[not-matthias]

TLDR: Agents had no guidance on writing valid CodSpeed benchmarks. This adds a compact skill with rules, a verify step and maintenance tasks, and points the existing skills at it.

New skill codspeed-write-benchmarks

• Language-agnostic rules: measure only the code under test, constant inputs, isolated state, sizing per instrument, stable identity.
• Verify step: static review always, one CodSpeed run checked through the MCP tools, reruns only for noisy benchmarks.
• Maintenance: noisy benchmarks, suite audits, renames and removals, changes to the code under test.
• Short notes for Rust, Node.js, Python and the exec harness. Details link to the CodSpeed docs.

Existing skills and plugin

• codspeed-setup-harness: description narrowed to first-time setup, so both skills no longer trigger on "create a benchmark". Its benchmark-writing step now points to the new skill.
• codspeed-optimize: fix noisy or invalid benchmarks with the new skill before optimizing.
• plugin.json: long description mentions the new skill.

Review notes

• "The old benchmark shows up as skipped" after an identity change is inferred, not stated in the docs.

Fixes COD-4129

[greptile-apps]

Confidence Score: 5/5

[Low impact] The PR appears safe to merge; no new actionable issues were found.

Summary

Adds codspeed-write-benchmarks to guide benchmark writing, checks, and maintenance.

• CodSpeed agents now have a dedicated guide for writing and maintaining benchmarks.
• Setup and optimization now point benchmark work to the dedicated guide.

_{Reviews (2) · Last reviewed commit: "docs(plugin): mention benchmark writing ..." · Reviewed by Greptile}

[codspeed]

Merging this PR will not alter performance

2 warnings

⚠️ Walltime results may be unreliable

Some benchmarks used the Walltime instrument outside CodSpeed Macro Runners, which can make the results inconsistent.

Learn more about Macro Runners

See affected benchmarks

⚠️ 8 benchmarks are dominated by system calls

System calls cannot be simulated consistently and are left out of the measure. The reported value understates the real cost. Switch to the Walltime instrument to measure them.

Learn more about system calls

See affected benchmarks

✅ 53 untouched benchmarks
⏩ 6 skipped benchmarks¹

---

_{Comparing cod-4129-provide-a-skill-for-creating-and-maintaining (28f6463) with main (21b0cda)}

Footnotes

1. 6 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[SuperMuel]

I think you meant MCP rather than CLI

If the CodSpeed CLI is installed and authenticated

If the CLI is not available,

[SuperMuel]

daily: make sure to sync the docs as well