skillos-cli

Runs the interactive first-time setup wizard for supported AI clients and optional local tools.

$ skillos init
  # Follow the prompts to configure this machine's AI client and optional local tools.

Prints the CLI command guide or examples for a named command. Use --help for Commander-generated option details.

$ skillos help connectors
  # Prints the connector command examples supported by this CLI release.

Aliases: skillos connect

Auto-detects supported AI clients (Claude Desktop, Antigravity/Gemini CLI, Cursor, Windsurf, Zed, Continue, Cline, Warp, Amp, Goose) on the current machine and writes the MCP server config to each client's config file. Existing files receive timestamped backups; malformed or incompatible JSON is left unchanged.

$ skillos install --dry-run
  # Lists only client config paths detected on this machine.
  # A dry run does not create or modify files.

Fetches the live registry from founderskillos.xyz/api/skills and prints all available skills.

$ skillos list --tier 2
  # Prints the Tier 2 skills currently returned by the live registry.
  # Results depend on the catalog and can change over time.

Fetches the full SDS manifest for a skill by slug. Displays tier, vertical, tools, prompt template, and all 13 SDS component statuses.

$ skillos inspect <skill-slug>
  # Prints the manifest and component status returned by the registry.
  # Use a slug shown by skillos list; unavailable slugs return an error.

Checks YAML or Markdown syntax, required metadata, each of the 13 canonical SDS components, declared-tool tier rules, and configured risk signatures. This static validation does not run model-behavior tests, prove safety, or approve publication.

$ skillos validate my-skill.sds.yml
  # Results are computed from this file and vary by its contents.
  # Add --json to emit a machine-readable validation report.

Validates a local DCS v1 JSON or YAML document. Validation does not issue a certificate or prove that a deployment is safe or operating as declared.

$ skillos validate-deployment deployment.json
  # Validates the supplied local document and reports errors and warnings.

Fetches and prints a skill's SDS prompt through the configured MCP client. Local mode uses stdio; remote entitlement checks run only when PLATFORM_API_URL is configured.

$ skillos activate <skill-name>
  # Retrieves the selected skill through MCP; availability and entitlement checks apply.

Starts the local MCP runtime using stdio by default. HTTP mode with --port requires the runtime source or built HTTP server from a SkillOS repository checkout; the published CLI package does not include that server.

$ skillos serve
  # Starts the local stdio runtime. Tool availability depends on its configuration.

Aliases: skillos test

Lists detected supported AI-client installations and checks the local stdio MCP handshake. It does not inspect the remote platform or report tool response times.

$ skillos doctor
  # Reports detected client config paths and the local stdio handshake result.
  # A failed handshake exits with a nonzero status.

Reads the live connector catalog. `connect` prints the signed-in dashboard setup link; it does not launch OAuth or save credentials. `test` and `disconnect` call owner-scoped platform APIs using SKILLOS_API_KEY.

$ skillos connectors list
  # Prints entries from the current platform catalog.
  # Account state is unknown unless SKILLOS_API_KEY is set.

$ skillos connectors connect <slug>
  # Prints a dashboard URL. Complete credential setup in the signed-in web app.

Prints the global command list or command-specific options. `skillos help [command]` also prints examples.

$ skillos --help
  # Lists commands and global options. Run `skillos serve --help` for transport-specific options.

Prints the installed CLI version.

$ skillos --version
  # Prints the version bundled in the installed CLI package.