The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data by performing a deep inventory and audit of a target software repository, including source code, documentation (blueprint/ folder), and deployment configurations. This content could contain malicious instructions designed to manipulate the agent's behavior or findings.
Ingestion points: The agent is instructed to read all surface-related files, repository code, deployment manifests, DNS records, and documentation from the target project.
Boundary markers: The instructions lack specific requirements for the agent to use delimiters or 'ignore' commands when processing text from the analyzed repository.
Capability inventory: The skill has the capability to write files to the local directory and execute a local Python script to process audit data.
Sanitization: There is no mention of sanitizing or escaping content retrieved from the target repository before it is processed or included in the final report.
[DYNAMIC_EXECUTION]: The skill relies on the execution of a local Python script (scripts/readiness.py) to validate the audit payload, compute scores, and generate the final report. While the script uses only standard libraries and performs legitimate data processing, executing scripts at runtime is a noteworthy behavior.
Evidence: SKILL.md contains the instruction: python <this skill's directory>/scripts/readiness.py <output folder>/production-readiness.json.
[COMMAND_EXECUTION]: The skill requires the ability to execute shell commands to run the project's own build tools, tests, linters, and scanners as part of the evidence-gathering phase, as well as its own report generation script.
Evidence: Step 4 ('Gather evidence') and Step 7 ('Score and render') in SKILL.md explicitly direct the agent to run external build tools and the provided Python script.

production