The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill defines patterns for capturing application errors and parsing browser cookies into the PostHog platform. This creates a data ingestion surface where external, untrusted content (such as error messages or cookie payloads) could potentially contain malicious instructions intended to influence the agent's behavior.
Ingestion points: posthog.captureException(error) and cookie parsing in the onRequestError hook within references/nextjs.md.
Boundary markers: No specific delimiters or "ignore embedded instructions" warnings are present in the provided implementation templates.
Capability inventory: The skill facilitates network transmission of captured data to vendor-owned API endpoints.
Sanitization: No specific sanitization or filtering logic is included in the implementation examples.
[EXTERNAL_DOWNLOADS]: The documentation guides the installation of posthog-js and posthog-node packages from standard Node.js registries. These are official libraries provided by the skill's author.
[CREDENTIALS_UNSAFE]: The skill promotes secure handling of sensitive information by instructing users to utilize environment variables (.env.local) for project tokens and host URLs, explicitly advising against hardcoding these values.

error-tracking-nextjs