The Agent Skills Directory

[SAFE]: The skill is authored by Harness and utilizes internal MCP tools (harness_list, harness_get) to perform deployment readiness assessments. It focuses on gathering telemetry and configuration data to provide decision support for canary rollouts and environment comparisons.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection by processing external data from the Harness platform to generate recommendations.
Ingestion points: The skill ingests untrusted data from the Harness API via harness_list (execution logs, artifact metadata) and harness_get (environment variables, config maps, Kubernetes manifests) in SKILL.md.
Boundary markers: Absent. There are no explicit delimiters or instructions provided to the agent to treat data from the MCP tools as untrusted or to ignore instructions embedded within retrieved configurations.
Capability inventory: The skill is restricted to read and analysis operations. It does not possess file-write, network exfiltration, or command execution capabilities; it is explicitly forbidden from running pipelines or performing debugging actions.
Sanitization: No evidence of validation, escaping, or filtering of the retrieved Harness data is present in the instructions.

deployment-readiness