executing-plans

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • Indirect Prompt Injection (LOW): The skill relies on reading task details from an external issue tracker via the bd show <id> command and instructs the agent to follow the steps exactly. This creates a surface for indirect prompt injection where a malicious actor could embed instructions in an issue description to manipulate the agent's behavior.\n
  • Ingestion points: Ingestion occurs in SKILL.md (Step 1) and references/batch-execution-detail.md (Step 2) when the agent reads issue details.\n
  • Boundary markers: Absent. The skill does not provide delimiters or instructions to treat issue content as untrusted data.\n
  • Capability inventory: The agent can perform file system modifications, execute local shell commands (git, bd), and invoke other development skills.\n
  • Sanitization: Absent. There is no sanitization or validation of the text retrieved from the issue tracker.\n- Command Execution (SAFE): The skill utilizes local development tools including git and the bd CLI. These operations are limited to project state management and local file operations and do not represent a high security risk in this context.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:44 PM