A command for explaining or running performance comparisons between Governor and Caveman, using a repeatable benchmark suite.

Use it to refresh captured comparison data, run the benchmark, and produce JSON and Markdown results. A logged-in Claude command-line account is needed for live captures.

It provides a defined way to compare the two output-handling approaches instead of relying on impressions.

benchmark — Command for coding agents by 0xhimanshu: A command for explaining or running performance comparisons between…

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods

npx agentmods add commands/0xhimanshu/governor/benchmark

Clone the repo

git clone --depth 1 https://github.com/0xhimanshu/governor

Or install governor, the plugin that ships this one along with the rest of its 1 skill, 11 commands, 5 hooks.

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.

README.md

[![agentmods](https://agentmods.dev/badge/commands/0xhimanshu/governor/benchmark.svg)](https://agentmods.dev/commands/0xhimanshu/governor/benchmark)

Your own site

<a href="https://agentmods.dev/commands/0xhimanshu/governor/benchmark"><img src="https://agentmods.dev/badge/commands/0xhimanshu/governor/benchmark.svg" alt="Measured on agentmods" height="20"></a>

Per session 11 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 763 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. Scan, not verified.

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.

Model	Per session	Once invoked
Fable 5	$0.00011	$0.00763
Opus 5	$0.00005	$0.00381
Sonnet 5	$0.00002	$0.00153
Haiku 4.5	$0.00001	$0.00076

Measured 5d ago against content hash a420187d9f0e, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

benchmark 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 5d 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.

commands/benchmark.md · 80 lines

What it actually says

Governor Benchmark

Use this command when the user wants an actual Governor vs Caveman performance comparison or wants to run the V2 benchmark suite.

If the user says /governor:benchmark refresh-caveman, refresh captured Caveman comparator files first, then rerun the V2 suite:

python3 "${CLAUDE_PLUGIN_ROOT}/scripts/capture_fixture_conditions.py" \
  --condition caveman \
  --model sonnet \
  --write-summary "${CLAUDE_PLUGIN_ROOT}/benchmarks/captured/caveman/latest-summary.json"
python3 "${CLAUDE_PLUGIN_ROOT}/scripts/run_benchmark.py" \
  --write-json "${CLAUDE_PLUGIN_ROOT}/benchmarks/v2-fixture-results.json" \
  --write-md "${CLAUDE_PLUGIN_ROOT}/benchmarks/v2-fixture-results.md"

If capture refresh reports claude-auth-unavailable, stop and explain that the benchmark can only use inline fallback comparators until Claude CLI is logged in.

If the user says /governor:benchmark refresh-governor, refresh the replayable Governor comparator files for reference-style fixtures, then rerun the V2 suite:

python3 "${CLAUDE_PLUGIN_ROOT}/scripts/capture_fixture_conditions.py" \
  --condition governor \
  --model sonnet \
  --write-summary "${CLAUDE_PLUGIN_ROOT}/benchmarks/captured/governor/latest-summary.json"
python3 "${CLAUDE_PLUGIN_ROOT}/scripts/run_benchmark.py" \
  --write-json "${CLAUDE_PLUGIN_ROOT}/benchmarks/v2-fixture-results.json" \
  --write-md "${CLAUDE_PLUGIN_ROOT}/benchmarks/v2-fixture-results.md"

If the user provides a CSV path, run:

python3 "${CLAUDE_PLUGIN_ROOT}/scripts/compare_benchmarks.py" "${CSV_PATH}"

If no CSV is provided, run the V2 fixture suite:

python3 "${CLAUDE_PLUGIN_ROOT}/scripts/run_benchmark.py" \
  --write-json "${CLAUDE_PLUGIN_ROOT}/benchmarks/v2-fixture-results.json" \
  --write-md "${CLAUDE_PLUGIN_ROOT}/benchmarks/v2-fixture-results.md"

If the user explicitly wants live Caveman captures and Claude CLI auth is available, refresh them first:

python3 "${CLAUDE_PLUGIN_ROOT}/scripts/capture_fixture_conditions.py" \
  --condition caveman \
  --model sonnet \
  --write-summary "${CLAUDE_PLUGIN_ROOT}/benchmarks/captured/caveman/latest-summary.json"

Then explain the protocol from benchmarks/README.md:

Run python3 scripts/run_benchmark.py for the fixture-based V2 suite.
Compare the same fixture inputs under control, caveman, and governor.
Use fresh sessions, the same model, the same repo commit, and exact prompt text when using model-judged evaluations.
Track VCLR, decision preservation, wrong-decision rate, and token savings in that order.
Treat inline comparator text as framework-grade only; captured Caveman files are needed for publishable head-to-head claims.

Keep the response practical. Do not claim Governor beats Caveman until the CSV contains measured runs. Token savings are a win only when success rate, requirement coverage, quality score, and verification evidence are retained. For coding benchmarks, call out runtime/selector/event-handler bugs as quality regressions even if output tokens drop.

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.

5d ago First seen · 80 lines · 11 tokens per session scan A a420187d9f0e

Subscribe to this mod's changes

benchmark is a command published in the GitHub repository 0xhimanshu/governor (133 stars, last pushed 2mo ago), licensed MIT. It adds 11 tokens to every session and 763 once invoked, about $0.0001 per session on Opus 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 2026-08-30.

Analytics and metrics LLM ai-tools claude-code cli context-window developer-tools

benchmark

Getting it into your agent

Wrote this? Show the measurements

What it costs to keep this loaded

Grade A, and why

Nothing flagged

What it actually says

Governor Benchmark

What this file has done since we first saw it

Other commands, from other repositories

benchmark

list-branches

remove-branch

/release

generate-rfcs

test-strategy