agent-bridge — Watchlist

MCP100 Assessment

Watchlist / Tracking

Why we’re tracking it

Monitor exact implementation activity, MCP interface, permission and setup changes.

Reviewed Oct 10, 2026 · Documentation review; complete setup workflow not independently tested

Best fit

Claude Code and Codex agent bridge

What we know

A local bridge for bidirectional collaboration between Claude Code and Codex. 连接 Claude Code 与 Codex 的本地实时协作桥接工具。 Public Radar listing describes capability; verify precise runtime setup and privilege boundaries upstream.

Maintenance / activity evidence

GitHub exact repo pushed_at 2026-10-09T20:15:06Z

Tracking evidence observed Oct 10, 2026. These dated signals do not establish current maintenance. Reviewed enrichment confidence: medium.

Verification gaps / watch-outs

Installation and comprehensive security have not been independently audited.
Bidirectional agent bridge may propagate commands across code agents; isolate local workspace and approvals. Reported action classes: agent-mediated code/file changes. Complete write authority is not verified.
Refer to first-party setup documentation; do not infer authorization from package existence.
Sensitive access: local code workspace and agent conversations

Bidirectional agent bridge may propagate commands across code agents; isolate local workspace and approvals.
Public tracking suitability is not an installation safety audit.

Full authority, access and evidence limits are available in Review context and Access & risk.

Popularity

374 repository stars

Stars observed Oct 10, 2026.

Repository total for raysonmeng/agent-bridge.

All limitations / Review context

Has agent-bridge changed its documented MCP interface, authentication or privilege boundaries, and does it merit promotion?

Limited editorial listing; known permission implications disclosed, no independently established publication blocker.

Bidirectional agent bridge may propagate commands across code agents; isolate local workspace and approvals.
Public tracking suitability is not an installation safety audit.
Refer to first-party setup documentation; do not infer authorization from package existence.
local code workspace and agent conversations
Reviewed write actions: agent-mediated code/file changes
Sensitive access: local code workspace and agent conversations
Complete write authority is not verified.
Installation, client compatibility and comprehensive security have not been independently audited. Authentication and access depend on the configured services.
Upstream credential and permission scopes must be validated before runtime use.
Precise setup/auth/permissions not verified
Write actions/tool inventory not confirmed against exact tool documentation; listed action classes remain illustrative or context-dependent. Empty inventory does not establish read-only.
Some integration/runtime instructions are not confirmed from a first-party install guide and therefore intentionally omitted.
Exact GitHub pushed_at 2026-10-09T20:15:06Z

Original activity observations

GitHub exact repo pushed_at 2026-10-09T20:15:06Z

Technical identity

github.com/raysonmeng/agent-bridge

Exact implementation source: https://github.com/raysonmeng/agent-bridge ↗
Reviewed resource type: Gateway
Repository identity: raysonmeng/agent-bridge

Evidence & Sources

Evidence basis: Recorded source evidence
Setup runtime verification: Not independently runtime-tested by MCP100 for this setup workflow.

Reviewed Oct 10, 2026 · Latest source observation Oct 10, 2026

GitHub repository ↗
raysonmeng/agent-bridge
Recorded repository material and project history.

Setup · Watchlist

Self-hosted gateway service setup

Publisher instructions.

Source evidence observed Oct 14, 2026.

Additional setup details required — see the specific gaps below.

Install / connect
Publisher instructions
AgentBridge links Claude Code and Codex sessions; it is a cross-agent bridge, not an isolated MCP stdio server.
publisher documented step · Publisher instructions
npm install -g @raysonmeng/agentbridge
publisher documented step · Publisher instructions
abg init
publisher documented step · Publisher instructions
abg claude --safe
publisher documented step · Publisher instructions
abg codex --safe
Configure / authenticate
MCP100 setup guide
Install Bun first (Node alone is insufficient). npm postinstall may register Claude Code plugin; abg init also writes .agentbridge/config.json. Use abg claude --safe and abg codex --safe in separate terminals; bare abg claude/codex bypass normal permission prompts.
Check connection
Suggested check
Run `abg doctor` (read-only diagnosis) and verify the Claude/Codex pair is connected; `abg init` configures plugin for the project. No generic mcpServers registration.
Remove / uninstall / revoke
Publisher instructions
Stop AgentBridge with its documented `abg kill` command for the chosen pair and remove the bridge registration from any configured clients. Uninstall its application only through the installation method originally used.
Client unregistration, program uninstall, persistent-data deletion, and credential revocation must be treated separately.
Stop deployment / service · Publisher instructions
abg kill

Official Documentation & Sources

What it does

A local bridge for bidirectional collaboration between Claude Code and Codex. 连接 Claude Code 与 Codex 的本地实时协作桥接工具。 Public Radar listing describes capability; verify precise runtime setup and privilege boundaries upstream.

Capability: Developer Tools
Resource type: Gateway
Best fit: Claude Code and Codex agent bridge

Representative uses

Claude Code and Codex agent bridge

Limitations

Bidirectional agent bridge may propagate commands across code agents; isolate local workspace and approvals.
Some integration/runtime instructions are not confirmed from a first-party install guide and therefore intentionally omitted.
Exact GitHub pushed_at 2026-10-09T20:15:06Z

Reviewed access

Access & risk

Review the exact access boundary before connecting this MCP to accounts, credentials, or sensitive data.

Authentication: Refer to first-party setup documentation; do not infer authorization from package existence.
Permissions: Bidirectional agent bridge may propagate commands across code agents; isolate local workspace and approvals.
Data handling: local code workspace and agent conversations
Cost: Provider or usage fees possible; verify current pricing
Risk and cautions: Bidirectional agent bridge may propagate commands across code agents; isolate local workspace and approvals.

Access scope and cautions

Write actions: Not verified agent-mediated code/file changes
Sensitive access: local code workspace and agent conversations
Risk level: high
Security notes: Bidirectional agent bridge may propagate commands across code agents; isolate local workspace and approvals. Public tracking suitability is not an installation safety audit.

The reviewed evidence does not establish the complete write authority of this MCP. Consult the unresolved evidence before granting write access.

FAQ

Based on reviewed evidence as of 2026-10-10T03:38:00Z. Watchlist tracking is not a recommendation or a security certification.

Which tasks are documented for this implementation?

The reviewed tasks include: Claude Code and Codex agent bridge.

What setup and authentication are documented?

A complete connection procedure is not verified. Client notes: Installation command omitted unless verified or clearly package-specific; do not infer npx/uvx entrypoints. These facts do not supply missing installation commands, credentials or a verified authentication procedure. Authentication: Refer to first-party setup documentation; do not infer authorization from package existence.

Can it modify data or perform write operations?

Not verified. The available evidence does not establish the complete write authority. The partial inventory lists: agent-mediated code/file changes. Recorded permission context: Bidirectional agent bridge may propagate commands across code agents; isolate local workspace and approvals.

What data-handling boundary should I consider?

local code workspace and agent conversations Recorded sensitive-access context: local code workspace and agent conversations.

What limitations and risks should I check before use?

Bidirectional agent bridge may propagate commands across code agents; isolate local workspace and approvals. Some integration/runtime instructions are not confirmed from a first-party install guide and therefore intentionally omitted. Exact GitHub pushed_at 2026-10-09T20:15:06Z. Public tracking suitability is not an installation safety audit. Upstream credential and permission scopes must be validated before runtime use.