Autousers
cloud_upload

Install the Skill in Claude.ai or the Claude API

Upload the Autousers Skill to Claude.ai or use it via the Claude API skills endpoint.

Pick your surface

  • Claude.ai (chat) — upload as a zip in Settings, available on Pro, Max, Team, and Enterprise plans with code execution enabled
  • Claude API (programmatic) — upload via the Skills API and reference by skill_id in your Messages requests

Get the Skill zip

The Skill is published as a release asset on the @autousers/mcp GitHub repository. Download autousers-skill.zip from the latest release page.

infoIf you have the npm package installed locally you can also build the zip yourself: cd node_modules/@autousers/mcp/skill && zip -r autousers-skill.zip autousers — both produce the same archive.

Install on Claude.ai

  1. Open Claude.ai and click your profile → Settings
  2. Go to Features → Skills
  3. Upload autousers-skill.zip
  4. Make sure code execution is enabled in your workspace settings (the Skill itself does not run code, but Claude.ai requires it for the Skills feature)
  5. Connect the @autousers/mcp MCP server via Connectors if you have not already — the Skill expects the MCP tools to be present
  6. Start a new conversation and try a prompt that mentions Autousers
warningCustom Skills on Claude.ai are per-user — each team member uploads their own zip. There is no central admin distribution surface today.

Install via the Claude API

Upload the Skill once with the Skills API, then reference its skill_id in any Messages request. The integration shape is identical to Anthropic's pre-built Skills.

curl -X POST "https://api.anthropic.com/v1/skills" \
  -H "x-api-key: $ANTHROPIC_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "anthropic-beta: skills-2025-10-02" \
  -F "display_title=Autousers" \
  -F "files[]=@autousers-skill.zip;filename=skill.zip"

Upload — returns a skill_id like skill_01...

import anthropic

client = anthropic.Anthropic()

response = client.beta.messages.create(
    model="claude-opus-4-7",
    max_tokens=4096,
    betas=["code-execution-2025-08-25", "skills-2025-10-02"],
    container={
        "skills": [
            {"type": "custom", "skill_id": "skill_01...", "version": "latest"}
        ]
    },
    messages=[{"role": "user", "content": "Summarise my latest Autousers eval"}],
    tools=[{"type": "code_execution_20250825", "name": "code_execution"}],
)

Reference the Skill in a Messages request

warningSkills via the Claude API are not eligible for Zero Data Retention. Definitions and execution data are kept under Anthropic's standard retention policy.

MCP connectivity from the Claude API

The Skill's workflows assume the @autousers/mcp tools are reachable. For the Claude API surface, that means your application code is responsible for either (a) running an MCP client that proxies to https://mcp.autousers.ai/mcp, or (b) translating the Skill's tool-call instructions into direct REST calls against the Autousers API. Most production setups use (a).

Update the Skill

Upload a new version with the Create Skill Version endpoint and either pin to that specific version in your container.skills entry or use 'latest' to always get the newest. New versions break prompt caching for that container, so keep the version pinned in production unless you intend to invalidate.

Was this article helpful?