perf

A command for investigating a slow or frequently used part of a program and proposing measured improvements. It examines timing, memory allocations, input/output, database queries, loops, and caching where relevant.

In plain words
What is it for?
Use it to analyze a target function or execution path, rank one to three possible changes, estimate their likely impact, and define the smallest benchmark that could confirm the result.
Why use it?
It prevents optimization based only on guesses by asking for baseline measurements and a benchmark. If the code is not actually a hot path, it recommends measuring first.

Command

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/juliusbrussee/caveman-code/perf
Clone the repo
git clone --depth 1 https://github.com/JuliusBrussee/caveman-code
Per session 15 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 270 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.

ModelPer sessionOnce invoked
Fable 5 $0.00015 $0.00270
Opus 5 $0.00008 $0.00135
Sonnet 5 $0.00003 $0.00054
Haiku 4.5 $0.00002 $0.00027

Measured 2d ago against content hash c942797e5eca, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

perf 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 2d 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.

packages/coding-agent/commands/perf.md · 33 lines

What it actually says

Investigate the performance of $ARGUMENTS.

  1. Locate the target function or path. Read its source and the immediate callers.
  2. Identify the measurable hot signal:
    • Wall-clock time per call,
    • Allocations / GC pressure,
    • Sync I/O on hot paths,
    • N+1 queries / quadratic loops,
    • Cache miss rate or hit ratio.
  3. Propose 1–3 concrete changes ranked by expected impact and implementation risk. For each change, give:
    • The exact diff or pseudo-diff.
    • Expected speedup with reasoning (Big-O if applicable).
    • The smallest benchmark that would prove the speedup.
  4. If a benchmark already exists for this path, re-run it and report baseline numbers; if not, sketch how to add one.
  5. Do not ship "premature optimization" — if the path is not actually hot, say so and recommend instrumentation first.

Output as a checklist the user can approve or reject item by item.

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. 2d ago First seen · 33 lines · 15 tokens per session scan A c942797e5eca

Subscribe to this mod's changes

perf is a command published in the GitHub repository JuliusBrussee/caveman-code (929 stars, last pushed 18d ago), licensed MIT. It adds 15 tokens to every session and 270 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.