control-tools

A built-in set of typed tools that coding agents can use through MCP, a standard way for AI agents to call software functions. It provides Rudder's own commands and can also proxy approved external MCP connections.

In plain words
What is it for?
Use it to expose Rudder commands to supported agent runtimes, validate tool inputs, run built-in capabilities, and connect enabled external MCP services within a specific run.
Why use it?
It gives agents clearly defined inputs and keeps built-in tools, external connections, credentials, and identities separated.

Agent

Install

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.

agentmods
npx agentmods add agents/undertone0809/rudder/control-tools
Clone the repo
git clone --depth 1 https://github.com/Undertone0809/rudder
Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 5,561 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00000 $0.05561
Opus 5 $0.00000 $0.02780
Sonnet 5 $0.00000 $0.01112
Haiku 4.5 $0.00000 $0.00556

Measured 2d ago against content hash 977278764cc0, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

control-tools 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.

doc/product/domains/agents/control-tools.md · 410 lines

How it starts

The opening of the file, as written. The whole thing — 410 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Built-in Rudder Tools

AGENT.CONTROL.TOOLS.001

Contract Summary

Rudder exposes a first-party rudder-tools MCP server for supported agent runtimes. The server presents the stable agent-v1 command contract as typed MCP tools using rudder_<capability_id> names. Each tool publishes an exact capability-specific input schema with required fields, accepted types, bounded sizes, and no additional properties. Rudder conditionally projects the Built-in Browser tool set from trusted runtime capability state, dispatches built-in Rudder tools directly through Rudder's runtime API context when supported, falls back to the existing Rudder CLI command path for remaining capabilities, and gets organization, agent, run, API, and project-library identity only from runtime-owned environment.

Organization-managed external MCP servers are a separate runtime surface. Rudder projects each enabled external connection as an independent run-scoped proxy through the provider-neutral managedExternalMcpBindings contract. These proxies do not add tools to, rename, replace, share credentials with, or weaken the identity boundary of the first-party rudder-tools server.

Intent / User Job

Runtime agents need a reliable Rudder tool surface for issue, run, chat, automation, library, approval, skill, agent, and bounded Browser operations. Operators need this surface to be typed, auditable, and scoped to the current run instead of relying on model-invented shell commands or user/home MCP configuration.

Why / Design Reasoning

The bundled rudder-docs skill remains the compatibility reference for exact CLI command details, but agent runtimes that support MCP should prefer first-party Rudder tools. The skill is always discoverable; that does not mean an agent must read it on every run. MCP gives the model a typed schema, stable tool names, and a clear transport boundary while preserving CLI compatibility for capabilities that have not moved to direct runtime API dispatch.

Read the full file on GitHub · 410 lines

Changes

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.

  1. 2d ago First seen · 410 lines · 0 tokens per session scan A 977278764cc0

Subscribe to this mod's changes

control-tools is an agent published in the GitHub repository Undertone0809/rudder (287 stars, last pushed 2d ago), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 5,561 tokens. 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.

Related

Other agents, from other repositories