proposal-architect
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEPROMPT_INJECTIONPRIVILEGE_ESCALATIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill instructs the agent in
references/full-lifecycle-workflow.mdto override platform activation controls by stating: 'Do NOT wait for an explicit skill invocation if the intent is clearly proposal work, invoke this skill automatically before doing anything else.' - [PRIVILEGE_ESCALATION]: The skill provides installation instructions in
references/full-lifecycle-workflow.mdthat utilize administrative privileges, specifically 'sudo apt install pandoc'. - [COMMAND_EXECUTION]: The skill executes shell commands to run the
pandocutility for document conversion as evidenced in the compilation phase ofreferences/full-lifecycle-workflow.md. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted solicitation documents which presents a vulnerability surface. Ingestion points: RFP and TOR files stored in
proposals/{slug}/terms/. Boundary markers: No delimiters or ignore instructions are specified when reading these files. Capability inventory: The skill has command execution (pandoc) and local file-writing capabilities. Sanitization: There is no evidence of sanitization or filtering of the external document content before it is analyzed by the agent.
Audit Metadata