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/jazelly/tokenless/agents-mdgit clone --depth 1 https://github.com/jazelly/tokenlessWhat 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 | $0.01910 | $0.01910 |
| Opus 5 | $0.00955 | $0.00955 |
| Sonnet 5 | $0.00382 | $0.00382 |
| Haiku 4.5 | $0.00191 | $0.00191 |
Grade A, and why
tokenless 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 2d 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 — 90 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Repository Development Rules
Language
- Use English for code, identifiers, package/config keys, and inline comments.
- Localize every reasonable user-facing surface in both English and Simplified Chinese. Never ship one language alone or assume users know the other.
- Keep commands, flags, identifiers, protocol fields, provider/model names, and clearer technical terms unchanged.
- Keep language selection and fallback consistent; cover externally visible localization through real boundaries.
- Keep paired docs, such as
README.mdandREADME.zh-CN.md, structurally and semantically aligned. - Internal technical docs may remain English when translation adds no user value.
Documentation style
- Do not ship walls of text. Keep paragraphs to one to three short sentences, and use a list, table, diagram, or screenshot when it makes the same point faster.
- Keep product READMEs as scannable landing pages: lead with the value proposition and product visual, then show providers, quick start, and links to deeper documentation.
- Put detailed reference material in the focused document that owns it and link there instead of duplicating long explanations.
- Prefer one clear sentence over a section, and a few concrete bullets over a paragraph that merely enumerates features or steps.
- Label illustrative or staged screenshots and metrics so they cannot be mistaken for benchmarks or production telemetry.
Tokenless skill
- Do not read or invoke the external
tokenlessskill during repository work. Use this file, source, and tests.
Branches and releases
- Land prerelease work on
devbeforemain; directdevcommits are allowed. - Ticket branches are optional; target their first PR to
dev. - Update
mainonly by PR, normally promoted fromdev; never commit or push directly. - Add a changeset for user-visible CLI changes requiring publication. A
mainPR may omit one only when no release should occur. - Never publish packages, releases, or artifacts manually unless explicitly asked; use release automation.
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.
- 2d ago First seen · 90 lines · 1,910 tokens per session scan A 6cbe104f8383
tokenless AGENTS.md is an instructions file published in the GitHub repository jazelly/tokenless (11 stars, last pushed 2d ago), licensed MIT. It adds 1,910 tokens to every session, about $0.0095 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
autoframe CLAUDE.md
Instructions for kint4/autoframe, covering autoframe — claude instructions, what is autoframe?, stack (non-negotiable), file locations and folder structure.
browserkit AGENTS.md
AGENTS.md instructions for browserkit-dev/browserkit, covering agents.md — workspace memory, project: browserkit, browser lifecycle, browser control and design process preferences.
browserkit CLAUDE.md
Claude Code instructions for browserkit-dev/browserkit, covering browserkit, repo layout, build & test, architecture and external adapter repos (all under browserkit-dev/).
Skill_Seekers CLAUDE.md
Instructions for yusufkaraaslan/Skill_Seekers, covering claude.md, project overview, essential commands, required before running tests or cli (src/ layout) and run all tests (never skip - all must pass before commits).
playwriter AGENTS.md
Instructions for remorses/playwriter, covering backward compatibility, architecture, development, running mcp locally and running cli locally.
stealth-browser-mcp AGENTS.md
Instructions for vibheksoni/stealth-browser-mcp, covering agent instructions, start here, recommended proxy services, environment and validation.