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 instructions/proxyfan/proxyfan/networkinggit clone --depth 1 https://github.com/Proxyfan/ProxyfanWhat 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 | $0.01645 | $0.01645 |
| Opus 5 | $0.00822 | $0.00822 |
| Sonnet 5 | $0.00329 | $0.00329 |
| Haiku 4.5 | $0.00164 | $0.00164 |
Grade A, and why
Proxyfan networking.instructions.md 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 yesterday.
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 — 159 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Networking and proxy-pipeline rules
The proxy engine is the highest-throughput, lowest-latency surface in the
codebase. Bugs here corrupt user traffic, leak memory at line rate, or stall
every captured connection. Every change in Framework.Networking or
Domain.Proxy is reviewed against this file.
Performance budget
The end-to-end overhead added by the proxy must remain under 1 ms per request (excluding rule evaluation and throttling). To stay inside that budget:
- Build on
System.IO.Pipelinesfor byte-stream I/O. Do not allocate a per-framebyte[]; useArrayPool<byte>.Sharedor the pipe's own buffer. - Use
Span<T>/ReadOnlySpan<T>for parsing. Avoid materialising astringfrom a header line until the rule pipeline actually inspects it. - Use
ValueTaskfor hot async paths that frequently complete synchronously (header parsing, frame reads, throttle waits). - Pool large buffers (
MemoryPool<byte>) for body relay; never allocate the full response body up front.
CancellationToken propagation
- Every async method in
Framework.Networkingaccepts aCancellationTokenas the last parameter (ATXTA008). The token from the proxy listener flows all the way down to socket reads, TLS handshakes, and HPACK parsing. - Never create a fresh
CancellationTokeninside a handler — link to the caller's token withCancellationTokenSource.CreateLinkedTokenSourceif a local timeout is needed. - Tokens never carry default values (
ATXCS005).
Sync-over-async is banned
.Result, .Wait(), .GetAwaiter().GetResult() on a Task are forbidden in
the proxy pipeline. They will deadlock under the synchronisation context of
the connection dispatcher and stall every captured connection on that listener.
async void is banned
The only legitimate async void is an Avalonia event handler in
Presentation code-behind. Networking code is never UI code; use
async Task and propagate exceptions through the result type.
Connection dispatch
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.
- yesterday First seen · 159 lines · 1,645 tokens per session scan A 5d439e5843a9
Proxyfan networking.instructions.md is an instructions file published in the GitHub repository Proxyfan/Proxyfan (2 stars, last pushed 2mo ago), licensed MIT. It adds 1,645 tokens to every session, about $0.0082 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 instructions, from other repositories
bifrost AGENTS.md
Instructions for bifrost-proxy/bifrost, covering bifrost 项目开发规则, 执行心法, 执行模式与完成定义, 任务模式判定 and 变更类型与验证路由(先判定,后执行).
quotio AGENTS.md
Instructions for nguyenphutrong/quotio, covering agents.md, stack, where to look, validate changes and project rules.
claude-tap AGENTS.md
Instructions for liaohch3/claude-tap, covering maintainer automation notes, agents 索引, documentation boundary, review guidelines and pre-commit hook.
telepresence AGENTS.md
Instructions for telepresenceio/telepresence, covering agents.md, project overview, git workflow, design plans and build artifacts.
GoModel CLAUDE.md
Instructions for ENTERPILOT/GoModel, covering core principles, follow postel’s law, follow the twelve-factor app, keep it simple and use good defaults.
GoModel AGENTS.md
Instructions for ENTERPILOT/GoModel, covering core principles, commit format — use conventional commits, code review and pr suggestion for the official repository.