generated: '2026-09-16'
method: searched
source: https://github.com/probabl-ai/skore-cli
name: skore-cli
binary: skore
license: MIT
version: 0.4.0
published: '2026-09-11'
install:
- method: pip
  command: pip install skore-cli
auth:
  interactive: OAuth device login on first `skore agent` run
  env:
  - SKORE_HUB_API_KEY
  - SKORE_HUB_URI
  flag: --hub-url
command_groups:
- group: skills
  purpose: discover, install and manage Agent Skills from probabl-ai/skills
  commands:
  - skore skills list
  - skore skills install [ids|all] [--global] [--agent <name>] [--repo owner/name]
  - skore skills update
  - skore skills remove
  agents:
  - agents
  - claude-code
  - cursor
  - codex
  - gemini
  - windsurf
  - cline
  - roo
  - amp
  - github-copilot
  - bob
  - bob-ide
- group: agent
  purpose: log in, pick a workspace + harness, mint a workspace API key, write harness config (MCP url {hub}/mcp
    with X-API-Key) and launch the agent
  commands:
  - skore agent
  - skore agent --harness claude
  - skore agent --workspace ./myapp
  harnesses:
  - Bob Shell
  - Bob IDE
  - Claude
  - Cursor
  - OpenCode
  - Pi
  - GitHub Copilot
  - Codex
- group: sync
  purpose: synchronize report projects across local storage, Skore Hub and MLflow
  commands:
  - skore sync <project> --to=hub --to-workspace=<ws> [--both|--dry-run]
  - skore sync <project> --from=hub --from-workspace=<ws> --to-project=<name>
  - skore sync <project> --to=mlflow --tracking-uri=<uri>
key_flows:
- Connect a coding agent to Skore Hub (skore agent)
- Install data-science skills into an agent (skore skills install all)
- Push local experiment reports to the hub (skore sync --to=hub); --dry-run available