The Agent Skills Directory

[SAFE]: No malicious patterns or security risks were identified. The skill demonstrates high security awareness by providing instructions to avoid Go module proxy poisoning (Failure Mode F7), prohibiting manual checksum manipulation (Failure Mode F9), and requiring validation of security errors (Failure Mode F15).- [INDIRECT_PROMPT_INJECTION]: The skill defines a process for ingesting external, untrusted data from third-party libraries.
Ingestion points: Migration guides and CHANGELOG.md files from external Go libraries (SKILL.md, Phase 0 and Failure Mode F6).
Boundary markers: The skill warns against trusting external documentation alone and mandates independent verification.
Capability inventory: The agent has access to file system operations, network access via go get and git, and execution capabilities via go test and templ generate.
Sanitization: The protocol explicitly instructs the agent to 'grep exported symbols yourself' (Failure Mode F6) to verify the actual API surface rather than relying on potentially malicious or incomplete external summaries.- [DYNAMIC_EXECUTION]: The skill utilizes standard development tools to generate and execute code as part of its verification process.
Evidence: The protocol requires running templ generate to refresh generated Go code (SKILL.md, Phase 3.4).
Evidence: The protocol mandates running the full test suite using go test ./... to verify behavioral consistency (SKILL.md, Phase 4.1).

go-ecosystem-upgrade