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/visnia-ai/browser-agent/agents-mdgit clone --depth 1 https://github.com/visnia-ai/browser-agentWrote 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/visnia-ai/browser-agent/agents-md)<a href="https://agentmods.dev/instructions/visnia-ai/browser-agent/agents-md"><img src="https://agentmods.dev/badge/instructions/visnia-ai/browser-agent/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 | $0.00529 | $0.00529 |
| Opus 5 | $0.00264 | $0.00264 |
| Sonnet 5 | $0.00106 | $0.00106 |
| Haiku 4.5 | $0.00053 | $0.00053 |
Grade A, and why
browser-agent 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 — 33 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
Architecture
- Keep reusable
src/core/APIs configurable through explicit typed inputs and dependency injection. Do not make core entrypoints depend on YAML files, CLI arguments, or implicit environment configuration. - Keep CLI YAML parsing and validation centralized in
src/utils.ts(loadConfig). - Expose new public core surfaces through
src/core/index.tsandsrc/core/types.ts. - Route provider execution through
src/agents/providers/ai-sdk.ts; do not add provider-specific router implementations. - Put runtime-configurable flags in
src/config-feature-flags.tsand internal static flags insrc/featureFlags.ts.
Model and Projection Contracts
- The harness uses only the canonical semantic projection from
src/browser/semantic-projection.ts; do not add alternate semantic serializers or simplified-DOM fallbacks. - Actions targeting page elements must use opaque
refvalues from the current canonical semantic projection and resolve them through the semantic ref registry. - Keep the standalone simplified-DOM library isolated under
src/browser/simplify-dom.tsandsrc/browser/simplify-dom-utils/. Canonical harness code insrc/agents/,src/core/, andsrc/auth/must not import it or depend on its identifiers. - Changes to semantic projection structure must preserve ref extraction, action resolution, projection-history reconstruction, and prompt contracts.
Authentication Safety
- Treat changes to auth runtime, executor integration, prompts, config parsing, or auth input handling as security-sensitive.
- Never expose real credentials to a model, prompt payload, screenshot, semantic projection, history, logs, thrown errors, or serialized results.
- Match the current URL to a configured auth domain before decrypting its identifier or password. Keep domain, identifier, and password encrypted independently.
- Plaintext credentials are allowed only at direct runtime entrypoints and must be encrypted immediately. YAML configuration must accept encrypted credentials only.
- Models may inspect only the redacted semantic projection to identify auth controls. Credential lookup, entry, and submission must remain in runtime code.
- Keep protected auth refs redacted and suppress screenshots while sensitive values may be present.
- When automated auth cannot proceed safely, return it as unhandled and use manual takeover only when enabled.
- Cover auth behavior changes with unit tests and maintain or extend the local fixture/e2e coverage.
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 · 33 lines · 529 tokens per session scan A 7c77da96f6eb
browser-agent AGENTS.md is an instructions file published in the GitHub repository visnia-ai/browser-agent (21 stars, last pushed 7d ago), licensed MIT. It adds 529 tokens to every session, about $0.0026 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
dev3000 AGENTS.md
Instructions for vercel-labs/dev3000, covering agents.md, runtime, local ui, production workflows, browser tools and development rules.
terminal-browser AGENTS.md
AGENTS.md instructions for zenbu-labs/terminal-browser, a project described as: A browser inside your terminal.
terminal-browser CLAUDE.md
Claude Code instructions for zenbu-labs/terminal-browser, a project described as: A browser inside your terminal.
Browser4 CLAUDE.md
Claude Code instructions for platonai/Browser4, covering browser4 — project context for claude, architecture, key dispatch chain (cli → browser), batch commands and e2e test structure.
cc-websearch CLAUDE.md
Claude Code instructions for Djarvur/cc-websearch, covering project, constraints, technology stack, what not to add and conventions.
sniff AGENTS.md
AGENTS.md instructions for Aboudjem/sniff, covering agents.md: sniff, what this repo is, how an agent should use sniff, handling the playwright setup gate and finding output schema.