The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill allows external, untrusted data from webhooks to be directly interpolated into agent prompts using templates like --prompt "New GitHub issue #{issue.number}: {issue.title}".
Ingestion points: Incoming HTTP POST requests from external services (GitHub, GitLab, Stripe, etc.) described in SKILL.md.
Boundary markers: The documentation does not specify the use of delimiters (e.g., XML tags, triple backticks) or specific instructions to the agent to ignore embedded commands within the interpolated payload data.
Capability inventory: Agents triggered by these webhooks can be configured with powerful skills (e.g., github-code-review), which could be subverted if the payload contains malicious instructions.
Sanitization: There is no evidence of input validation, filtering, or escaping for the data fields extracted from the JSON payloads.
[COMMAND_EXECUTION]: The documentation instructs users to perform administrative tasks via shell commands, including manual configuration of system services (systemctl), gateway execution, and network health checks (curl). While these are standard for setup, they represent the skill's operational footprint on the host system.

webhook-subscriptions