The Agent Skills Directory

[COMMAND_EXECUTION]: The skill makes extensive use of the Bash tool to execute pylot CLI commands. These commands control infrastructure deployment, mission dispatching, and interaction with remote Fargate workers.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by ingesting untrusted data from external sources.
Ingestion points: The agent is instructed to read content from remote environments using pylot workers output, pylot workers logs, and pylot workers view (SKILL.md).
Boundary markers: The instructions do not define boundary markers or explicit warnings to ignore instructions embedded in worker output.
Capability inventory: The skill has access to shell execution (Bash), file reading (Read), and high-privilege infrastructure management via the pylot CLI.
Sanitization: There is no mention of sanitization or validation of the data retrieved from workers before it is processed by the agent.
[DATA_EXPOSURE]: The skill accesses the local directory ~/.claude/session-env/ to identify conversation IDs. While this is used for legitimate session tracking within the Pylot environment, it involves reading from sensitive platform-specific environment paths.
[EXTERNAL_DOWNLOADS]: The skill uses curl to interact with the Pylot gateway API and to upload local files to dynamically generated presigned URLs for asset management. While these are directed at the vendor's infrastructure, they facilitate the transmission of local data to external endpoints.

pylot-cli