project-management-specialist
Warn
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- COMMAND_EXECUTION (MEDIUM): The
project-analyst,team-configurator, andtech-lead-orchestratorsub-agents are granted access to theBashtool. These agents use Bash to scan directory structures and sample code files. This presents a risk of command injection if the agent is tricked into executing commands derived from maliciously crafted filenames or file contents within an analyzed project. - PROMPT_INJECTION (LOW): The skill exhibits a high surface area for Indirect Prompt Injection (Category 8) due to its core purpose of analyzing untrusted external data.
- Ingestion points:
project-analyst.md(package manifests, source code sampling),team-configurator.md(dependency files, build configurations),tech-lead-orchestrator.md(project requirements and codebase analysis). - Boundary markers: Absent. The instructions do not specify the use of delimiters or warnings to the agent to disregard instructions found within the files it reads.
- Capability inventory:
Bash(command execution),WriteFile(filesystem modification),WebSearch(network access). - Sanitization: Absent. Data read from the environment is processed directly to generate reports and configuration updates (e.g., modifying
CLAUDE.md), providing a path for malicious instructions in the project to influence agent behavior.
Audit Metadata