agent-friction-ledger
Pass
Audited by Gen Agent Trust Hub on Aug 6, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is purely instructional and contains no executable scripts (Python, JavaScript, or Shell). All analysis is based on the Markdown guidelines provided.
- [DATA_EXFILTRATION]: The skill explicitly forbids external data transmission. Instructions state: 'Never POST to a viewer, telemetry endpoint, or external API' and 'Never sends session data outside the workspace.'
- [PROMPT_INJECTION]: No prompt injection patterns, bypass instructions, or 'DAN' style overrides were detected. The instructions focus on restrictive scoping and opt-in triggers.
- [CREDENTIALS_UNSAFE]: The skill includes explicit instructions to 'Redact or omit sensitive content before writing' and 'Never record... private URLs, credentials, personal data.'
- [EXTERNAL_DOWNLOADS]: No external URLs or remote dependencies are referenced for download or execution.
- [COMMAND_EXECUTION]: The skill does not perform any system commands. It instructs the agent to write local Markdown files to specific paths (e.g.,
.git/agent-friction/) which is a standard file-writing operation within the agent's workspace.
Audit Metadata