The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute a local Node.js script located at ${CLAUDE_PLUGIN_ROOT}/src/tasks.js. These executions are used to manage task lifecycle and state tracking within a project context. The script is a vendor-owned resource used for the skill's primary functionality.\n- [INDIRECT_PROMPT_INJECTION]: The skill defines a surface for processing untrusted data by ingesting task names and JSON state patches which are re-injected into the agent's context during session compaction.\n
Ingestion points: Task names in the add command and JSON objects in the patch command defined in SKILL.md.\n
Boundary markers: The instructions specify the use of quotes for shell arguments, which provides basic separation between the command and the data.\n
Capability inventory: The skill is limited to running a specific local script via Bash; it does not request network access or sensitive file system permissions.\n
Sanitization: No explicit validation or sanitization of the JSON payload is described, relying on the platform's tool-calling logic to prevent command injection.

cco-task