The Agent Skills Directory

Command Execution: The skill performs command-line operations to discover release artifacts using standard tools such as svn, gh (GitHub CLI), and aws (AWS CLI). These operations are restricted to listing and discovery tasks, adhering to a read-only policy on the distribution surface.
Dynamic Execution: A pre-flight validation step executes a local Python module (setup_preflight) to verify the project environment. Additionally, the skill supports a 'self-hosted' backend that allows for command templates defined in configuration files, which is a common pattern for extensible administrative tooling.
Indirect Prompt Injection Surface: The skill ingests external data from release distribution listings and issue tracker descriptions. While this presents an injection surface, the skill includes specific instructions for the agent to treat all external content as untrusted data and provides a mechanism to flag potential injection attempts.
Ingestion points: Release distribution listings (SVN, GitHub, S3) and release planning issue bodies.
Boundary markers: The skill contains explicit documentation instructing the agent to treat external content as data, never as instructions.
Capability inventory: Shell command execution for release discovery tools and a pre-flight Python script.
Sanitization: The instructions provide a protocol for detecting and flagging suspicious patterns in external data without executing them.
External References: The skill references official Apache Software Foundation infrastructure and well-known services like GitHub and AWS, which is appropriate for a release management tool developed for that organization.

archive-sweep