validation failure, and then submitting `snouty launch` with sane metadata. Use when the user wants to send, submit, or launch an Antithesis run. This skill takes duration in minutes as input. compatibility: Requires Docker Compose v2, a container engine (docker or podman), and snouty (https://github.com/antithesishq/snouty). metadata: version: "2026-10-15 84b188a" --- # Antithesis Launch **Skill version:** `2026-10-15 84b188a` ## Prerequisites - DO NOT PROCEED if `snouty` is not installed. See `https://raw.githubusercontent.com/antithesishq/snouty/refs/heads/main/README.md` for installation options. - DO NOT PROCEED if `snouty` is older than 0.8.0. Use `snouty --version` to find the version. Use `snouty update` to update. - Run `snouty doctor`. It checks the container runtime and the compose CLI. Report both to the user and build with that same pair. This skill writes `docker compose`; use `docker-compose` instead when that is the CLI snouty names. ## Goal Launch an Antithesis run in this order only: 1. `docker compose build` (skip if directed to use pre-built images) 2. `snouty validate` 3. if validation fails, stop and report the error 4. `snouty launch` ## Required Input - `duration` in minutes is required. If the user did not provide it, ask before submitting the run. ## Discovery - Start from any user-provided path, command, or Antithesis directory name. - Otherwise, inspect the repo to understand how the harness is wired. Check nearby `AGENTS.md`, `README*`, `Makefile*`, and Antithesis-specific scripts before choosing commands. - Find the config directory by locating the `docker-compose.yaml` intended for Antithesis. Prefer directories like `antithesis/config`, but support non-standard layouts. - Treat these as strong Antithesis signals: nearby `scratchbook/` or `test/` directories, compose content mentioning `/opt/antithesis`, `ANTITHESIS_` env vars, `setup_complete`, or existing `snouty` examples. - If multiple compose files look plausible, prefer the one referenced by repo docs or existing `snouty launch` examples. If the choice is still ambiguous, ask the user instead of guessing. - Use the directory containing `docker-compose.yaml` as the `snouty validate <CONFIG>` and `snouty launch --config <CONFIG>` argument. - Build against that exact file: `docker compose -f <CONFIG>/docker-compose.yaml build`.