The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external sources such as a task backlog and a work ledger. If these sources contain malicious instructions, the agent could potentially be manipulated.
Ingestion points: Data is fetched via the backlog_view tool (referenced in dispatch_input) and the sam_cli plan read command output (referenced in work_ledger).
Boundary markers: The instructions do not specify delimiters or safety warnings for handling content retrieved from these external sources.
Capability inventory: The skill utilizes the sam_cli tool to perform stateful operations, including updating task sections and finalizing task status.
Sanitization: There is no mention of sanitizing or validating the retrieved content before it is processed or used in subsequent commands.
[COMMAND_EXECUTION]: The skill explicitly directs the agent to interact with the system using shell commands via the sam_cli utility to manage task state, including reading, updating, and finishing tasks.

subagent-contract