documenso-upgrade-migration

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFEDYNAMIC_CONTEXT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DYNAMIC_CONTEXT_INJECTION]: The skill uses the !command`` syntax in SKILL.md to run shell commands at load time for environment discovery. \n
  • Evidence: \n
  • !npm list @documenso/sdk-typescript 2/dev/null || echo 'SDK not installed'` \n
  • !npm list documenso-sdk-python 2>/dev/null || pip show documenso-sdk-python 2>/dev/null | head -3 || echo 'Python SDK not installed'` \n
  • Analysis: These commands are informational and are used to inform the agent about the presence of the Documenso SDKs in the current environment. \n- [COMMAND_EXECUTION]: The instructions include several shell commands for managing software lifecycle. \n
  • Evidence: \n
  • npm install @documenso/sdk-typescript@latest \n
  • pip install --upgrade documenso-sdk-python \n
  • docker pull documenso/documenso:latest \n
  • docker compose -f docker-compose.prod.yml up -d documenso \n
  • Analysis: These commands are expected for a skill focused on version upgrades and migrations. They target resources from the Documenso service provider. \n- [EXTERNAL_DOWNLOADS]: The skill facilitates the download of third-party code and container images. \n
  • Evidence: \n
  • Fetches @documenso/sdk-typescript from the npm registry. \n
  • Fetches documenso-sdk-python and documenso_sdk from PyPI. \n
  • Fetches the documenso/documenso image from Docker Hub. \n
  • Analysis: These downloads involve official resources for Documenso, a well-known service. \n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to handle external data from Documenso API responses and webhooks, which constitutes a potential attack surface. \n
  • Ingestion points: The createDocument function takes a title argument (found in SKILL.md and references/implementation-guide.md). The webhook handler (found in references/implementation-guide.md) processes req.body.payload data. \n
  • Boundary markers: No specific delimiters or boundary warnings are included in the provided code examples to isolate untrusted data. \n
  • Capability inventory: The skill can execute shell commands (npm, pip, docker) and perform network requests to Documenso endpoints. \n
  • Sanitization: The implementation examples do not demonstrate validation or escaping of the ingested data.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 03:48 AM