agentsclimarketplace

Mission control benchmark comparison

Skill MN755/Codex-Mission_Control/apps/mcp-server/src/mission_control_mcp_server/_bundled/skills/mission-control-benchmark-comparison

An Open Source Platform For Building Projects With OpenAI's Codex, Claude CLI, And Other Coding LLMs Using Multi-Agent Swarms And Customizable Workflows. Runs On Windows, MacOS, And Linux.

Install
npx -y skills add MN755/Codex-Mission_Control --skill mission-control-benchmark-comparison

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Compare before-and-after performance evidence through Mission Control without turning benchmark claims into theater.

SKILL.md

2.1 KB, as published. Nobody here has run it

Mission Control Benchmark Comparison

Purpose

Use Mission Control to compare GPU benchmark results before and after code changes with explicit evidence and known variance.

The Codex chat agent is not the Mission Control Manager. It is the bridge between the user and the Mission Control Manager.

Use when

  • The user asks whether a CUDA change made things faster.
  • A kernel optimization needs before-and-after comparison.
  • Performance work risks trading away correctness or stability.

Workflow

  1. Read the current validation summary and diagnostics.
  2. Ask Mission Control to identify the benchmark path that matches the changed code.
  3. Keep the benchmark loop paired with focused correctness validation.
  4. Record the delta and the confidence level honestly.
  5. Feed the result back into optimization, review, or handoff output.

Mission Control calls

Tools:

  • mission_control_start_task
  • mission_control_get_status
  • mission_control_get_handoff_summary

Resources:

  • mission-control://projects/{project_id}/validation-summary
  • mission-control://projects/{project_id}/handoff
  • mission-control://projects/{project_id}/diagnostics

User-facing output

  • Show the benchmark target and the claimed delta.
  • Show whether the evidence supports speedup, regression, or inconclusive results.
  • Show whether infrastructure issues contaminated the run.

Approval behavior

Benchmark commands and any supporting build or profile steps still use Mission Control approval rules.

Never do

  • Do not compare unmatched workloads.
  • Do not call noisy or one-off results a win.
  • Do not hide correctness regressions behind faster numbers.

Failure and fallback

If the repo has no clear benchmark target, say that plainly and fall back to a focused profiling or validation task first.

Example invocation

Use Mission Control to compare the current GPU benchmark against the previous baseline.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.