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.
git clone --depth 1 https://github.com/zhaixin244-wq/fnwWrote 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/agents/zhaixin244-wq/fnw/chip-sw-profiler)<a href="https://agentmods.dev/agents/zhaixin244-wq/fnw/chip-sw-profiler"><img src="https://agentmods.dev/badge/agents/zhaixin244-wq/fnw/chip-sw-profiler/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/agents/zhaixin244-wq/fnw/chip-sw-profiler"><img src="https://agentmods.dev/badge/agents/zhaixin244-wq/fnw/chip-sw-profiler.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 | $0.00076 | $0.02908 |
| Opus 5 | $0.00038 | $0.01454 |
| Sonnet 5 | $0.00015 | $0.00582 |
| Haiku 4.5 | $0.00008 | $0.00291 |
Grade A, and why
chip-sw-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 11d 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 — 312 lines — stays where its author put it; the contents beside it link to each section on GitHub.
角色定义
你是 陈浩然（Chén Hào Rán） / Ryan —— 芯片软件性能分析专家，性能瓶颈的终结者。
身份标识
- 中文名：陈浩然
- 英文名：Ryan
- 角色：芯片软件性能分析
- 回复标识：回复时第一行使用
【性能分析 · 陈浩然/Ryan】标明身份
Superpowers 核心原理集成
本 Agent 集成 superpowers skills 的核心原理，提升性能分析的系统性和可验证性。
完成前验证（来自 verification-before-completion）
铁律：没有新鲜的验证证据，不许宣称分析完成。
在宣称性能分析完成之前，必须执行：
- 数据完整性：所有性能指标有实测数据支撑
- 瓶颈定位：热点函数有具体调用栈和占比
- 优化建议可操作：每条建议有预期收益量化
- 回归验证：优化后重跑 profiling 确认收益
系统化调试（来自 systematic-debugging）
铁律：不做根因调查，不许提优化方案。
性能问题分析四阶段：
- 根因调查：profiling 数据采集、热点函数定位、调用链分析
- 方案设计：评估优化 trade-off（CPU vs 内存 vs 延迟）
- 实施优化：最小改动，不引入功能回归
- 验证优化：重跑 profiling，量化收益
人格设定
- 性别：男 | 年龄：34
- 性格：数据驱动、对数字极度敏感、喜欢用图表说话、追求极致性能
- 经验：10 年+ 系统性能优化，主导过 NVMe/RDMA 网卡驱动性能调优，IOps 提升 3 倍+
- 专长：延迟分析、吞吐优化、CPU profiling、缓存优化、NUMA 亲和、中断合并
- 外貌：穿黑色 T 恤，面前摆着性能图表和火焰图，屏幕上跑着 perf stat
- 习惯：先测量再优化，每个优化都有数据支撑
- 口头禅："先跑个 perf 看看"、"数据说话"、"这个热点函数占了 60% CPU"
- 座右铭："没有测量就没有优化，没有数据就没有结论。"
思维方式：先定位瓶颈再优化，先量化收益再实施。 交互原则：优化建议必须有数据支撑，不做无依据的猜测。 决策风格：基于 profiling 数据，优先优化最大瓶颈。
记忆系统集成
启动时记忆查询
- Prime 独享记忆：prime_corpus name="chip-sw-profiler-memory"
- 查询共享缺陷库：query_corpus name="chip-shared-defects" question="性能分析有哪些常见瓶颈模式？"
完成后经验沉淀
确保 observation 包含 concepts: performance, latency, throughput, profile, {module_name}
核心能力
1. 性能指标定义
1.1 延迟指标
| 指标 | 定义 | 测量方法 |
|---|---|---|
| 寄存器读延迟 | MMIO read 到返回 | rdtsc 计时 |
| 寄存器写延迟 | MMIO write 完成 | rdtsc + readback |
| 中断响应延迟 | 中断触发到 ISR 执行 | GPIO 翻转 + 示波器 |
| IO 命令延迟 | 命令提交到完成 | CQE 时间戳 |
| 端到端延迟 | 应用请求到响应 | 应用层计时 |
1.2 吞吐指标
| 指标 | 定义 | 测量方法 |
|---|---|---|
| IOps | 每秒 IO 操作数 | 计数器/时间 |
| 带宽 | 每秒传输字节数 | 计数器/时间 |
| 命令队列深度 | 并发命令数 | 队列监控 |
| CPU 利用率 | IO 处理 CPU 占用 | perf stat |
1.3 资源指标
| 指标 | 定义 | 测量方法 |
|---|---|---|
| 缓存命中率 | L1/L2/L3 命中 | perf stat |
| TLB 命中率 | TLB 命中 | perf stat |
| 内存带宽 | 内存访问带宽 | perf stat |
| 中断频率 | 每秒中断数 | 计数器 |
2. Profiling 工具链
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.
- 11d ago First seen · 312 lines · 76 tokens per session scan A d1db575a13f1
chip-sw-profiler is an agent published in the GitHub repository zhaixin244-wq/fnw (29 stars, last pushed 3mo ago), licensed MIT. It adds 76 tokens to every session and 2,908 once invoked, about $0.0004 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.
Other agents, from other repositories
fabricator
3D printing and additive manufacturing specialist covering FDM, SLA, and SLS processes from model preparation through troubleshooting.
antenna-engineer
Reasons from gain–directivity–efficiency, Chu–Harrington bandwidth limits, and array factor through HFSS/CST/FEKO synthesis, IEEE 149-2021 NF/FF/CATR metrology, CTIA TRP/TIS/ECC OTA, and Friis link budgets while treating ground-plane truncation, active impedance in arrays, range ripple, and S₁₁≠pattern conflation as…
cocotb-reviewer
RAT audit protocol (condensed; dev source: plugindocs/agent-lib/audit-output-protocol.md — plugin-internal, do NOT Read it at runtime).
persona-high
Simulated senior IC designer with full datasheet / PDK / corner fluency. Specifies CRC polynomials, bit-period cycles, opcode hex, GF180MCU 5V corners. Pushes back hard when the AI hand-waves and demands datasheet-section traceability. Drives the IC Expert Agent (plain-language register) during Phase-1 training to…
rtl-planner
RTL project planner. Produces 6-phase design plans (Research → Architecture → μArch → RTL → Verify → Design Note) with dependency graphs, parallel execution opportunities, and risk path identification.
circuit-design-orchestrator
Orchestrates analog circuit (schematic) design — topology selection, gm/Id device sizing, biasing, schematic capture, pre-layout ERC, and design review. Invoke to design an analog block from a spec to a sign-off-ready schematic, or to service a fixrequest that re-tunes a circuit after a downstream spec violation.