The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute various shell commands to build a 'red-capable loop'. Explicitly mentioned commands include curl, CLI diff, headless-browser script, git bisect run, and fuzz/property loop. These are used to verify and minimize bugs within the user's project environment.
[DYNAMIC_EXECUTION]: The skill requires the generation and execution of temporary scripts and test harnesses, such as 'throwaway harnesses' and 'headless-browser scripts'. This involves the agent creating executable code based on its diagnosis and then running it to confirm the bug reproduction.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data in the form of user-provided symptoms and project-specific files.
Ingestion points: The [symptom] argument hint and the project's foundation/glossary.md file.
Boundary markers: None identified. The skill does not specify delimiters or instructions to ignore embedded commands within the symptoms or files being analyzed.
Capability inventory: The skill has the ability to execute shell commands (curl, git bisect, etc.) and write files to the project directory (diagnosis.md).
Sanitization: No explicit sanitization or validation of the input symptom or project data is mentioned before these are used to formulate reproduction commands.

dx-diagnose