The Agent Skills Directory

[COMMAND_EXECUTION]: The skill involves the execution of various development tools and scripts to build feedback loops for bug reproduction. Key activities include:
Using git bisect run for automated bisection.
Executing network tests with curl or HTTP scripts.
Running browser automation scripts (Playwright/Puppeteer).
Generating and running local shell scripts based on provided templates to capture user input during debugging sessions.\n- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it processes untrusted external data during the diagnosis process.
Ingestion points: The agent is instructed to read HAR files, log dumps, and core dumps (documented in Phase 1) to identify symptoms.
Boundary markers: There are no explicit instructions to use delimiters or ignore embedded commands within the ingested data.
Capability inventory: The skill utilizes shell execution, network access, and file system modification (across multiple phases).
Sanitization: The instructions do not include steps to sanitize or filter the content of captured artifacts before analysis.

diagnose