The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions for the agent to execute administrative shell commands using ssh and docker exec to manage remote servers and service containers.
[DATA_EXFILTRATION]: The skill accesses sensitive file paths and credentials as part of its management workflows.
It instructs the agent to use SSH keys located at ~/.ssh/coolify_key.
It includes commands to read sensitive .env files inside containers using docker exec <container> cat /app/.env.
It facilitates access to private keys and credentials via the Coolify API (/security/keys).
[INDIRECT_PROMPT_INJECTION]: The skill creates a vulnerability surface by ingesting and processing untrusted external data.
Ingestion points: Application deployment and runtime logs are fetched and analyzed in references/deploy-verify.md and references/turborepo-deploy.md using curl ... /logs.
Boundary markers: The instructions do not define delimiters to separate log data from agent instructions.
Capability inventory: The skill possesses powerful system capabilities including SSH access, Docker management, and network request tools (curl).
Sanitization: While the skill mentions masking secrets in logs using sed, it lacks validation or filtering of the log content itself.
[PRIVILEGE_ESCALATION]: The skill utilizes privileged access methods such as SSH connections to host servers and Docker container execution to perform deployments and infrastructure debugging, giving the agent broad control over the user's environment.

coolify-orchestrator