The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes curl commands to interact with the Bull Board API and uses node to perform local data processing and authentication setup.
[DYNAMIC_EXECUTION]: A Node.js script is executed via a heredoc in references/http-api.md to generate a temporary authentication configuration file for curl. This script implements security measures like credential escaping and file permission restriction (0o600).
[INDIRECT_PROMPT_INJECTION]: The skill retrieves and processes data from external API endpoints (queues, task details, logs), which may contain untrusted content that could influence agent behavior.
Ingestion points: JSON responses from Bull Board API endpoints such as /api/queues and task-specific detail paths.
Boundary markers: The instructions in SKILL.md advise the agent to output only a 'desensitized summary' of retrieved data, which acts as a manual instruction-based boundary.
Capability inventory: The skill possesses network access via curl (including write operations like PUT and POST) and local execution via node.
Sanitization: The skill relies on LLM-level summarization for safety rather than programmatic sanitization of the ingested API data.

bullboard