The Agent Skills Directory

[PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection.
Ingestion points: It is designed to read and analyze source code and pull request content during coding, review, and audit modes (SKILL.md).
Boundary markers: There are no explicit instructions for the agent to use delimiters or boundary markers to differentiate between its own instructions and untrusted content from the ingested files.
Capability inventory: The skill has access to sensitive tools including file modification (Edit, Write) and shell execution (Bash), which could be leveraged if the agent is manipulated by instructions embedded in processed code.
Sanitization: No specified mechanisms for sanitizing or escaping content from external data sources are described.
[EXTERNAL_DOWNLOADS]: The skill references established open-source libraries and reputable monitoring platforms.
It fetches configuration and alerting rules from the author's official GitHub repositories and associated documentation.
It recommends official and well-known Go packages for OpenTelemetry, Prometheus, and Grafana.
It integrates with verified analytics providers such as PostHog and Segment using official SDKs.

golang-observability