custom-integrations

A system for connecting coding agents to external services through custom APIs or managed MCP connections. MCP is a standard way for an agent to discover and call tools provided by another service.

In plain words
What is it for?
Use it to connect agents to Supabase, Linear, Notion, GitHub, or custom MCP servers, manage authorization, discover available tools, and bind connections to organizations or agents.
Why use it?
It keeps connection settings and credentials organized while limiting secret exposure. It also records which agents can use each connection and evidence of tool calls.

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/custom-integrations
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 6,121 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 $0.06121
Opus 5 $0.00000 $0.03061
Sonnet 5 $0.00000 $0.01224
Haiku 4.5 $0.00000 $0.00612

Measured 2d ago against content hash 7e339261008c, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

custom-integrations 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/custom-integrations.md · 460 lines

How it starts

The opening of the file, as written. The whole thing — 460 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Agent Custom Integrations

AGENT.CUSTOM.INTEGRATIONS.001

Contract Summary

Rudder supports the existing custom_api and legacy mcp_server records plus organization-owned managed MCP connections for Supabase, Linear, Notion, GitHub, and custom servers. Managed connections own provider, transport, external scope, access mode, lifecycle, safe non-secret configuration, discovered tools, agent bindings, and redacted dispatch evidence. OAuth grants and temporary OAuth sessions are separate encrypted credential records; GitHub uses a personal access token stored in the same encrypted managed-secret boundary. Public connection summaries expose credential presence only, never secret identifiers or values. External MCP credentials stay server-side throughout authorization, discovery, and dispatch. Every managed connection has immutable organization or agent scope and its own credential boundary. Supabase, Linear, Notion, and GitHub allow one canonical connection per provider and target: one Organization connection plus one independent connection for each Agent. Custom MCP remains multi-instance within either scope. When both official scopes exist, the Agent connection takes precedence. Operators assign coarse-grained Agent access instead of managing individual tool checkboxes. Plugin import may prepare a disabled custom MCP draft through this same service, but package metadata never supplies runtime admission, credentials, or Agent access. The operator continues setup and activation through the managed MCP boundary.

Intent / User Job

Operators need to connect internal APIs, remote MCP servers, and organization specific tools to agent work loops without waiting for each provider to become a first-party Rudder integration. Some tools are shared infrastructure; others are personal or identity-bound and must attach to exactly one agent.

Why / Design Reasoning

Custom integrations are agent-visible runtime capabilities, not generic global settings. The product separates them from fixed provider integrations such as Feishu because fixed providers may have provider-specific identity, chat, and runtime semantics. Custom integrations need explicit organization, owner-agent, binding, tool, credential, and audit records so Rudder can enforce scope before any prompt or tool-call surface sees them.

Read the full file on GitHub · 460 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 · 460 lines · 0 tokens per session scan A 7e339261008c

Subscribe to this mod's changes

custom-integrations 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 6,121 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.