documenso-debug-bundle
Pass
Audited by Gen Agent Trust Hub on Sep 9, 2026
Risk Level: SAFEDYNAMIC_CONTEXT_INJECTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_CONTEXT_INJECTION]: The SKILL.md file contains shell commands using the '!' syntax (
node --version,python3 --version,uname -a) that execute automatically when the skill is loaded to provide environment context. - [DATA_EXFILTRATION]: Diagnostic scripts included in the implementation guide, such as
logged-client.tsandtest-webhook.ts, log API request arguments and full webhook payloads to the console. This could lead to the exposure of PII or sensitive document metadata if these logs are shared for support without proper redaction. - [INDIRECT_PROMPT_INJECTION]: The skill ingests and displays document titles and recipient information from external API responses. Maliciously crafted metadata within these documents could attempt to influence the agent's behavior during a debug session.
Audit Metadata