The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the execution of various system-level utilities for service management and network diagnostics, such as systemctl, ss, lsof, nginx, and caddy. These operations are fundamental to the skill's primary purpose of host and service administration.
[INDIRECT_PROMPT_INJECTION]: The skill has a vulnerability surface where it ingests and processes untrusted or externally influenced data from local state files and system observations. 1. Ingestion points: Data is read from ~/Clawic/data/server/memory.md, config.yaml, servers.md, domains.md, and profile.yaml at the start of every session. 2. Boundary markers: The instructions do not define explicit delimiters or headers to distinguish between trusted instructions and data retrieved from system observations. 3. Capability inventory: The skill possesses capabilities to write to system configuration paths (e.g., /srv, /etc/myapp/env) and execute service reloads or restarts. 4. Sanitization: The skill incorporates best practices by validating configurations using tools like nginx -t and systemd-analyze verify before application, which helps mitigate accidental misconfiguration caused by malformed input.

Server