The Agent Skills Directory

[Data Protection]: The skill incorporates a PII redaction mechanism within the Python script, using regular expressions and key-based filtering to strip sensitive information like email addresses and IP addresses from API responses.
[Credential Management]: Authentication is managed through environment variables rather than direct chat input. The instructions explicitly advise against requesting tokens in the conversation, aligning with secure secret management practices.
[External Communication]: The skill interacts with the official Sentry API (sentry.io), a well-known service, for fetching observability data. The operations are limited to read-only GET requests.
[Potential for Indirect Prompt Injection]: This skill ingests data from an external API, which is a common surface for indirect prompt injection. Ingestion points: External data enters through the request_json function in scripts/sentry_api.py. Boundary markers: While there are no specific markers for the output, the instructions provide clear guidance on redacting PII and suppressing stack traces. Capability inventory: The skill has no capabilities to execute commands or write files; it only performs GET requests to the Sentry API. Sanitization: The Python script implements a redact_data function to sanitize the data before it is presented to the agent context.

sentry