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 skills/tslateman/duet/performancenpx skills add tslateman/duet --skill performancegit clone --depth 1 https://github.com/tslateman/duetWhat 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 | $0.00050 | $0.01214 |
| Opus 5 | $0.00025 | $0.00607 |
| Sonnet 5 | $0.00010 | $0.00243 |
| Haiku 4.5 | $0.00005 | $0.00121 |
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 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.
How it starts
The opening of the file, as written. The whole thing — 151 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Performance as Measurement
Overview
Optimize what you've measured, not what you suspect. Performance work without profiling is superstition. Measure first, hypothesize second, optimize third, measure again.
The Performance Loop
- Define the goal, What metric matters? Latency, throughput, memory, startup time?
- Measure the baseline, Quantify current performance with reproducible benchmarks
- Profile, Identify where time and resources actually go
- Hypothesize, What change would improve the bottleneck?
- Optimize, Make one change
- Measure again, Did it help? By how much? Any regressions elsewhere?
Never skip from step 1 to step 5.
For concrete profiler commands and a step-by-step run of this loop per language, see references/profiling-checklist.md.
Trade-off Framework
Every optimization trades one resource for another. Make the trade explicit.
| Trade-off | Example |
|---|---|
| Latency vs. throughput | Batching increases throughput, raises individual latency |
| Memory vs. CPU | Caching trades memory for fewer computations |
| Simplicity vs. speed | Hand-rolled loops beat abstractions but obscure intent |
| Startup vs. runtime | Lazy loading delays startup cost to first use |
| Bandwidth vs. latency | Compression saves bandwidth, costs CPU time |
| Consistency vs. speed | Eventual consistency is faster than strong consistency |
Ask: "Which resource is scarce in this context?" Optimize for the scarce one.
Profiling Strategy
Where to Look
Start with the outermost measurement, narrow inward:
- End-to-end timing, Total wall-clock time for the operation
- Component breakdown, Which phase takes the most time?
- Hot path analysis, Which functions dominate the profile?
- Allocation analysis, Where is memory allocated and freed?
What ships with it
2 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 2d ago First seen · 151 lines · 50 tokens per session scan A 0be2becab2c0
performance is a skill published in the GitHub repository tslateman/duet (1 stars, last pushed 5d ago), licensed MIT. It adds 50 tokens to every session and 1,214 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 2026-08-31.
Other skills, from other repositories
aip-tracker
Track Airflow Improvement Proposal (AIP) implementation progress by comparing Confluence specs against codebase evidence. Use when asked to assess, report on, or compare AIP status.
html-ppt-hermes-cyber-terminal
OpenDesign + BYOK: choosing and wiring your own model, hands-on — cost, quality, and the routing decision. Built as a decision-grade AI literacy deck for engineers, IT, applied-AI teams.
development
开发语言能力索引。Python、Go、Rust、TypeScript、Java、C++、Shell。当用户提到编程、开发、代码、语言时路由到此。.
devops
DevOps 能力索引。Git、测试、DevSecOps、数据库。当用户提到 DevOps、CI/CD、Git、测试时路由到此。.
post-build-flow
Handles workflow verification and setup after build-workflow succeeds, or when the message contains workflow-verification-follow-up or workflow-setup-required. Load after direct builds, when verificationReadiness requires action, or on orchestrator verify/setup follow-up turns.
n8n:nathan
Deploy a temporary n8n test instance (or generate a local docker run command) via the internal "Nathan" bot, from the repo instead of Slack. Use after opening a PR to offer the user a live test instance, or whenever someone asks to spin up / deploy a test instance for a branch.