Borrowing it
Nothing to install: this file belongs to smithersai/smithers. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/smithersai/smithers/main/AGENTS.mdgit clone --depth 1 https://github.com/smithersai/smithersWrote 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/smithersai/smithers/agents-md)<a href="https://agentmods.dev/instructions/smithersai/smithers/agents-md"><img src="https://agentmods.dev/badge/instructions/smithersai/smithers/agents-md/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/instructions/smithersai/smithers/agents-md"><img src="https://agentmods.dev/badge/instructions/smithersai/smithers/agents-md.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 | $0.05089 | $0.05089 |
| Opus 5.5 | $0.02036 | $0.02036 |
| Sonnet 5.5 | $0.01018 | $0.01018 |
| Haiku 4.5 | $0.00509 | $0.00509 |
Grade A, and why
smithers 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 today.
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 — 346 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Permanent product interaction rules
Everything is a durable flow (Will, 2026-10-19, permanent)
Smithers is built on its own durable flows (@smthrs/flow). Every action the
app or an agent takes runs as a durable flow step, and that includes every model
call. Flows run on the user's machine, on Cloudflare, or on our sandbox
provider.
- Durable by default. The flow engine supplies retries, resumption after a crash, idempotency and deadlines. App and library code never write their own retry loops, background jobs, sweeps, queues or "re-dispatch" logic.
- Model calls go through the durable layer only. That means a flow step,
through
@smthrs/agentor@smthrs/flow. No handler or app module callsModel.streamdirectly. - Never substitute another runtime for one. No Cloudflare Workflows,
Durable Object job loops, bare
waitUntilwork, Temporal, Inngest, Trigger.dev, Vercel AI SDK, LangChain or LangGraph, or Mastra. If the flow engine can't run where a job must run, the fix is a host for it in the SDK (@smthrs/platform-node,-bun,-browser, and the Cloudflare and sandbox hosts). A workaround in an app is not a fix. - Before choosing any library, find the job in the library catalog, and say in the change which entry you used. A brief written for another agent states this rule.
Doneish first; release continuously (Will, 2026-10-08)
Make it work, then make it good. The goal is doneish, and every doneish
main ships as a silent prerelease. Where another rule here or in .specs
sets a stricter bar or a different order, this section wins until the
prerelease is out.
- Doneish: to the best of our knowledge the spec is implemented. On one
maincommit a real install on an ordinary Mac takes a TODO to a merged pull request, the J1 to J11 journey suites pass, and every ticket is either believed complete or listed inPRERELEASE.mdas off, broken, missing or unverified. A suite that runs on fixtures is not the real install. It may be unverified on the reference host and unreviewed. - Done is unchanged: owner sign-off and a closed issue. Done gates the real release, never a prerelease. That work continues after each prerelease.
- Nothing strict blocks doneish: no wiped or reference Mac mini, independent
operator, receipt,
Ready:restamp, owner pre-review, second-model review or coverage campaign. Each becomes a line under Unverified. - Security hard rules and contributor trust rules still hold. A feature with a known hole ships switched off and listed.
- Work in this order: the real install run, failing journey rows, the release path, then the rest. Work that serves none of them waits for the prerelease.
- A prerelease is a suffixed version such as
1.0.0-rc.2, published under npmnextwith no announcement;latestdoes not move. Only gates that prove it builds, installs and starts can block it; every other gate reports. Cut one whenever the journeys pass, at least daily, so Will sees the product instead of waiting on it.
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.
- today Changed · +40 lines · +593 tokens per session c3bb18f1c795
- 3d ago Changed · -20 lines · -246 tokens per session ac88587631b6
- 4d ago Changed · +34 lines · +539 tokens per session 2ec5b81df349
- 7d ago Changed · +15 lines · +188 tokens per session 439b1046a0c6
- 12d ago Changed · +33 lines · +478 tokens per session 7375a2299f5e
- 16d ago Changed · -1 lines · +3 tokens per session 83fc9f0e6e05
- 17d ago Changed · +19 lines · +270 tokens per session bc4b1dd2acfa
- 18d ago Changed · +73 lines · +1,191 tokens per session 8e0a012cf929
- 20d ago Changed · +17 lines · +158 tokens per session 7802dc4b7c51
- 21d ago First seen · 136 lines · 1,915 tokens per session scan A af8eb7fc0543
smithers AGENTS.md is an instructions file published in the GitHub repository smithersai/smithers (439 stars, last pushed today), licensed MIT. It adds 5,089 tokens to every session, about $0.0204 per session on Opus 5.5. 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 2026-09-03.
Other instructions, from other repositories
vscode buildNext.instructions.md
Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).
next.js AGENTS.md
AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.
vscode oss-third-party-notices.instructions.md
Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, cache eligibility and rollout and applying the notice (cutover).
langchain AGENTS.md
AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.
vscode copilot-instructions.md
Copilot instructions for microsoft/vscode, covering vs code copilot instructions, project overview, root folders, core architecture (src/ folder) and built-in extensions (extensions/ folder).
deepseek-harness AGENTS.md
AGENTS.md instructions for deepseek-ai/deepseek-harness, covering agents.md, pre-stable apis and released session data, repository layout, commands and host sandbox failures.