The Agent Skills Directory

[COMMAND_EXECUTION]: The skill generates a local shell script (squadron-inbox-poll.sh) designed to poll the Squadron API. This script is intended to be executed periodically via a task scheduler to notify the agent of new tasks.
[DATA_EXPOSURE]: The skill is configured to read credentials such as SQUADRON_API_KEY and configuration like SQUADRON_URL from the environment and the /data/workspace/.env file. This follows standard security practices for managing secrets in automated environments.
[INDIRECT_PROMPT_INJECTION]: The skill ingests external data from the Squadron API, including task titles, descriptions, and inbox messages, which could potentially contain malicious instructions.
Ingestion points: API responses from community.iamstarchild.com containing task and knowledge base content.
Boundary markers: None explicitly implemented in the provided code snippets to distinguish external content from system instructions.
Capability inventory: The skill possesses the ability to execute shell commands (via scheduled tasks), perform network operations (via requests and curl), and modify persistent storage (via knowledge base write operations).
Sanitization: The skill uses JSON parsing in both Python and Bash implementations to isolate specific data fields (e.g., priority, title) before they are processed or displayed.

@1892/squadron