verify

verify is a skill for Claude Code from melodic-software/claude-code-plugins. It costs 92 tokens per session (2,937 once invoked), scanned A, original, MIT.

An independent check of a reported performance result. It repeats the measurement from a fresh context and checks whether the change preserved existing behavior.

In plain words
What is it for?
Use it to verify optimization claims, compare before-and-after behavior, and decide whether a performance target was actually met.
Why use it?
It helps catch incorrect numbers and defects that normal tests or green continuous integration checks can miss. It also avoids treating a near miss as a success.

Skill for Claude Code

Written for Claude Code: disable-model-invocation in frontmatter. Also seen: reads .claude/ paths; mentions Claude Code.

Runs only inside its plugin — its command needs a path that Claude Code sets for a plugin’s own hooks and for nothing else. Install the plugin, not this.

Part of the performance plugin — 4 skills shipped together

Good fit Use it to verify optimization claims, compare before-and-after behavior, and decide whether a performance target was actually met.

Compare 6 skills from other repositories ↓
Install

Getting it into your agent

This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.

Claude Code
/plugin marketplace add melodic-software/claude-code-plugins
Claude Code
/plugin install performance

Made for: Claude Code.

Or install performance, the plugin that ships this one along with the rest of its 4 skills.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for verify

README.md
[![agentmods](https://agentmods.dev/badge/skills/melodic-software/claude-code-plugins/verify/github.svg)](https://agentmods.dev/skills/melodic-software/claude-code-plugins/verify)
Your own site
<a href="https://agentmods.dev/skills/melodic-software/claude-code-plugins/verify"><img src="https://agentmods.dev/badge/skills/melodic-software/claude-code-plugins/verify/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for verify

Your own site · 80×15
<a href="https://agentmods.dev/skills/melodic-software/claude-code-plugins/verify"><img src="https://agentmods.dev/badge/skills/melodic-software/claude-code-plugins/verify.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 92 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,937 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00092 $0.02937
Opus 5.5 $0.00037 $0.01175
Sonnet 5.5 $0.00018 $0.00587
Haiku 4.5 $0.00009 $0.00294

Measured 18d ago against content hash f655c63ca1d0, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

verify scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 18d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

plugins/performance/skills/verify/SKILL.md · 193 lines

How it starts

The opening of the file, as written. The whole thing — 193 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Arguments. [<claim or target>]. e.g. /performance:verify the 4-to-1 spawn reduction

Purpose

Answers "is this result real, and what does it actually say?"

This phase is where blocking correctness defects that the implementer, the implementer's own test suite, and a full green CI run all missed get caught. That is why it is a separate phase from measurement and not a step inside it.

Read ${CLAUDE_PLUGIN_ROOT}/reference/harness-integrity.md first.

1. Fresh context, and adversarial by construction

Dispatch a verifier that does not inherit the implementer's numbers. Give it the trees and the claim; withhold the reasoning that produced the figures. A verifier shown the expected answer verifies the answer, not the work.

The brief should say, in substance: distrust the reported numbers, re-derive them yourself, and report what you actually observe including the ways you could not reproduce it.

The brief also sets a turn budget, a limit the verifier keeps for itself: 30 turns, stop gathering and re-measuring by turn 22, and spend the remaining turns writing the report. Run the highest-value checks first: re-derive the headline number and the differential before any secondary check. A report that ends short names what it did not reach under Not covered:. A verification that ran out of turns is not complete, so anything unreached stays NOT MET or unverified.

Two independent verifiers routinely find different defects, so one verifier is the floor, not the target.

2. Prove behavior did not change, with a differential

A passing test suite is not a behavior proof. It proves nothing asserted broke. It does not prove behavior is unchanged, because it only checks what someone thought to assert.

Run a differential: the pre-change and post-change subject over a harvested corpus of real inputs, requiring byte-identical output. Use the bundled one rather than writing one: python3 "${CLAUDE_PLUGIN_ROOT}/scripts/differential.py" --baseline <path> --candidate <path> --config <json>. Its config takes a matrix of modes, and it refuses a run where neither arm ever produced output. Run it from the Bash tool, whose shell resolves python3 (type -P python3 names the one it found); a bare name inside another process's subprocess call searches a different PATH (harness-integrity rule 6). The module docstring documents the config.

Read the full file on GitHub · 193 lines

Files

What ships with it

1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 18d ago Changed · +6 lines f655c63ca1d0
  2. 20d ago Changed · +31 lines 5ec7a075177d
  3. 21d ago Changed · +35 lines · -80 tokens per session 9e41c81c1053
  4. 25d ago Changed · +16 lines 45456588b82a
  5. 1mo ago Changed · +6 lines d2dde6ee21b0
  6. 1mo ago Changed · -1 tokens per session b510cf8810dc
  7. 1mo ago First seen · 99 lines · 173 tokens per session scan A 54f20272e24c

Subscribe to this mod's changes

verify is a skill published in the GitHub repository melodic-software/claude-code-plugins (26 stars, last pushed today), licensed MIT. It adds 92 tokens to every session and 2,937 once invoked, about $0.0004 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.