bmad-project-context

Warn

Audited by Gen Agent Trust Hub on Nov 13, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DYNAMIC_EXECUTION]: The skill executes Python scripts located within the target repository's directory (specifically {project-root}/_bmad/scripts/resolve_customization.py and resolve_config.py) using the uv run command. This pattern involves executing code found in the project being audited, which could lead to unauthorized actions if the repository is malicious.
  • [COMMAND_EXECUTION]: The skill configuration in customize.toml supports workflow hooks such as activation_steps_prepend, activation_steps_append, and on_complete which allow for the execution of arbitrary shell commands.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests data from various repository files and configuration manifests to generate agent instructions, creating a surface for indirect prompt injection.
  • Ingestion points: AGENTS.md, package.json, Makefile, pyproject.toml, README, and CI configuration files are read from the target project.
  • Boundary markers: The skill uses <!-- bmad:context --> and <!-- /bmad:context --> markers to delimit the managed block in AGENTS.md.
  • Capability inventory: The skill possesses file-writing capabilities and the ability to execute project-local Python scripts and arbitrary shell commands.
  • Sanitization: Risk is mitigated by a mandatory human-in-the-loop approval process where the user must review a "settled ledger" and the complete instruction block before any file writes are performed.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 13, 2026, 10:20 PM