task-management-protocol

Pass

Audited by Gen Agent Trust Hub on Apr 30, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes task metadata and descriptions which may contain instructions from untrusted sources (Indirect Prompt Injection surface).
  • Ingestion points: The skill instructions in SKILL.md (Phase 1 and 2) command the agent to read task details using TaskGet and TaskList.
  • Boundary markers: There are no instructions to use delimiters or ignore embedded directives within the task descriptions or metadata being read.
  • Capability inventory: The agent is granted capabilities to Write and Edit files (specifically in .claude/context/plans/) and use TaskUpdate to modify system state based on the ingested content.
  • Sanitization: The protocol lacks instructions for sanitizing or validating external task content before it is processed or reflected in subsequent updates.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 30, 2026, 10:36 AM