The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection through its interaction with project source code.
Ingestion points: The paths configuration in SKILL.md allows the agent to ingest all **/*.go files within a project directory. Maliciously crafted content within these files (such as comments or string literals) could potentially influence the agent's behavior.
Boundary markers: Absent. There are no instructions in SKILL.md defining specific delimiters or warning markers to help the agent distinguish between static code data and its own control instructions.
Capability inventory: According to the allowed-tools in SKILL.md, the agent has Write and Edit permissions, and can execute Bash commands (specifically go, golangci-lint, and git). These capabilities could be misused if an injection successfully subverts the agent's instructions.
Sanitization: Absent. The skill does not provide guidelines for sanitizing, validating, or escaping data retrieved from the codebase before it is integrated into the agent's reasoning or output.

golang-concurrency