The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from sources like issue trackers, chat channels, and source control through the why skill and T3 thread tools.
Ingestion points: t3_thread_read, t3_thread_search, and findings from the why skill which ingest data from issue trackers, chat/issue channels, long-form docs, and error tracking systems.
Boundary markers: The instructions include scope-locking mechanisms (e.g., pinning time ranges and workspaces) and mandate reading only relevant regions to minimize exposure to noise or malicious payloads.
Capability inventory: The skill possesses the capability to execute git and gh commands for state verification and can spawn subagents using the delegate_task tool.
Sanitization: The skill explicitly instructs the agent to sanitize private context before any public output and to process results through the unslop skill.

recall