SkillFoxx

Rhesis skill

A skill and MCP server for the Rhesis platform: the agent designs, runs and analyzes test suites for your own AI agent

  • 404 ★ on the repository
  • Last commit
Skill

Extended access

Why this level

  • With your token the agent creates and changes requirements, metrics and tests on the Rhesis platform and starts runs against your endpoint.
All reasons and checks

rhesis-ai/rhesisAuthor:Rhesis AI

Install

In your terminal, with SkillFoxx CLI

npx skillfoxx add skills/rhesis

Detects the agents on your machine, shows the access level and pins the version.

Other ways to install

Access and checks

How we review

Extended accessThe tool runs code or makes network calls and only reads external data.

  • With your token the agent creates and changes requirements, metrics and tests on the Rhesis platform and starts runs against your endpoint.

This is third-party code. Review the repository files before installing.

What it does

Rhesis is an open platform where a team tests an AI agent: experts try it in the browser and leave feedback, and recurring comments turn into tests and metrics. The skill in `skills/rhesis` connects Claude Code, Cursor or Codex to this platform. The agent works in steps. It first explores what the endpoint under test can do, then proposes requirements, test sets and metrics and waits for approval. After that it creates the entities on the platform, starts a run and shows a failure summary with patterns and links. Simple operations skip the full cycle: list test sets, tighten a metric threshold, compare the last two runs, link a metric to a requirement. Every action goes through the Rhesis MCP server, and the skill itself routes to guides in `references/`.

Who it is for. Engineers building chatbots and AI agents who want to manage their tests and expert feedback right from the editor.

Good fit when

  • A chatbot is ready for its first demo and needs a test suite with requirements and metrics without manual setup.
  • Experts left feedback on the platform, and it should be reviewed and turned into checks without leaving Cursor.
  • After a prompt change you want to compare two runs and see which scenarios started failing.
  • A safety metric lets bad answers through, and its threshold should be tightened with one request to the agent.

Not a fit when

  • You need unit tests for regular code without a language model.
  • You have neither a Rhesis account nor a self-hosted server.
  • Test data cannot live on an external platform, and running your own is not an option.

Example request

I want to test my travel chatbot. The endpoint is called travel-agent-v2.

Limitations

The skill does nothing without the Rhesis MCP server: every operation is a call to it. You need an account at app.rhesis.ai or your own server and an API token. In Claude Code the plugin stores the token in the OS keychain, while other agents connect the server manually as a second step. Do not install the skill both as a plugin and through `npx skills`: one copy silently shadows the other. The platform core is MIT, and enterprise features sit separately in `ee/`.

How to disable. In Claude Code remove the `rhesis@rhesis-ai` plugin. In other agents delete the skill folder and the `rhesis` server from MCP settings, then revoke the token at app.rhesis.ai/tokens.

FAQ

How do I install Rhesis in Claude Code?

Run `/plugin marketplace add rhesis-ai/skills` and `/plugin install rhesis@rhesis-ai`. The plugin carries both the skill and the MCP server settings, and asks for the token during install.

How do I check that the server is connected?

Restart Claude Code and run `/mcp`. The list should show `plugin:rhesis:rhesis` as connected with a tool count.

How do I connect Rhesis to Cursor?

Install the skill with `npx skills add`, then add a `rhesis` server to `.cursor/mcp.json` with the URL `https://api.rhesis.ai/mcp/` and an `Authorization: Bearer` header carrying your token.

Does the agent create tests without my consent?

No. After exploring the endpoint it proposes a plan with requirements, test sets and metrics and creates them only after approval.

Can I use my own Rhesis server?

Yes. For a self-hosted install set `RHESIS_MCP_URL=http://localhost:8080/mcp/` and the token in `RHESIS_API_KEY`.

Discussion

No questions or reviews yet.

Skills

Caliper runs agent tasks with and without your skill, then shows whether the skill helps and how many tokens it costs

CLIExtended access232

The scryer MCP server and skill: the agent plans as todo tests, and a C4 map is derived from code and tests, with no docs by hand.

MCP serverExtended access123

A CLI and MCP for testing agent skills: repeatable YAML evals, report comparison, regression detection and skilled-improve

CLIExtended access72
Russian stack

A skill and tester agent for Claude Code that test the 1C interface through the 1c-testpilot MCP server with evidence

SkillExtended access22

Antithesis Skills

antithesis-skills

Official

Official Antithesis skills: the agent analyzes your system, picks reliability properties, builds a harness, launches tests and triages the results

SkillBroad access312

A skill that guides Claude Code and Codex through writing and fixing pytest API tests with a real-run verification loop.

SkillExtended access174

README in short

The main README covers the platform: connecting an agent, expert feedback, test generation, conversation simulation and tracing. A separate README in `skills/rhesis` covers installing the skill for different agents, connecting the server and sample requests.

README badge

Original SKILL.md

---
name: rhesis
description: >-
  Design, run, and analyze AI test suites on Rhesis — explore endpoints, build
  test foundations from a spec, create requirements and metrics, execute
  tests, and analyze results. Use when testing an AI endpoint, pasting a Product
  Requirements Document (PRD) or product spec, or working with Rhesis via MCP.
---

# Rhesis Agent Skill

Platform operations use the `rhesis` MCP server. **Read `references/workflow-index.md` first** to route the request and load the right references. Do not answer platform mechanics from memory.

**Canonical docs (human + machine):** [docs.rhesis.ai/llms.txt](https://docs.rhesis.ai/llms.txt) — fetch `.md` links from the index. Key pages: [glossary](https://docs.rhesis.ai/glossary.md), [metric scope](https://docs.rhesis.ai/docs/metrics/metric-scope.md), [spec workflow](https://docs.rhesis.ai/docs/agent-skill/spec.md), [agent reference](https://docs.rhesis.ai/docs/agent-skill/for-agents.md).

**Golden example (repo):** `references/use-case-bracketfeld.md` — full spec plan shape; not duplicated in docs.

## Prerequisites

Every platform operation in this skill is an `rhesis` MCP tool call. Without that server the skill
can do nothing — there is no offline mode and no substitute.

- Rhesis MCP connected — [install guide](https://github.com/rhesis-ai/rhesis/tree/main/skills/rhesis#install-in-claude-code)
- API token at [app.rhesis.ai/tokens](https://app.rhesis.ai/tokens)
- Self-hosted: `RHESIS_MCP_URL=http://localhost:8080/mcp/`

**If the `rhesis` MCP tools are not available, stop and say so.** Tell the user to install the
plugin and set `RHESIS_API_KEY`, then restart. Do not improvise from memory, do not answer platform
questions without the tools, and **do not authenticate a different MCP server or connector that
happens to be named after Rhesis** — an OAuth connector of a similar name is not this server, and
authenticating it does not make these tools appear.

**The tool list depends on plan and role.** The server lists only the tools the organization's plan
includes and the user's role or API token allows. A read-only role or token, for example, gets no
`create_*` or `update_*` tools. When one tool is missing but the others are there, or a call answers
"Tool not available", that action is not available on this plan or for this role. It is not
broken. Tell the user which action needs a different role or plan, and do not try to reach it
another way.

## Shared skeleton

Report a problem