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/gvkhosla/pi-tinker/agents-mdgit clone --depth 1 https://github.com/gvkhosla/pi-tinkerWrote 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/gvkhosla/pi-tinker/agents-md)<a href="https://agentmods.dev/instructions/gvkhosla/pi-tinker/agents-md"><img src="https://agentmods.dev/badge/instructions/gvkhosla/pi-tinker/agents-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 | $0.01100 | $0.01100 |
| Opus 5 | $0.00550 | $0.00550 |
| Sonnet 5 | $0.00220 | $0.00220 |
| Haiku 4.5 | $0.00110 | $0.00110 |
Grade A, and why
pi-tinker AGENTS.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 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.
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 — 116 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agent guide for pi-tinker
This repo is a Pi package that helps people run and fine-tune Inkling and other open-weight models with Tinker.
Product promise
Help a non-expert go from:
data → prepared JSONL → validation → baseline eval → smoke training → checkpoint → before/after comparison → chat in Pi
Do not turn this into a separate training framework. Keep Tinker/Tinker Cookbook as the real training layer and generate normal editable Python.
Agent compatibility
This AGENTS.md is the canonical guide for every coding agent. Pi is the primary and richest interface, but Claude Code, Codex, Cursor, Copilot, Gemini CLI, and other shell-capable agents can run the same operator non-interactively:
node scripts/agent-cli.mjs inkling
node scripts/agent-cli.mjs doctor
node scripts/agent-cli.mjs validate data/train.jsonl --model thinkingmachines/Inkling-Small
When installed from npm, use pi-tinker-agent instead. Read docs/coding-agents.md for details. Other agents should inspect and edit the generated Python directly; they must not imitate hidden Pi UI state or trigger API-using stages without explicit user approval. --force is file-only; never add --accept-regression, --allow-retired-model, or --allow-unapproved unless the user explicitly requests that exact safety override.
Best first commands for users
For Inkling itself:
/tinker inkling
/tinker inkling sweep --prompt "a representative task" --efforts low,medium,high,xhigh --yes
/model
If the user has no data yet:
/tinker demo
/tinker next
If the user has CSV/JSON/JSONL/docs, prefer the managed operator first:
/tinker improve <input> --goal "what should improve" --budget demo
Then, only after the user understands API usage and evals:
/tinker improve <input> --goal "what should improve" --budget smoke --eval-reviewed --yes
/tinker improve <input> --goal "what should improve" --budget small --yes
/tinker deploy latest
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 · 116 lines · 1,100 tokens per session scan A dc59af148e9a
pi-tinker AGENTS.md is an instructions file published in the GitHub repository gvkhosla/pi-tinker (24 stars, last pushed 7d ago), licensed Apache-2.0. It adds 1,100 tokens to every session, about $0.0055 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 instructions, from other repositories
axolotl AGENTS.md
AGENTS.md instructions for axolotl-ai-cloud/axolotl, covering axolotl, tech stack, commands, training methods and config pattern.
tuning-engines-cli CLAUDE.md
Instructions for cerebrixos-org/tuning-engines-cli, covering tuning engines mcp server, setup, claude desktop configuration, claude code and available tools.
Eval CLAUDE.md
Instructions for ai-twinkle/Eval, covering twinkle eval — 專案規範手冊（claude.md）, 目錄, 1. 專案定位與設計理念, 1.1 誕生背景 and 1.2 核心設計哲學.
llm-ide-rules react.instructions.md
Instructions for iloveitaly/llm-ide-rules, covering react, mock data, react hook form and styling.
quant.cpp CLAUDE.md
Claude Code instructions for quantumaikr/quant.cpp, covering quant.cpp — agent development guide, project vision, project overview, architecture and key documents.
The-Ideal-Harness AGENTS.md
AGENTS.md instructions for bharat3645/The-Ideal-Harness, covering agents.md, read this before you write code, 1. enforce below the model, 2. zero overlap and 3. clean-room.