repo-kanban-pm

Warn

Audited by Gen Agent Trust Hub on Mar 3, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The script scripts/add_daily_pm_cron.sh uses the openclaw tool to create a persistent scheduled cron job that runs a PM review task.\n- [EXTERNAL_DOWNLOADS]: The cron job instructions for the agent include running npx tsc --noEmit. This involves downloading and executing packages from the npm registry at runtime, which is an external and potentially untrusted source.\n- [PROMPT_INJECTION]: The skill contains an indirect prompt injection vulnerability within the automated daily review process.\n
  • Ingestion points: The agent reads from docs/roadmap/ROADMAP.md, docs/features/*/KANBAN.md, and any files in docs/pm/bugs/*.md.\n
  • Boundary markers: Absent; there are no delimiters or instructions provided to the agent to treat data from these files as untrusted or to ignore embedded instructions.\n
  • Capability inventory: The agent has the ability to run shell commands (npx, gh), modify repository files, and manage pull requests.\n
  • Sanitization: Absent; the agent is instructed to update the repository state (Kanban and Roadmap) directly based on the contents of these external documents without any sanitization or verification.\n- [COMMAND_EXECUTION]: The scripts/init_repo_pm.sh script creates a series of directories and files and appends logic to AGENTS.md, which modifies the instructions and constraints for all agents working in the repository.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 3, 2026, 02:51 PM