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 skills add PrabhdeepSingh/claude-plugins --skill performancegit clone --depth 1 https://github.com/PrabhdeepSingh/claude-pluginsWrote 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/skills/prabhdeepsingh/claude-plugins/performance)<a href="https://agentmods.dev/skills/prabhdeepsingh/claude-plugins/performance"><img src="https://agentmods.dev/badge/skills/prabhdeepsingh/claude-plugins/performance/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.
<a href="https://agentmods.dev/skills/prabhdeepsingh/claude-plugins/performance"><img src="https://agentmods.dev/badge/skills/prabhdeepsingh/claude-plugins/performance.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.00099 | $0.01486 |
| Opus 5.5 | $0.00040 | $0.00594 |
| Sonnet 5.5 | $0.00020 | $0.00297 |
| Haiku 4.5 | $0.00010 | $0.00149 |
Grade A, and why
performance 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 1mo 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.
How it starts
The opening of the file, as written. The whole thing — 80 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Performance — the measurement is the work
Most "optimizations" are guesses wearing confidence: a change lands because it should be faster, nobody measures, and the codebase accretes complexity that never bought anything. The discipline that prevents this is not knowing the tricks — it's the loop around them: measure a baseline, change one thing, re-measure the same way, and keep only what provably paid. Code you keep, you maintain forever; make it pay for itself.
How to apply this
Run the loop in order: baseline → identify the bottleneck → one change → verify against the baseline → keep or revert → record the attempt. When invoked directly as /sonu:performance, apply it to $ARGUMENTS — the text typed after the invocation; if that token appears literally or is empty, apply it to the performance concern in the current discussion.
1. Baseline before touching anything
No baseline, no optimization — without a starting number, "faster" is a feeling. Capture the metric that matters to the user (page-load milestones, interaction latency, API p95, job duration, bundle bytes — whatever the complaint names), under stated conditions (dataset size, cache state, hardware/environment), with a fixed budget (sample count, wall-clock, or request count). Write the number down; the verdict in §4 is computed against it.
Profile, don't deduce. The bottleneck is where the time measured goes, not where the code looks slow — profilers, query plans, and waterfall traces exist because intuition about hot paths is reliably wrong. Route by symptom first: slow first render → network and render path; slow interaction → main-thread work; slow API → the server and its queries; then profile inside that region.
2. One change at a time
Land one optimization per measurement cycle. Three optimizations measured together produce one number that can't be attributed — if the total improved, you may be keeping two regressions paid for by one win ([[debugging]]'s one-change rule, applied to speed). Small, separately-verified changes also revert cleanly when §4 says revert.
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.
- 1mo ago First seen · 80 lines · 99 tokens per session scan A 8a6469143afc
performance is a skill published in the GitHub repository PrabhdeepSingh/claude-plugins (4 stars, last pushed 1mo ago), licensed MIT. It adds 99 tokens to every session and 1,486 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 2026-08-31.
Other skills, from other repositories
tg-account
A skill for using Telegram as the account owner through a local Telegram Desktop session. It can read chats and export messages, while writing actions require showing the target and exact text for confirmation.
unity-coder
Use when implementing Unity C# code to follow proper coding guidelines, naming conventions, member ordering, and Unity-specific patterns.
agent-teams
Set up and orchestrate Claude Code agent teams - multiple Claude Code instances working together with shared tasks, inter-agent messaging, and centralized management. Use when the user wants to (1) create or start an agent team for parallel work, (2) configure agent team settings like display mode or teammate mode or…
python-coder
Implements Python code following PEP 8, type hints, and project conventions. Use when implementing features designed by python-architect.
python-testing
Python testing patterns with pytest including unit tests, integration tests, fixtures, mocking, and coverage. Use when writing Python tests.
typescript-coder
Implements TypeScript/JavaScript code following established architecture and coding guidelines. Use when implementing features designed by typescript-architect.