The Agent Skills Directory

[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by instructing the agent to process untrusted external content.
Ingestion points: The agent reads project manifest files (package.json, requirements.txt, go.mod), lockfiles, and external changelogs or release notes from third-party repositories.
Boundary markers: The instructions lack explicit boundary markers or warnings to disregard instructions that may be embedded within the external changelogs or manifests.
Capability inventory: The skill utilizes the agent's ability to execute shell commands via package managers (npm, pip, go), which can trigger installation scripts or binary execution.
Sanitization: No specific sanitization or verification steps are outlined for external text content before it is incorporated into the agent's context.
[COMMAND_EXECUTION]: The skill workflow involves the execution of shell commands for package management (e.g., npm install, pip install, go get). While these are standard development tools, they represent a capability that could be abused if the input (package names or versions) is manipulated.

dependency-updates