recoup-internal-consulting-linkedin-publisher
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill instructions define a standard business workflow for LinkedIn publishing. It includes specific defensive instructions, such as 'Credentials belong in the user's connector or selected workspace environment, never in this skill or a chat message,' which directly mitigates credential exposure risks.
- [SAFE]: While the skill interacts with a local script (
integrations/linkedin/_work/publish.py), it instructs the agent to inspect the tool's help and account mappings rather than executing arbitrary or hidden code. This is a standard extension pattern for workspace-specific adapters. - [SAFE]: The skill includes instructions to 'verify claims' and 'resolve [ambiguous profiles]' before creating external posts, which serves as a manual validation step to prevent accidental or malicious data publishing.
- [SAFE]: The skill explicitly restricts the agent's scope to the current workspace ('Never search another private checkout for missing inputs'), preventing potential data leakage across different projects or environments.
Audit Metadata