The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes local CLI tools such as gcx, k6, jq, and curl to manage load tests, fetch metrics, and process results. These commands are restricted to the management of k6 tests and are used in accordance with the vendor's documentation.
[EXTERNAL_DOWNLOADS]: The skill fetches resources including OpenAPI specifications, test logs, performance traces, and script files from official k6 and Grafana domains (api.k6.io, cloudlogs.k6.io). It also downloads artifacts such as browser screenshots from AWS S3 using pre-signed URLs generated by the official API.
[REMOTE_CODE_EXECUTION]: The skill includes a workflow for downloading k6 test scripts from the user's Grafana Cloud account and executing them locally using the k6 run command. This is used for validating scripts (smoke testing) before cloud deployment. As the scripts are sourced from the user's authenticated account on the vendor's platform, this represents intended operational functionality.
[PROMPT_INJECTION]: The skill processes external data from the k6 API, including test logs and script snapshots, which serves as an indirect prompt injection surface. 1. Ingestion points: API responses for logs, traces, and scripts in SKILL.md. 2. Boundary markers: Not explicitly used in prompt instructions. 3. Capability inventory: Execution of CLI tools (gcx, k6, curl). 4. Sanitization: Uses jq for structured data extraction.
[SAFE]: No malicious patterns such as credential exfiltration, persistence, or obfuscation were detected. The skill follows best practices for credential management by instructing the agent to prompt the user for API tokens rather than searching for them in sensitive files.

k6-manage