The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses local CLI tools and system commands (hermes, systemctl, curl, ps, grep) to manage the webhook gateway service and local configuration files. These operations are restricted to the local user environment and aligned with the skill's stated purpose.
[PROMPT_INJECTION]: The skill creates an entry point for untrusted external data, which is a vector for indirect prompt injection.
Ingestion points: External data from webhook payloads (e.g., GitHub issue titles, PR descriptions) is processed in SKILL.md examples.
Boundary markers: The skill lacks explicit boundary markers or safety instructions in its prompt templates to prevent the agent from obeying instructions embedded in the external payload.
Capability inventory: While the skill itself is a management interface, it triggers agent runs that may possess significant capabilities (e.g., file system access, tool execution).
Sanitization: There is no evidence of sanitization or validation of external content before it is interpolated into the prompt templates (e.g., {issue.body}).

webhook-subscriptions