tensorpm-agentic-pm
Pass
Audited by Gen Agent Trust Hub on Mar 13, 2026
Risk Level: SAFEREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: Installation instructions in README.md and SKILL.md suggest using piped shell execution (curl | bash and irm | iex) to download and run setup scripts from the author's repository at github.com/Neo552. These are standard deployment methods provided by the vendor for the TensorPM application.
- [EXTERNAL_DOWNLOADS]: The skill documentation references external binaries and installer scripts hosted on GitHub Releases and the vendor's domain for Windows, macOS, and Linux setup.
- [PROMPT_INJECTION]: The skill handles project initialization via strings or files, creating an ingestion surface for indirect prompt injection. 1. Ingestion points: create_project (documentPath, prompt), message/send (A2A messages). 2. Boundary markers: None identified in skill metadata. 3. Capability inventory: MCP CRUD tools for projects and action items. 4. Sanitization: Not specified in documentation; reliant on the host application.
- [COMMAND_EXECUTION]: The skill facilitates interaction with the local host to communicate with the TensorPM desktop app and performs file system access to process project briefs.
Audit Metadata