The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted performance data that could be influenced by external factors, creating a surface for indirect prompt injection in CI output.
Ingestion points: The scripts/run_perf_gate.py script ingests data from k6-summary.json and .lighthouseci/lhr-*.json (specifically the finalUrl and metric names).
Boundary markers: The script does not use specific boundary markers or delimiters when interpolating data into the Markdown summary report.
Capability inventory: The script has the capability to write to $GITHUB_STEP_SUMMARY and control CI job execution status via sys.exit().
Sanitization: There is no evidence of sanitization or escaping of the values (like subject or url) extracted from the performance artifacts before they are printed as Markdown.

perf-budget-gate