The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use git ls-remote to confirm the existence of specific version tags on the author's official GitHub repository. This is a read-only network operation used to ensure that links in review reports are correctly pinned to the version of the skill currently in use.
[EXTERNAL_DOWNLOADS]: The skill directs the agent to fetch the README file from the author's repository (github.com/ctxr-dev) when companion skills are missing. This is used to provide the user with up-to-date installation instructions and is limited to trusted sources associated with the skill author.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze Go source code and read local agent configuration files, creating an attack surface for indirect prompt injection.
Ingestion points: The skill ingests Go files, .golangci.yml, and agent installer lockfiles (skills-lock.json) as defined in corpus/onboarding.md and references/companion-skills.md.
Boundary markers: The skill relies on structured review patterns and the agent's ability to distinguish between code content and instructions during the review process.
Capability inventory: Capabilities are restricted to reading project files, reading agent configuration, writing reports to a specific home directory path (~/.skill-golang/), and performing Git version checks.
Sanitization: The skill utilizes specific parsing logic for Go syntax and lockfile JSON structure to minimize the impact of potentially malicious embedded instructions.

golang