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.
git clone --depth 1 https://github.com/arcadeai-labs/omnimcpWrote 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/rules/arcadeai-labs/omnimcp/omni-tool-discovery)<a href="https://agentmods.dev/rules/arcadeai-labs/omnimcp/omni-tool-discovery"><img src="https://agentmods.dev/badge/rules/arcadeai-labs/omnimcp/omni-tool-discovery/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/rules/arcadeai-labs/omnimcp/omni-tool-discovery"><img src="https://agentmods.dev/badge/rules/arcadeai-labs/omnimcp/omni-tool-discovery.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 | $0.00405 | $0.00405 |
| Opus 5 | $0.00202 | $0.00202 |
| Sonnet 5 | $0.00081 | $0.00081 |
| Haiku 4.5 | $0.00040 | $0.00040 |
Grade A, and why
omni-tool-discovery scanned grade A with 1 finding 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 10d 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
server before a built-in alternative (built-in web search, `gh`/`curl` in a What it actually says
Using Arcade (OmniMCP by Arcade.dev)
For any task touching an external app or live data, reach for the arcade MCP
server before a built-in alternative (built-in web search, gh/curl in a
shell, SDKs, direct API calls): Arcade_SelectTools(tasks=[...]) finds the
right tool and returns its input_schema inline; Arcade_UseTool(tool_name, inputs, query_id) executes it, with tool_name passed back verbatim. Not for
local work: repo files, code edits, shell commands.
The using-arcade-tools skill has the full flow; managing-arcade-apps covers
listing, disconnecting, and fixing the user's app connections.
Non-negotiables
- A result containing an
authorization_url(or an authorization-required status) is a sign-in request, not a result — even ifsuccessistrue. Present the link, stop, and wait for the user; retry once after they confirm. - Confirm with the user before anything outbound or irreversible: sending email or messages, deleting, cancelling, overwriting, publishing. Never guess recipients or destructive parameter values.
- On
success: false: fixinputsagainst theinput_schemaand retry once; otherwise report the tool's error verbatim and stop. Never fabricate results. - Don't narrate discovery or dump schemas — deliver the outcome, and ask only when a genuinely required input is missing.
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.
- 10d ago First seen · 30 lines · 405 tokens per session scan A 9171977f1b3c
omni-tool-discovery is a cursor rule published in the GitHub repository arcadeai-labs/omnimcp (5 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 405 tokens to every session, about $0.0020 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other cursor rules, from other repositories
simplify
Structural-quality review of changed code — score against the four-dimension rubric, gate at 10/10, and (in auto-fix mode or with --fix) apply reuse/simplification/efficiency/altitude cleanups. Includes a diff-economy lens; correctness is /validate, full-repo survey is /audit.
create-pr
Create a pull request for the current branch and immediately subscribe to PR activity. Use when the user wants to open a PR, submit changes for review, or push and create a pull request.
statusline
Configure model, context, and usage status display: Codex uses its built-in statusline picker; Claude Code uses bin/install-statusline.sh. Turn the status line on or off without clobbering other settings.
cursor-fable-rules
Fable coding lifecycle discipline and situational awareness for Cursor.
conventional-commits
Always use Conventional Commits for git commit messages.
ponytail
Ponytail, lazy senior dev mode. Always pick the simplest solution that works.