The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from the target repository environment which could contain malicious instructions.
Ingestion points: The skill discovers shell commands from the repository and collects their raw output to serve as evidence for stage completion as described in SKILL.md.
Boundary markers: The skill does not define specific delimiters or instructions to treat the ingested command output as untrusted data or to ignore natural language instructions within that output.
Capability inventory: The skill is authorized to use the Agent tool and other sub-skills to execute discovered shell commands and perform repository modifications.
Sanitization: No specific sanitization, filtering, or validation logic is specified for the data retrieved from the environment before it is processed by the agent.
[DYNAMIC_EXECUTION]: The skill performs runtime discovery and execution of commands based on the repository environment.
Evidence: SKILL.md instructs the agent to "Discover what this repository can actually verify" and "Record the command... test runner, coverage, complexity or lint gates, mutation tooling, build, failure injection" for execution during the staged process.

software-factory