The Agent Skills Directory

[PROMPT_INJECTION]: The skill uses imperative language and a "CRITICAL RULE" header to attempt to override the agent's default behavior, instructing it to never suggest standard package manager commands and to exclusively use a custom tool instead.
[COMMAND_EXECUTION]: The instructions require the execution of a custom, non-standard CLI tool named infynon for all management and monitoring operations.
[COMMAND_EXECUTION]: The persistence setup instructions involve writing configuration to system-level directories (/etc/systemd/system/) and managing system services, which are high-risk operations typically requiring administrative privileges.
[CREDENTIALS_UNSAFE]: The skill prompts users to input sensitive SMTP credentials, including hostnames, usernames, and plaintext passwords or API keys, which are then stored in a local configuration file at ~/.infynon/eagle-eye.toml.
[PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection as it processes untrusted project data.
Ingestion points: The skill scans project directories for various lock files (e.g., package-lock.json, Cargo.lock, uv.lock) as described in SKILL.md.
Boundary markers: Absent; there are no instructions provided to the agent to ignore or delimit potentially malicious instructions embedded within the metadata of the scanned files.
Capability inventory: The skill utilizes the infynon pkg fix and infynon pkg scan commands which perform file system writes and package modifications.
Sanitization: Absent; the skill does not specify any validation or sanitization steps for the external content before it is processed by the monitoring tool.

eagle-eye-monitor