custom-war-room
Pass
Audited by Gen Agent Trust Hub on Feb 25, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The README suggests installation via 'npx skills add ericrisco/war-room-builder', which fetches code from the author's repository. This is a vendor-owned resource.- [COMMAND_EXECUTION]: The skill generates a local folder structure containing a secondary SKILL.md file. This generated file includes its own YAML frontmatter and debate instructions, creating a standalone derivative skill that can be installed or shared independently.- [COMMAND_EXECUTION]: The PRD documentation references a Python script (scripts/create_war_room.py) for scaffolding the persistent war room directory structure. While the script content is not provided, its use is part of the core functionality for file and directory management.- [PROMPT_INJECTION]: In 'INVOKE' mode, the skill loads external expert profiles and configuration data to initialize the simulation.
- Ingestion points: Local files (expert-N.md and war-room.config.json) read from user-specified directories.
- Boundary markers: Absent; the agent loads profile content directly into its persona context without delimiters.
- Capability inventory: File system read/write operations and the ability to execute complex role-play simulations.
- Sanitization: No sanitization or schema validation is performed on the profile content before it is interpolated into the agent's instructions.
Audit Metadata