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/dimetron/pi-go/claude-mdgit clone --depth 1 https://github.com/dimetron/pi-goWrote 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/instructions/dimetron/pi-go/claude-md)<a href="https://agentmods.dev/instructions/dimetron/pi-go/claude-md"><img src="https://agentmods.dev/badge/instructions/dimetron/pi-go/claude-md.svg" alt="Measured on agentmods" 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 | $0.04720 | $0.04720 |
| Opus 5 | $0.02360 | $0.02360 |
| Sonnet 5 | $0.00944 | $0.00944 |
| Haiku 4.5 | $0.00472 | $0.00472 |
Grade A, and why
pi-go CLAUDE.md scanned grade A with 1 finding 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 5d 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
- **Profiling endpoints are on localhost**, so `curl localhost:6060/...` is How it starts
The opening of the file, as written. The whole thing — 406 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md — pi-go
Guidance for coding agents working in this repo. Applies to Claude Code and to pi-go's own agent; where the two differ, both are described.
Work in a git worktree, not the primary checkout
Do not make uncommitted edits in /Users/dimetron/p6s/pi-dev/pi-go and leave
them there. The primary checkout has its branch switched frequently, and
git checkout discards uncommitted changes in tracked files without warning.
Work has been lost to this. A worktree gives each task its own working
directory and its own branch, so a switch in one cannot destroy another.
Create one before starting any task that edits tracked files:
git worktree add -b fix/<topic> .worktrees/fix-<topic> HEAD
cd .worktrees/fix-<topic>
Remove it when the branch is merged or abandoned:
git worktree remove .worktrees/fix-<topic>
git worktree list # verify; prune stale metadata with `git worktree prune`
Where worktrees live
Three conventions coexist. Match the one that fits who is doing the work.
| Creator | Path | Branch | Notes |
|---|---|---|---|
| Human / Claude Code | <repo>/.worktrees/<branch-with-dashes> |
fix/…, feat/… |
Inside the repo, so it stays within the sandbox; .worktrees/ is gitignored |
pi-go agent (/run, subagents) |
<repo>/.pi-go/tasks/<pathID> |
pi-agent-<shortID>, or the sanitized requested name |
Created by internal/subagent/worktree.go; .pi-go/ is gitignored |
arbor tool |
~/.arbor/worktrees/pi-go/<name> |
matches dir name | External tool, listed here only so git worktree list output is not surprising |
.pi-go/ and .worktrees/ are both gitignored, so agent worktrees never show
up as untracked noise in git status.
What pi-go's agent already does, and why it matters
WorktreeManager.Create (internal/subagent/worktree.go:117) stashes
uncommitted changes before git worktree add and pops them afterwards, with
a unique stash message so the pop is deterministic
(stashMessage, popStashByMessage). It does this because worktree add from
HEAD fails on a dirty tree.
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.
- 5d ago First seen · 406 lines · 4,720 tokens per session scan A 88238cac82a2
pi-go CLAUDE.md is an instructions file published in the GitHub repository dimetron/pi-go (151 stars, last pushed yesterday), licensed MIT. It adds 4,720 tokens to every session, about $0.0236 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
awesome-go AGENTS.md
AGENTS.md instructions for avelino/awesome-go, covering awesome-go · llm contribution guide, project snapshot, when modifying the awesome list, coding guidelines and testing & validation.
go-micro CLAUDE.md
Claude Code instructions for micro/go-micro, covering claude.md - go micro project guide, project overview, build & test, run all tests and run tests for a specific package.
azure-sdk-for-go go-examples.instructions.md
Instructions for Azure/azure-sdk-for-go, a project described as: This repository is for active development of the Azure SDK for Go. For consumers of the SDK we recommend visiting our public developer docs at.
chatgpt-cli AGENTS.md
AGENTS.md instructions for kardolus/chatgpt-cli, covering agents.md, 1) repo map (where things usually live), 2) build & test workflow (use makefile first), 3) go version and 4) testing expectations (required).
phi AGENTS.md
Instructions for pulseaiclub/phi, covering phi, communication preferences, constraints, contributor guidelines and commands.
peri CLAUDE.md
Claude Code instructions for KonghaYao/peri, covering claude.md — perihelion, 先读什么, 跨仓库契约, 任务路由 and 代码定位.