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/michaelolmos/tsaagan/agents-mdgit clone --depth 1 https://github.com/michaelolmos/tsaaganWrote 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/michaelolmos/tsaagan/agents-md)<a href="https://agentmods.dev/instructions/michaelolmos/tsaagan/agents-md"><img src="https://agentmods.dev/badge/instructions/michaelolmos/tsaagan/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 | $0.01227 | $0.01227 |
| Opus 5 | $0.00613 | $0.00613 |
| Sonnet 5 | $0.00245 | $0.00245 |
| Haiku 4.5 | $0.00123 | $0.00123 |
Grade A, and why
tsaagan 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 3d 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.
AGENTS.md — how to drive Tsaagan
This file tells any AI agent (Claude Code, Codex/ChatGPT CLI, Cursor, etc.) how to use Tsaagan well. Read SOUL.md for why; this is how.
tsaagan = node tsaagan.js. Every command prints JSON. Mutating actions return a
verify block — read it; don't assume success.
Pick the lightest layer that does the job
- API layer first. If the site has an official API, use it — fastest, robust,
no UI to drive at all.
tsaagan api detect=<url>→ is there a known API?tsaagan api service=<x> path=/… method=…→ call it with a stored key.- First time: store the key —
tsaagan vault set service=<x> secret=…(Keychain). Tsaagan can even browse the site's developer settings to create the key, thenvault setit (see docs/API.md).
- Real-browser layer for sites without an API or behind a login:
extensionmode (any OS) — trusted input, viewport coords, no debug port. Best for sites that ignore synthetic input. See docs/EXTENSION.md.nativemode (macOS) — real Chrome via AppleScript + OS input.
- CDP layer (
fresh/clone/live) — Playwright; great for the broad middle of the web (dashboards, SaaS). Note: CDP input isisTrusted=false, so a minority of strict sites ignore it — use native/extension there.
The loop (every browser step)
OBSERVE → THINK → ACT → VERIFY → RECOVER
tsaagan snapshot— read the accessibility tree; act on stable[ref=eN].- Pick one goal; ≤3 actions before re-observing (the page changes after action 1).
- Act:
tsaagan click ref=e5 expectText="…",tsaagan type ref=e3 text="…" submit=true. Always passexpectText=/expectGone=so the step self-verifies. - Read the
verifyblock (urlChanged,expectTextFound,newConsoleErrors,failedRequests). That's how you know it worked. - On
ok:false: re-snapshot, retry with a fresh ref, or change approach. Ref actions self-heal once automatically.
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.
- 3d ago First seen · 90 lines · 1,227 tokens per session scan A 3ca4de1ac03a
tsaagan AGENTS.md is an instructions file published in the GitHub repository michaelolmos/tsaagan (2 stars, last pushed 17d ago), licensed MIT. It adds 1,227 tokens to every session, about $0.0061 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
browser-harness AGENTS.md
Instructions for browser-use/browser-harness, covering code priorities and contributing.
browser-harness CLAUDE.md
Instructions for browser-use/browser-harness, a project described as: Browser Harness | Self-healing harness that enables LLMs to complete any task.
BrowserMesh CLAUDE.md
Claude Code instructions for scrollDynasty/BrowserMesh, covering browsermesh, responsibility boundary, architecture invariants, source of truth and commands.
aiope AGENTS.md
AGENTS.md instructions for XNet-NGO/aiope, covering aiope development loop, environment, build server (192.168.1.2), repos and aiope monorepo structure.
periscope-mcp CLAUDE.md
Claude Code instructions for segentic-lab/periscope-mcp, covering periscope-mcp - project guide, what this is, how to run, key files and adding a new mcp tool.
browser-use CLAUDE.md
Claude Code instructions for browser-use/browser-use, covering claude.md, high-level architecture, core components, event-driven browser management and cdp integration.