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.
npx agentmods add commands/random6913/claude-code-superkit/benchmarkgit clone --depth 1 https://github.com/RaNDoM6913/claude-code-superkitWrote 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.
[](https://agentmods.dev/commands/random6913/claude-code-superkit/benchmark)<a href="https://agentmods.dev/commands/random6913/claude-code-superkit/benchmark"><img src="https://agentmods.dev/badge/commands/random6913/claude-code-superkit/benchmark.svg" alt="Measured on agentmods" height="20"></a>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 | $0.00009 | $0.01477 |
| Opus 5 | $0.00005 | $0.00739 |
| Sonnet 5 | $0.00002 | $0.00295 |
| Haiku 4.5 | $0.00001 | $0.00148 |
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 today.
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.
How it starts
The opening of the file, as written. The whole thing — 131 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Go Benchmark Runner
Role
Run Go benchmarks with memory stats and 6 repetitions; in --compare mode, benchmark two git refs and diff them with benchstat. Go-only command: on non-Go projects it exits at Step 1 with zero side effects.
Hard Rules
- No
go.modfound → report "No Go project detected" and STOP. - No
func Benchmarkfunctions found → report "No benchmark functions found" and STOP. - Compare mode: record
ORIGINAL_REFandSTASHED(Step 4) BEFORE anygit checkoutorgit stash. Never switch first. - Once Step 4 switched refs, the Restore epilogue (Step 6) ALWAYS runs — after success, after a failed benchmark run, after a missing tool, after any error. Never finish with the repo on another ref or with changes still stashed.
- Never run destructive git commands (
reset --hard,clean -f, checkout that overwrites uncommitted changes). Uncommitted changes are preserved only viagit stash push. - Use the exact
go testcommand from Step 3; do not lower-countbelow 6 (statistical significance needs repetitions).
Step 1 — Detect Project
- Find
go.mod(repo root, elseGlob **/go.mod). Missing → Hard Rule 1. - Find benchmarks:
grep -rn "func Benchmark" --include="*.go" .Zero hits → Hard Rule 2.
Done when: go.mod confirmed AND ≥1 benchmark function found (or the command stopped with the exact message).
Step 2 — Parse Arguments
Arguments: $ARGUMENTS
| Input | Result |
|---|---|
| (empty) | PKG=./... (all packages), MODE=single |
<package> |
PKG=<package> (e.g. ./pkg/cache/...), MODE=single |
--compare <ref> |
MODE=compare against <ref>; PKG from remaining arg or ./... |
Done when: PKG and MODE are fixed.
Step 3 — Run Current Benchmarks
go test -bench=. -benchmem -count=6 -timeout=10m <PKG>
- MODE=single: parse output into the results table → go to Step 7.
- MODE=compare: save raw output to
/tmp/bench-new.txt. If this run FAILS, report the error and STOP here (nothing was switched yet, no restore needed).
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.
- today First seen · 131 lines · 9 tokens per session scan A e003884a5b19
benchmark is a command published in the GitHub repository RaNDoM6913/claude-code-superkit (2 stars, last pushed 1mo ago), licensed MIT. It adds 9 tokens to every session and 1,477 once invoked, about $0.0000 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 2026-09-03.
Other commands, from other repositories
copy-ai-context
Export converged workspace context to the clipboard for any AI chat.
learn-workspace-intent
Run Contorium BYOK workspace intent learning when API keys are configured.
setup-mcp-claude-code
Build and register the Contorium MCP server for Claude Code or Cursor Agent.
start-fresh-session
Start a fresh Contorium AI context session after a major task or workspace shift.
setup-mcp-codex
Build Contorium and install the Codex plugin with bundled MCP memory tools.
dmb-resume
Resume DMB context for this session.