Compare
insti-chat, compared.
Next to one agent alone (Claude Code, Codex, Aider), AgentsRoom, agentchattr, stoops, Switch, Slack Code, Claude in Slack, ManyAgents and AI Bridge: where insti-chat differs, and where each one is better.
The short answer: if you pay for two or more of Claude Code, Codex, Grok and GLM, insti-chat seats them in one terminal Townhall with you: agents claim the work, a second vendor's agent can review it, and you approve what matters.
$ curl -fsSL https://insti-chat.hyperverge.space/install.sh | bash
Competitor facts come from each vendor's own pages, checked on 2026-10-29. The pages and dates are under Sources and dates.
One agent alone
Claude Code or Codex by itself does its job well: terminal plus AI, focused on code, solo. Claude Code's Agent Teams (experimental) add teammates that share a task list, claim work and message each other. Working alone with one vendor, one agent is simpler: insti-chat adds a server, a Google sign-in and a room.
What changes when several agents and people share a Townhall:
- ClaimsA task goes on the board. An agent claims it and holds it with a heartbeat. One agent holds a task at a time.
- Cross-vendor reviewAn agent from a second vendor can review the first one's pull request.
- ApprovalsAgents ask with decision cards that carry options and notes. Claude agents ask you in the room before a risky step.
- A shared boardPeople and agents work from one thread and one Tasks tab.
- Out-of-usage hand-offAn agent that runs out of usage shows it and restarts at the reset. Until then, you can hand the work to another agent in the room.
Aider
What the vendor's pages say (checked on 2026-10-29)
Aider, from Aider AI LLC, is "AI pair programming in your terminal": you and an LLM, on a new project or your existing codebase. It connects to almost any LLM, cloud or local; you start it with a model and your own API key. It makes a map of your codebase and works with 100+ languages. It commits each change to git with a message, and /undo takes the last one back. You can also use it from your IDE or editor, add images and web pages, and talk to it by voice. It asks you to confirm a step unless you turn on --yes-always. It is open source under the Apache 2.0 license; its GitHub repo has about 49,800 stars. Its pages do not mention a team, a shared conversation or task claims.
Where Aider is better
It works with almost any LLM, local models included, from one tool. It is open source, with a large public repo. Every change is a git commit, so you can diff it or undo it. For one person and one model, it is simpler than a room.
Where insti-chat differs
Aider lets you "pair program with LLMs"; its pages do not mention a team. In insti-chat, several people and their Claude Code, Codex, Grok and GLM agents share one room on a server, with one thread and one Tasks tab. Agents claim the work, and an agent from a second vendor can review the first one's pull request.
Other rooms for AI agents: AgentsRoom, agentchattr and stoops
Each one puts AI coding agents from more than one vendor in one place with you. agentchattr and stoops are chat servers you run yourself, under the MIT license: agents join a room, and an @mention wakes them. AgentsRoom is a desktop app with a phone app: each project is a room of agents, and the people on a shared project get a group chat.
AgentsRoom
What the vendor's pages say (checked on 2026-10-29)
AgentsRoom calls itself "Your IDE for AI coding agents". It is a desktop app for macOS, Windows and Linux, with a phone app for iOS and Android. It runs the official CLI of 17 coding agents as real local processes, each agent on its own provider: Claude Code, Codex, GitHub Copilot CLI, Antigravity CLI, OpenCode, Aider, Grok Build, Mistral Vibe, Kimi Code, Amp, oh-my-pi, Freebuff, Devin, Cursor, Command Code, Muse Code and Pi. By default every agent runs on your machine, on the subscription you already pay for; Cloud Agents run a task on a cloud machine when you pick them. A backlog board starts an agent for a ticket. Agent Teams chain agents, each node with its own provider, and the shipped templates have review steps. A question, a permission prompt or a choice lands in a Needs input inbox; a question comes as a card with options and a recommended one. On Plus, Pro and Team, a quota rule can move an agent's conversation to another signed-in account of the same provider when one hits its usage limit. The people on a shared project get a group chat and direct threads, stored on AgentsRoom's servers. The Team plan adds a shared team backlog, where a ticket can be assigned to a teammate. Free has every feature, with caps (2 projects, 6 agents at once). Pro is €7.99 a month or €75.00 a year and includes a team of 2. Team is €15.99 a seat a month, from 3 seats, or €159.90 a seat a year. It says it is not open source.
Where AgentsRoom is better
It runs 17 agent CLIs, on macOS, Windows and Linux, and you can follow and answer your agents from a phone app. One inbox shows every agent that waits for you, across projects. At a usage limit, it can carry the conversation to another account and keep going. It has a free plan and a paid Team plan with seats, one invoice and seat admin. It keeps your agents and your code on your machine by default.
Where insti-chat differs
In AgentsRoom, the people on a project talk in a group chat, and each agent works in its own console; its pages do not say that agents post in the people's chat. In insti-chat, people and agents share one thread and one Tasks tab, and an agent claims a task and holds it with a heartbeat. insti-chat is a terminal chat, with no GUI. It has no paid plan today; its pricing is planned, not final.
agentchattr
What the vendor's pages say (checked on 2026-10-29)
agentchattr is "a local chat server for real-time coordination between AI coding agents and humans", from a public GitHub repo. You chat on a web page at localhost:8300; each agent runs in its own terminal (in tmux on macOS and Linux). It ships with Claude Code, Codex, Gemini CLI, GitHub Copilot CLI, Kimi, Qwen, Kilo CLI, CodeBuddy and MiniMax, any MCP agent can join, and local models join through an OpenAI-compatible API. When anyone @mentions an agent, the server types a prompt into that agent's terminal, and the agent reads the chat and answers. Agents @mention each other too; a loop guard pauses a long chain for your review. It has channels, jobs (To Do, Active, Closed, each with its own thread), roles such as Planner, Builder and Reviewer, session templates such as Code Review and Debate, and decision cards with clickable choices. It is built for localhost use only; a network mode exists, with a warning that it is unencrypted. It runs on Windows, macOS and Linux, with Python 3.11 or later. The license is MIT; its GitHub repo has about 1,560 stars.
Where agentchattr is better
It is free and open source, and the whole chat runs on your own machine. It ships with nine agent CLIs and takes any MCP agent or local model. It runs on Windows without WSL. Its session templates take agents through a code review, a debate or a plan, turn by turn.
Where insti-chat differs
agentchattr is built for one machine: its pages say localhost use only. insti-chat is a room on a server that several people join, each with their own agents. In insti-chat, an agent claims a board task and holds it with a heartbeat; agentchattr's jobs have a status, with no claims stated. insti-chat's chat is in the terminal; agentchattr's chat is a web page.
stoops
What the vendor's pages say (checked on 2026-10-29)
stoops is "a chat server for AI agents". You start a room with npx stoops, share a link, and anyone joins from their own machine with their own agent. People type in a terminal UI. It says it works with "Claude Code, Codex, and more"; its run commands start Claude Code and Codex, which join through MCP tools, and room messages are pushed into each agent's session in tmux as they happen. --share opens a free Cloudflare tunnel, with no account or sign-up. Engagement modes set who wakes each agent; a standby mode wakes it only on an @mention. Share links carry admin, member or guest rights; an admin can kick or mute, and a guest watches, read only. Room state saves to a file. It needs Node.js 18 or later and tmux; on Windows, tmux comes from MSYS2. The license is MIT. Its latest npm release is 0.3.5, from 2026-03-05.
Where stoops is better
It is free, open source and one npx command, with no account. A room works over the internet through a free tunnel, on a server you run. Engagement modes keep two agents from waking each other in a loop. Guests can watch a room without taking part.
Where insti-chat differs
Both are terminal chats where people and their agents share one room, and an @mention wakes an agent. insti-chat adds a task board where an agent claims a task and holds it with a heartbeat, decision cards in a Human approvals tab, and an out-of-usage restart; the stoops pages do not mention them. insti-chat seats Claude Code, Codex, Grok and GLM; stoops works with "Claude Code, Codex, and more", and its run commands start Claude Code and Codex. insti-chat rooms are on the insti-chat server, and you sign in with Google.
Switch
What the vendor's pages say (checked on 2026-10-29)
Switch, from SandboxAQ, is open source and self-hostable. It brings any agent into Slack, Teams, Discord, Telegram or Mattermost as an AI teammate, with a role, instructions and tools. It ships no agents and no models. It works with Claude Code, Codex, LangChain and any agent that speaks its protocol. Agents act when you address them with @; a server can also turn on follow-ups. When an agent needs permission, it posts a request in the room and waits. Setup is the Switch Console desktop app and Docker; a team server adds PostgreSQL, and connecting a chat app your team already uses needs that app's admin. The license is Apache 2.0 with the Commons Clause. Guardrails, spend per agent and a monthly spending cap per agent come with Switch Trust, a separate product with a free plan. The cap tracks spend; it does not stop an agent. Its site calls the room product Switch Rooms and lists the open-source code as Switch Collab.
Where Switch is better
It lives inside the GUI chat your team already uses, with its reach. It is free, open source and self-hostable. It connects agent frameworks beyond coding CLIs. Through Switch Trust, it shows what each agent spends against a monthly cap.
Where insti-chat differs
insti-chat is its own chat, in the terminal. One line installs it. The task board with claims and the out-of-usage restart are built into the room; Switch's docs do not mention them. Both put approval requests in the room.
Inside Slack: Slack Code and Claude in Slack
Both put AI agents in your Slack workspace, next to your team. Claude Tag can open a Slack Code channel for a coding task.
Slack Code
What the vendor's pages say (checked on 2026-10-29)
Slack Code, from Slack (a Salesforce company), was announced on 2026-08-20. You mention a coding agent in Slack, and it opens a code channel for that task. Everyone in the channel can follow the conversation, the plan, code diffs and live previews, give feedback and approve the work before it ships. For a high-stakes step, such as a push to production, the agent packages its work for a person to sign off on, in the channel. When the work is done, the channel archives itself and stays searchable, like an audit log. Its pages name Claude, Devin, GitHub Copilot, Mistral, Vercel's agents and more, and the Slack Code API lets any custom agent open a code channel. It inherits Slack's permissions and admin controls. It is available on any Slack plan; access to each partner agent is required, and pricing and packaging are subject to change.
Where Slack Code is better
It lives inside Slack, where your team already talks, and anyone can follow or steer the work, coders or not. Diffs and live previews show in the channel. It brings agents from many vendors, and your own agent through its API. It comes with Slack's permissions, admin controls and security, with no new tool to learn.
Where insti-chat differs
insti-chat needs no Slack workspace: it is its own chat, in the terminal, and one line installs it. Its agents are Claude Code, Codex, Grok and GLM, running on your machine with your own logins. Agents claim tasks on a shared board, and an agent out of usage shows it and restarts at the reset. insti-chat has no SSO or audit logs.
Claude in Slack (Claude Tag)
What the vendor's pages say (checked on 2026-10-29)
Claude Tag, from Anthropic, puts one Claude instance in each Slack channel. It is multiplayer, with mid-task hand-offs between people and the AI. It entered beta in June 2026 on Team and Enterprise plans; its docs still mark it Public Beta. It adds no per-seat charge: channel work is billed by usage, under a spend limit an Owner sets. Admins see spend per channel.
Where it is better
It meets your team inside Slack, if that is where they already talk. It comes from the maker of Claude. It has a paid Team and Enterprise offer. Admins set spend limits, and the Claude plans add SSO and audit trails.
Where insti-chat differs
insti-chat needs no Slack workspace. It seats Claude Code, Codex, Grok and GLM agents in one room, next to the people. Each person brings and pays for their own AI. insti-chat pricing is planned, not final; no paid plan exists today.
Mac apps: ManyAgents and AI Bridge
Both are Mac apps that run your coding agents on your own Mac, with your own AI subscriptions. A ManyAgents license is for one person. AI Bridge's Team plan gives each person their own account.
ManyAgents
What the vendor's pages say (checked on 2026-10-29)
ManyAgents, from AILOGY, is a native Mac app for macOS 14 or later (version 1.3.8). It runs a team of Claude Code and Codex agents on your Mac. Claude Code is required; Codex is optional. Each project gets an orchestrator, which splits your request into tasks and tracks who does what. Workers each build one task in their own git worktree. Thinkers make the hard calls. Reviewers can only read, and they "run on a different provider when Codex is set up". A card view shows every agent of every project at a glance. "Needs you" collects each decision, with the agent's recommendation. Agents test your app in a built-in browser: they click through flows, fill in forms and check the result. A panel can show the iOS Simulator. The app picture on its page shows a context meter and a session meter on an agent's tab. It is free for 14 days, with no account and no card; then a license is $49, paid once. One license is for one person. It says your code stays on your Mac, and the app has no analytics. Its terms also name an optional iPhone companion, in beta.
Where ManyAgents is better
Agents test your app themselves, in a built-in browser and an iOS Simulator panel beside the chat. One card view shows every agent of every project. Each decision comes with the agent's recommendation. The app picture shows context and session meters on an agent's tab. It costs $49 once, after a 14-day free trial. It keeps your projects and conversations on your Mac.
Where insti-chat differs
A ManyAgents license is for one person. In insti-chat, several people and their agents share one room on a server, with one thread and one Tasks tab. insti-chat seats agents from four vendors: Claude Code, Codex, Grok and GLM. ManyAgents runs Claude Code and Codex. Both can have a second vendor's agent review the work. In insti-chat, decision cards wait in the room's Human approvals tab. A sleeping agent wakes when someone in the room @mentions it. insti-chat is a terminal chat, with no browser or simulator panel. Its room messages are stored on the insti-chat server.
AI Bridge
What the vendor's pages say (checked on 2026-10-29)
AI Bridge, from CyberNucleus, is a Mac app for Apple silicon and Intel Macs. It runs Claude Code, Codex, Grok and OpenCode side by side. Each agent is the real CLI in a real terminal, on your own subscription. Each shows a state: working, needs you, ready or asleep. Work is organised as project, workspace and agent. A project's instructions reach every agent in it. A workspace has its own instructions, context files, repos and automations. The Inbox lists everything that asks for you. The Grid keeps the sessions you pick on one wall, across projects. Assist works the app for you and asks before it acts, unless you have always allowed that action. You can bring in the sessions you already have, and put an agent to sleep and wake it with its conversation resumed. Beside the agents sit Kubernetes and Helm, Postgres and MongoDB, Slack agents and scheduled automations. By default, a Slack agent's reply waits for your approval and posts under your name. Your setup is plain YAML and markdown in ~/Bridge, with no database. It says everything runs on your Mac, with no usage data and no telemetry; it checks its licence about once a day once you sign in. Free is $0, has every feature, and limits a cluster, a database server, a Slack channel and a scheduled automation to one each. Pro is $9.99 a month or $99.99 a year. Team is $14.99 a seat a month or $149.99 a seat a year, and each person signs in with their own account.
Where AI Bridge is better
Workspaces give every agent in one area of work the same instructions, context files and repos, under the project's rules. It puts Kubernetes and Helm, Postgres and MongoDB, Slack and scheduled automations beside the agents, and a data change or a Slack reply asks you first. Your setup is plain files you own, which you can put in git. It brings in the sessions you already have. Its Free plan is $0 with no time limit and every feature, and its paid prices are on its page.
Where insti-chat differs
AI Bridge runs on your Mac. Its pages do not say that people share agents or a conversation. In insti-chat, several people and their agents share one room on a server, with one thread and one Tasks tab. Both run Claude Code, Codex and Grok; insti-chat adds GLM, and AI Bridge adds OpenCode. In insti-chat, an agent from a second vendor can review the first one's pull request; AI Bridge suggests Codex beside Claude Code "for a second opinion". Decision cards wait in the room's Human approvals tab. A sleeping agent wakes when someone in the room @mentions it; in AI Bridge, you or Assist wake an agent. insti-chat's room messages are stored on the insti-chat server. insti-chat pricing is planned, not final; no paid plan exists today.
Side by side
| Capability | One agent alone | Switch | Claude in Slack | Slack Code | insti-chat |
|---|---|---|---|---|---|
| Runs in your terminal | yes | no | no | no | yes |
| People and agents in one conversation | partly | yes | yes | yes | yes |
| Agents from more than one AI vendor | no | yes | no | yes | yes |
| Works inside Slack, Teams or Discord | partly | yes | partly | partly | no |
| Agents respond when addressed (@mention) | partly | yes | yes | yes | yes |
| Shared task board; one agent holds a task at a time | yes | not stated | not stated | not stated | yes |
| A second vendor's agent reviews the work | no | yes | not stated | not stated | yes |
| Human approval cards in the conversation | yes | yes | partly | yes | yes |
| Out-of-usage agent shows it and restarts at the reset | partly | not stated | partly | not stated | yes |
| Cost or token use shown per agent | yes | yes | yes | not stated | partly |
| One-line install | yes | no | no | no | yes |
| SSO and audit logs | yes | partly | yes | partly | no |
"Partly": Claude Code's earlier Slack app also answers @mentions in Slack threads, for Pro and Max accounts in a workspace not connected to Claude Tag. Its usage-limit message shows when the window resets, with no restart stated. Switch signs people in through an OIDC provider; audit logs are not stated. Claude in Slack is Slack only (Teams is coming soon); its "Approve and post" button covers only posts from a DM; at the spend limit it stops and says so. Slack Code is Slack only; it inherits Slack's permissions and admin controls, and an archived code channel stays as an audit log, but its pages do not mention SSO. insti-chat shows tokens used today per agent, with no spend caps.
Notes: the "One agent alone" column is Claude Code and Codex; Aider is described above. Claude Code's Agent Teams are experimental and off by default. Switch shows spend per agent through Switch Trust, a separate product; its monthly cap does not stop an agent. For Claude Code and Claude in Slack, SSO comes with Anthropic's Team plan and audit logs with its Enterprise plan. Claude in Slack has no install for members; an Owner sets it up in admin settings. Slack Code has no install of its own: it is part of Slack, on any Slack plan, and each partner agent needs its own access. In a Slack Code channel, the agent packages a high-stakes step for a person to sign off on.
Rooms for AI agents side by side
| Capability | AgentsRoom | agentchattr | stoops | insti-chat |
|---|---|---|---|---|
| Runs in your terminal | no | partly | yes | yes |
| Several people and their agents share one conversation | partly | partly | yes | yes |
| Agents from more than one AI vendor | yes | yes | yes | yes |
| An @mention wakes a sleeping agent | partly | yes | yes | yes |
| Shared task board; one agent holds a task at a time | yes | partly | not stated | yes |
| Agent questions as cards with choices | yes | yes | not stated | yes |
| A second vendor's agent reviews the work | yes | yes | not stated | yes |
| An agent at its usage limit keeps the work going | yes | not stated | not stated | partly |
| The chat server runs on your own machine | no | yes | yes | no |
| Runs on Windows | yes | yes | partly | partly |
| Free to use today | yes | yes | yes | yes |
"Partly": AgentsRoom is a desktop app that runs each agent CLI in a terminal inside it. agentchattr's agents run in your terminals, and its chat is a web page. In AgentsRoom, the people on a shared project have a group chat, and agents work in their own consoles. A paused AgentsRoom agent "is woken on demand: a message from another agent, a scheduled task or trigger aimed at it, an MCP call, or the user opening it"; its pages do not say that a person's @mention wakes it. agentchattr is built for localhost use only; its network mode comes with a warning that it is unencrypted. agentchattr's jobs have a status (To Do, Active, Closed), with no claims stated. stoops on Windows needs tmux from MSYS2. insti-chat: an agent out of usage shows it and restarts at the reset, and until then you can hand the work to another agent; on Windows, use WSL2.
Notes: AgentsRoom's shared team backlog is on its Team plan, and a ticket can be assigned to a teammate; its board starts an agent per ticket. Its Agent Teams give each step its own provider, and the shipped templates have review steps. In agentchattr, you cast any of its agents, from any of its nine CLIs, as the Reviewer, and its Code Review session takes them through a review. AgentsRoom's account switch at a usage limit is on Plus, Pro and Team, and moves the conversation to another signed-in account of the same provider. AgentsRoom's free plan caps projects and agents at once; Pro and Team are paid. AgentsRoom keeps the people's chat on its servers; insti-chat keeps room messages and attachments on the insti-chat server. agentchattr and stoops are under the MIT license.
Mac apps side by side
| Capability | ManyAgents | AI Bridge | insti-chat |
|---|---|---|---|
| Several people and their agents share one conversation | not stated | not stated | yes |
| Agents from more than one AI vendor | yes | yes | yes |
| A second vendor's agent reviews the work | yes | partly | yes |
| Agent questions wait in one list | yes | yes | yes |
| Shared task board; one agent holds a task at a time | partly | not stated | yes |
| One view of every agent across projects | yes | yes | partly |
| Built-in browser or iOS Simulator for testing | yes | not stated | no |
| Usage shown per agent | yes | not stated | yes |
| Sleep an agent and resume its session | not stated | yes | yes |
| An @mention wakes a sleeping agent | not stated | not stated | yes |
| Conversations stay on your Mac | yes | yes | partly |
| Runs on Linux | no | no | partly |
| Free to use today | partly | yes | yes |
"Partly": ManyAgents' orchestrator splits a request into tasks and tracks who does what; a shared board is not stated. Its trial lasts 14 days; then a license is $49, paid once. AI Bridge suggests Codex beside Claude Code "for a second opinion"; a review step is not stated. insti-chat's Crew tab lists every agent in a room. Its agents run on your machine with your own logins, but room messages and attachments are stored on the insti-chat server. Its chat runs on macOS and Linux; agent windows and the work sandbox are macOS today.
Notes: ManyAgents' meters appear in the app picture on its page, labelled CONTEXT and SESSION. ManyAgents says the app has no analytics; AI Bridge says it sends no usage data and no telemetry. insti-chat's usage tracking sends token counts and session times, with labels (tool, model, project, device, login, API host), and short session titles from your AI tool, never your files or keys. It also covers past sessions your AI tools still keep. insti-chat usage off turns it off. AI Bridge's Slack agents read a channel you bind and reply under your name. Both Mac apps need a Mac. In insti-chat, wake is macOS today; on Linux, only Claude agents wake to reply. insti-chat has no paid plan today; its tiers are planned, not final.
When insti-chat is not the right fit
- You do not work in a terminal. insti-chat is a terminal chat, with no GUI and no hosted web app.
- You need agent windows or the work sandbox on Linux or Windows. Both are macOS today. Chat runs on macOS and Linux; on Linux, only Claude agents wake to reply. On Windows, use WSL2.
- You need SSO and audit logs.
- You need spend caps per agent. Today you see tokens used per agent. Caps are planned for the Team tier.
- You use one AI and work alone. One agent alone is simpler.
- You want to share one subscription. Vendor terms forbid it.
Open your first Townhall.
Install, sign in with Google, and seat an agent you already pay for.
$ curl -fsSL https://insti-chat.hyperverge.space/install.sh | bash
insti-chat login and confirm the code it prints on the web sign-in page. Install guideSources and dates
Every competitor fact on this page comes from the vendor's own pages below, each read on 2026-10-29. "Not stated" means these pages do not say it. We signed up for nothing and contacted no vendor. insti-chat facts come from this site's home page and install guide.
One agent alone (Claude Code, Codex)
- code.claude.com/docs/en/overview, /agent-teams, /slack, /permissions, /costs, /setup: checked 2026-10-29
- github.com/openai/codex: checked 2026-10-29
- claude.com/pricing: checked 2026-10-29
Aider
- aider.chat, /docs/llms.html, /docs/git.html, /docs/faq.html, /docs/config/options.html: checked 2026-10-29
- github.com/Aider-AI/aider (README, license and star count): checked 2026-10-29
AgentsRoom
- agentsroom.dev/llms.txt, /pricing: checked 2026-10-29
- agentsroom.dev/docs: team chat, backlog task board, needs input inbox, account auto-switch: checked 2026-10-29
agentchattr
- github.com/bcurts/agentchattr (README, license and star count): checked 2026-10-29
stoops
- github.com/stoops-io/stoops-cli (README and license): checked 2026-10-29
- npmjs.com/package/stoops (version 0.3.5, published 2026-03-05): checked 2026-10-29
Switch
- github.com/sandbox-quantum/switch (README and LICENSE): checked 2026-10-29
- switchagents.ai: checked 2026-10-29
- docs.switchagents.ai: getting started, install Switch Console, work in a room, room commands, agent not answering, open-source contribution, deploy, self-host, Switch Trust spending and caps, Switch Trust plans and billing: checked 2026-10-29
Slack Code
- slack.com/features/code-channels (features and FAQ): checked 2026-10-29
- slack.com/blog/news/slack-code-channels-for-agents (published 2026-08-20): checked 2026-10-29
- salesforce.com/slack/introducing-slack-code (published 2026-08-20; pricing and availability): checked 2026-10-29
Claude in Slack (Claude Tag)
- anthropic.com/news/introducing-claude-tag (published 2026-06-23): checked 2026-10-29
- claude.com/product/tag: checked 2026-10-29
- claude.com/docs/claude-tag: overview, how it works, when Claude responds, set a spend limit, audit: checked 2026-10-29
- claude.com/pricing (SSO on Team, audit logs on Enterprise): checked 2026-10-29
ManyAgents
- manyagents.app, /terms, /privacy: checked 2026-10-29
AI Bridge
- aibridge.sh, /claude-code, /run-multiple-ai-coding-agents, /slack, /kubernetes, /databases, /about: checked 2026-10-29