Sorry

Sorry integration. Manage data, records, and automate workflows. Use when the user wants to interact with Sorry data.

This skill is coherent: it uses the Membrane CLI to talk to a 'sorry' connector and requests no extra credentials. Before installing or running commands: 1) verify the @membranehq npm package and the project homepage/repository (getmembrane.com / github.com/membranedev) to ensure you trust the CLI source; 2) prefer using npx to avoid a global npm install if you want less permanent change to your system; 3) review Membrane's privacy/auth documentation to understand where credentials/tokens are stored (local auth state or Membrane server-side) and whether that fits your security posture; 4) confirm the connectorKey 'sorry' corresponds to the service you expect — the SKILL.md contains a stray sentence about 'Official docs' that looks like a template error, so if uncertain, double-check the Membrane action list for the connector's capabilities before using it. If you need a deeper assurance, provide the Membrane repo/package links and I can point out specific files/commands to audit.

Type: OpenClaw Skill Name: sorry Version: 1.0.1 The skill bundle appears to be a low-quality or AI-generated placeholder for a non-existent service, as evidenced by the inclusion of a literal LLM refusal message in SKILL.md ('I am sorry, but I cannot provide an API... for an app called "Sorry"'). It instructs the agent to perform high-privilege actions, including a global installation of the Membrane CLI (npm install -g @membranehq/cli) and authentication steps. The contradictory documentation and the requirement for global system changes for a service that the documentation itself claims is not a software application make this bundle highly suspicious.