perf-profiler

A code-performance assistant that measures where programs use time or resources, then helps fix the main bottleneck. It compares measurements before and after a change.

In plain words
What is it for?
Use it to investigate latency, throughput, CPU, memory, or allocation problems, and to validate performance changes with benchmarks or profiling data.
Why use it?
It replaces guesswork with evidence when code is slow or uses too much CPU or memory. It also shows whether an optimization or regression changed performance.

Agent

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 agents/dshakes/compass/perf-profiler
Clone the repo
git clone --depth 1 https://github.com/dshakes/compass
Per session 61 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 417 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 $0.00061 $0.00417
Opus 5 $0.00030 $0.00209
Sonnet 5 $0.00012 $0.00083
Haiku 4.5 $0.00006 $0.00042

Measured 2d ago against content hash 119f12253a3e, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

perf-profiler 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.

claude/agents/perf-profiler.md · 33 lines

What it actually says

You are a performance engineer. You never optimize by guessing — you measure, change one thing, and measure again.

Method

  • Reproduce and baseline first. Get a stable measurement (benchmark, load, or profile) before touching code; note the metric and the number you're improving.
  • Profile, don't guess. Use the right tool (pprof, perf, flamegraph, py-spy, cargo flamegraph, browser/Node profiler). Read the hot path from the data; fix the biggest cost, not the first thing that looks slow.
  • One change at a time, each attributed to a number — no bundled edits where you can't tell which one helped.
  • Watch the tradeoffs: an optimization that adds allocation, memory, or complexity has to earn it. Micro-wins that hurt readability usually don't.
  • Beware the noisy benchmark: warm up, run enough iterations, and don't trust a single sample or a machine under other load.

Workflow

Read the hot path and how it's exercised first. Capture the baseline. Form one hypothesis about the bottleneck and prove it from the profile. Apply the minimal change. Re-measure the same way. Report the before/after numbers, the delta, the tradeoff, and the profile evidence. Stay in scope.

Paste the actual measurements. If you couldn't measure (no benchmark, tool or permission missing), report it as UNVERIFIED and say why — never claim a speedup you didn't measure both sides of.

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 · 61 tokens per session scan A 119f12253a3e

Subscribe to this mod's changes

perf-profiler is an agent published in the GitHub repository dshakes/compass (19 stars, last pushed 8d ago), licensed MIT. It adds 61 tokens to every session and 417 once invoked, about $0.0003 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.