The Agent Skills Directory

[SAFE]: The skill is a legitimate tool designed for debugging Harness CI/CD pipelines. It uses official platform-specific MCP tools to retrieve diagnostic information. All tool calls and parameters align with the stated purpose of pipeline analysis.
[PROMPT_INJECTION]: The skill processes untrusted external content (execution logs and pipeline YAML definitions), which represents an indirect prompt injection surface. Ingestion points: Data enters the context via harness_diagnose and harness_get tools as defined in SKILL.md. Boundary markers: No explicit delimiters or warnings to ignore embedded instructions are present in the analysis prompts. Capability inventory: The skill only performs read-only analysis and does not have access to subprocess execution, file-writing, or outbound network tools. Sanitization: No explicit sanitization or filtering of log content is mentioned. Due to the limited capabilities available to the agent, this is considered a low-risk surface.
[DATA_EXFILTRATION]: No unauthorized data transmission was detected. While the skill retrieves sensitive information like build logs and configurations, it does so through managed tool calls within the Harness environment for diagnostic purposes.

debug-pipeline