菜单
Run, inspect, import, and export ControlKeel benchmark suites and multi-subject matrices. Use this when comparing governed and external agents or validating policy changes.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Run objective, repeatable improvement work under an immutable verifier and hard iteration, cost, time, and no-progress limits. Use for approved experiments with an automated metric and sandboxed worker.
Write concise, calm technical responses whose structure matches the decision, evidence, or sequence being communicated. Use when drafting substantial plans, reviews, findings, documentation, or completion summaries.
Operate inside a ControlKeel-governed session. Use this before code edits, shell execution, delegation, deploy work, or any task that needs CK validation, findings, budget, proof, or routing context.
Keep a governed session within budget. Use this before long-running agent work, bulk processing, or any task where spend pressure could change the plan.
Close a governed work session with validation, proof, findings, budget, digest, learning, and handoff checks. Use when the user asks to wrap up, stop for the day, or leave work ready for another agent.
Audit tests that may pass without proving the claimed behavior. Use for periodic test-quality reviews or when coverage looks healthy but regressions still escape.
| name | benchmark-operator |
| description | Run, inspect, import, and export ControlKeel benchmark suites and multi-subject matrices. Use this when comparing governed and external agents or validating policy changes. |
| when_to_use | Activate when the user asks to benchmark, compare, or validate agent behavior, or when running regression tests after policy changes. |
| argument-hint | [benchmark suite name or agent to compare] |
| license | Apache-2.0 |
| compatibility | ["codex","claude-standalone","claude-plugin","copilot-plugin","github-repo","open-standard","cline-native","cursor-native","windsurf-native","continue-native","letta-code-native","pi-native","roo-native","goose-native","opencode-native","gemini-cli-native","kiro-native","kilo-native","amp-native","augment-native","hermes-native","multica-native","openclaw-native","devin-terminal-native","warp-native","droid-bundle","forge-acp"] |
| disable-model-invocation | true |
| metadata | {"author":"controlkeel","version":"2.0","category":"benchmark","ck_mcp_tools":["ck_observability"]} |
Use this skill when the task is benchmark orchestration instead of normal governed delivery work.
For generated observability coverage, prefer the human-gated loop:
ck_observability reports for saved_evals, benchmark_drafts, benchmark_scenarios, and benchmark_history.promotions as advisory evidence; do not mutate policy, router, prompt, or autofix artifacts from benchmark results alone.