The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructions direct the agent to fetch documentation from official and well-known sources, including PestPHP (pestphp.com) and Laravel (laravel.com). These references are found in several files: rules/assertions.md, rules/endpoint-tests.md, rules/finding-features.md, rules/isolation.md, and rules/performance.md.
[INDIRECT_PROMPT_INJECTION]: The skill requires the agent to ingest and process untrusted data from local project files (code under test and existing tests) and external documentation to generate or review tests. This creates a surface where malicious instructions hidden in these sources could influence the agent's output.
Ingestion points: Reads local source code files and existing test files (SKILL.md, rules/naming.md); fetches external documentation from the web (rules/finding-features.md).
Boundary markers: The instructions do not specify any delimiters or warnings to ignore embedded instructions within the processed data.
Capability inventory: The skill facilitates file reading, analysis, and the generation of test code within the local development environment.
Sanitization: No explicit sanitization or filtering of the ingested content is described.
[METADATA_POISONING]: The SKILL.md file contains metadata identifying the author as "laravel," which conflicts with the actual skill author identity. This is deceptive metadata but does not appear to contain executable malicious instructions.

testing-best-practices