The Agent Skills Directory

[COMMAND_EXECUTION]: The skill extensively uses shell commands, docker, and docker-compose to manage the lifecycle of a local multi-node cluster. This includes starting containers, executing commands inside them (docker exec), and managing networks.
[EXTERNAL_DOWNLOADS]: The instructions guide the user to download a geolocation database (GeoLite2-City.mmdb) from MaxMind, which is a well-known service. This is a legitimate dependency for the proxy plugin mentioned in the skill.
[DYNAMIC_EXECUTION]: The skill employs python3 -c one-liners to parse JSON output from cluster API calls (e.g., extracting node IDs). This is a routine method for handling structured data in shell-based workflows.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect injection surface as it ingests and processes data from local API endpoints (e.g., http://localhost/api/internal/nodes).
Ingestion points: Data enters the agent's context through curl requests to local API services, which is then parsed by Python scripts in SKILL.md.
Boundary markers: No explicit delimiters or safety instructions are used when interpolating this external data into subsequent commands.
Capability inventory: The skill can execute shell commands, perform Docker operations, and make network requests via curl.
Sanitization: No explicit sanitization or validation of the API-provided data is performed before it is used to construct shell variables (e.g., NIDS, PID). However, given the data source is the user's own local development cluster, the risk is negligible.

start-temps-cluster