platform-friction-detector
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill operates as a static analysis and monitoring tool. It uses Sourcegraph for code searches and the GitHub API to fetch pull request context, which are legitimate operations for a development-focused agent.
- [DATA_EXPOSURE_EXFILTRATION]: While the skill accesses potentially sensitive areas such as Slack channels and internal code repositories, this is within its stated purpose of detecting user friction signals. The data gathered is used to generate a report for the platform team and does not involve exfiltration to untrusted external domains.
- [COMMAND_EXECUTION]: The skill utilizes
gh apianddiff_searchcapabilities. These are used for data gathering (fetching PR details and code changes) rather than executing arbitrary system commands. No privilege escalation or persistence patterns were found. - [INDIRECT_PROMPT_INJECTION]: The skill processes external data (PR bodies, commit messages, Slack messages). It identifies this as a surface for classification. While it infers root causes from this untrusted text, the impact is limited to the generated report summary and does not drive high-privilege automated actions. The design instructions include classification and grouping logic to handle this data safely.
Audit Metadata