proposal-write
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from user idea notes and workspace guidelines.md files to generate proposal content. This creates a surface where malicious instructions in the input could influence the agent's behavior.\n
- Ingestion points: User-provided idea notes and workspace-specific guidelines.md override files as described in SKILL.md.\n
- Boundary markers: The skill does not define explicit delimiters or instructions to ignore embedded directives in the ingested content.\n
- Capability inventory: The agent has the ability to write to the local filesystem and execute a provided Python validation script (scripts/check.py).\n
- Sanitization: The skill provides instructions for safe 'slug' generation (lowercase, hyphenated) to mitigate filename-based injection, but does not specify content-level sanitization for the notes or descriptions.\n- [COMMAND_EXECUTION]: The workflow involves the agent executing a local Python script with a filename argument derived from user-influenced title content. This pattern introduces a potential risk of command injection if the agent or execution environment does not properly escape the filename argument during shell invocation.
Audit Metadata