The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute several shell commands for environment setup, file management, and validation. These include using mkdir and rsync to initialize documentation structures from local templates, and grep to audit document language for non-blameless terms. It also executes local Python scripts (scripts/build_status.py and scripts/build_docs.py) to perform consistency checks on the generated documentation.- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process potentially untrusted external data, such as chat transcripts, incident logs, and alert histories, to generate postmortem summaries. This creates an attack surface where malicious content embedded in logs could attempt to influence the agent's output. However, the skill provides a structured framework for extraction, which mitigates the risk of arbitrary instruction following.
Ingestion points: incidents/incidents.json, incident logs, alert history, chat transcripts.
Boundary markers: None explicitly defined in the prompt templates.
Capability inventory: File system write access (docs/sre/), local script execution (python3 scripts/).
Sanitization: The instructions focus on factual extraction but do not explicitly specify sanitization or escaping for the ingested content.

postmortem