flow-parallel
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEDYNAMIC_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill dynamically generates shell scripts (launch.sh) and markdown instructions (instructions.md) based on task decomposition, which are subsequently executed as background processes.\n- [COMMAND_EXECUTION]: The skill invokes several local scripts expected to be present in the user's home directory (e.g., ~/.claude-octopus/plugin/scripts/agent-registry.sh). These are part of the vendor's required infrastructure for the skill's operation.\n- [INDIRECT_PROMPT_INJECTION]: The skill takes user-provided task descriptions and interpolates them directly into instruction files for sub-agents. \n
- Ingestion points: User input from AskUserQuestion (compound task description) in SKILL.md.\n
- Boundary markers: No explicit delimiters or instructions are used to separate user-provided task content from system instructions in the generated instructions.md.\n
- Capability inventory: The skill has the capability to write files, execute shell commands, and spawn new AI agent processes (claude -p).\n
- Sanitization: There is no evidence of sanitization or escaping of the user-provided task string before interpolation.
Audit Metadata