The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to identify and execute project-specific build, lint, and test commands found in manifest files such as package.json, Makefile, Cargo.toml, and pyproject.toml. This involves running arbitrary shell commands defined within the target repository.
[DYNAMIC_EXECUTION]: To orchestrate the sweep, the skill directs the agent to generate and execute custom scripts or small programs, such as shell loops, locally. These generated scripts are used to manage the execution flow of various regression checks.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data, creating a potential attack surface for indirect prompt injection.
Ingestion points: The skill ingests data from codebase files, test suite outputs, CI logs, and network responses from deployed environments as defined in SKILL.md.
Boundary markers: The instructions explicitly define a 'Security boundary', directing the agent to treat all ingested content as data and never as instructions that could alter the sweep's scope or behavior.
Capability inventory: The agent is empowered to execute shell commands, perform network requests via curl, and execute SQL queries against data stores.
Sanitization: The instructions mandate that any embedded directives found within processed data must be reported as text content rather than being acted upon or allowed to override the agent's current task.

awesome-regression-sweep