plannotator-inbox

Pass

Audited by Gen Agent Trust Hub on Nov 6, 2026

Risk Level: SAFEDYNAMIC_CONTEXT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DYNAMIC_CONTEXT_INJECTION]: The skill uses the dynamic context injection pattern !\plannotator inbox --open`` in SKILL.md. This command is executed when the skill is loaded to open the tool's interface in the user's browser. This is a legitimate use of project-specific tooling and does not involve user-controlled input or data exfiltration.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes and reports the output of a local command to the user, creating a potential surface for indirect prompt injection if the tool's output could be manipulated.
  • Ingestion points: The output of the plannotator inbox --open command is used directly as context for the agent's response.
  • Boundary markers: None identified for the command output.
  • Capability inventory: The skill uses the plannotator_inbox tool and the plannotator inbox mcp server to send and read messages.
  • Sanitization: The skill instructions tell the agent to tell the user the address or the error message, implying a pass-through of information.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 6, 2026, 06:19 AM