Wesight
Open-source desktop workspace that runs Claude Code, Codex, OpenClaw, and other CLI coding agents side by side with per-task model routing.
If you are already paying for Claude Code and Codex and still opening four terminals to watch them, WeSight removes real friction: one workspace, one permission log, per-task engine switching, local SQLite persistence, MIT license. Routing routine work to GLM-5.3 through Ollama or OpenRouter while sending hard problems to frontier APIs is the cost story, and it is a real one. The trade against cloud-first assistants is support and polish — community channels rather than a vendor SLA, and a macOS Apple Silicon build rather than every platform. Worth an afternoon if you're terminal-fluent; skip it if you want an assistant, not a control panel.
Verified 5d ago · liveness 67/100 · cite: rightaichoice.com/tools/wesight
- Developers already running two or more CLI coding agents who want them supervised from one window
- Engineers comparing how different models handle the same coding task before committing to one
- Privacy-conscious developers who want agent state stored locally rather than in a vendor cloud
- Power users who want MIT-licensed tooling they can inspect and modify
- Non-technical users wanting a plug-and-play AI assistant with no setup
- Teams needing cloud collaboration, shared sessions, or browser-based access
- Anyone who needs a mobile app — WeSight is a desktop workspace
We scan live Reddit threads, YouTube comments, X posts, G2 reviews and other communities — and hand you an honest verdict in under a minute.
- Honest verdict, not marketing
- Real pros & cons from real users
- Attributed quotes with receipts
3 free scans · no card needed
Skip WeSight if you want a finished AI assistant rather than a control panel for agents you already run from the terminal — it assumes you'll supply your own CLI agents, provider accounts, and API configuration.
WeSight itself is MIT-licensed, but every routed provider is billed separately by that provider — heavy Claude Code or Codex use shows up on Anthropic's and OpenAI's invoices, not WeSight's.
No pricing page was reachable for this refresh, so treat cost here as $0 for the MIT-licensed WeSight app plus whatever your chosen providers charge. That makes WeSight cheaper than per-seat orchestrators or cloud agent platforms at any team size, and roughly equal to a plain terminal — the real budget line is API spend, especially if you route everything to frontier models instead of sending routine work to GLM-5.3 via Ollama or OpenRouter.
In short
Wesight — Open-source desktop workspace that runs Claude Code, Codex, OpenClaw, and other CLI coding agents side by side with per-task model routing. Best for Developers already running two or more CLI coding agents who want them supervised from one window, Engineers comparing how different models handle the same coding task before committing to one, Privacy-conscious developers who want agent state stored locally rather than in a vendor cloud. Free to use.
What people actually say about Wesight — is it worth it?
We scanned public community sources for Wesight on Sep 24, 2026 and could not establish that the discussion we found is about this tool rather than something else sharing its name. Only 4 of the posts we fetched could be positively tied to Wesight. Rather than publish a sentiment score built on the wrong subject, we publish nothing here and re-run the scan.
Viability Score
How well maintained and how widely used is Wesight? Built from what the vendor actually publishes (docs, changelog, tutorials, integrations, pricing), whether the site is live, and how much real users discuss it. How we calculate this
Last calculated: October 2026
How we score →Key Features
- One-click connection for Claude Code, Codex, OpenCode, Qwen Code, DeepSeek-TUI, OpenClaw, Hermes Agent, and Claude Agent SDK
- Per-task execution engine selection from the task header, switchable mid-session
- Unified model provider settings covering OpenAI-compatible endpoints, Anthropic, DeepSeek, Qwen, Gemini, Ollama, OpenRouter, and custom providers
- Configurable API Base URL, model list, and compatible format per provider
- Route to open-weight models such as GLM-5.3 via Ollama locally or OpenRouter hosted
- Visual permission events for file access, shell commands, and network requests reviewed before execution
- Inspected tool execution panels showing tool calls, command output, and file operations
- Chat and workroom tabs with runtime status visible alongside the prompt box
- Built-in file explorer and terminal
- Skills system including web-search, docx, xlsx, pptx, pdf, playwright, and imap-smtp-email
- Slash commands such as /run, /plan, /docs, and /search
- Scheduled tasks for recurring research, reports, reminders, and automation
- Memory extraction for continuity across sessions
- Agent logs and output history for later review
- Obsidian plugin connecting your vault to desktop agent sessions
About Wesight
WeSight is an open-source desktop app for macOS Apple Silicon that pulls your terminal coding agents — Claude Code, Codex, OpenCode, Qwen Code, DeepSeek-TUI, OpenClaw, Hermes Agent, and the Claude Agent SDK — into one visual workspace instead of leaving them scattered across terminal tabs. Each runtime reuses your existing local CLI configuration and accounts, so setup is a matter of connecting what you already have rather than re-authenticating everything. The part that earns its keep is per-task model routing. You pick the engine from the task header and configure providers, API base URLs, model lists, and compatible formats from visual settings — OpenAI-compatible endpoints, Anthropic, DeepSeek, Qwen, Gemini, Ollama, OpenRouter, and custom providers. Following the August 2026 open-weight release of GLM-5.3 on Hugging Face, that routing list picked up a genuinely inexpensive option you can reach locally through Ollama or hosted through OpenRouter, which matters if you're paying frontier API rates for routine work. Control sits alongside convenience. File access, shell commands, network requests, and tool calls surface as permission events you can review before they proceed. Conversations stream into a desktop chat with a separate workroom view, tool output and command results render in traceable panels, and skills like web-search, docx/xlsx/pptx, pdf, playwright, and imap-smtp-email turn repeat work into callable capabilities. Slash commands, scheduled tasks, and memory extraction keep long jobs moving across sessions; everything persists to a local SQLite layer under an MIT license. An Obsidian plugin connects your vault to the same sessions, and there's an optional desktop companion you can enable from Appearance settings. It suits developers already running two or more CLI agents who want them supervised from one window, plus anyone comparing how different models handle the same coding task before committing. It is not a cloud product and not a plug-and-play assistant.
Behind the Verdict
WeSight is best understood as an orchestration layer, not a model. It doesn't try to beat Claude Code or Codex at coding; it assumes you've already chosen them and asks why you're managing them across four terminal windows. That framing is honest and the execution follows it: runtimes connect to visual chat, reuse local CLI sessions and accounts, and stay switchable from the task header mid-session. The routing configuration is the most concrete thing here. Providers, API base URLs, model lists, and compatible formats are managed in one settings surface covering OpenAI-compatible endpoints, Anthropic, DeepSeek, Qwen, Gemini, Ollama, OpenRouter, and custom providers. That's a real answer to a real problem — teams routinely overpay because switching a job from a frontier model to a cheap one means editing config, and config edits don't happen at 11pm. With GLM-5.3 open-weight on Hugging Face as of August 2026, the cheap end of that list got materially more capable, and WeSight is one of the few surfaces where that swap is a dropdown rather than a rewrite. Observability is the second pillar. Permission events for file access, shell commands, and network requests appear before the operation proceeds, tool calls and command output render in panels, and agent logs persist for later review. If you've ever watched an agent quietly rewrite a config file, this is the feature you were missing. Skills — web-search, document formats, PDF, Playwright browser operation, IMAP/SMTP email — plus scheduled tasks and memory extraction extend that from a session tool to something that keeps working while you're away. The weaknesses are structural rather than fixable. It's a desktop app prioritising Apple Silicon, so no browser access, no mobile, no shared cloud sessions. Support is community-driven open-source. And there's an inherent dependency risk in wrapping other vendors' CLIs: WeSight manages OpenClaw and Hermes runtimes itself, but Claude Code and Codex are Anthropic's and OpenAI's to change, and a breaking CLI update lands on WeSight's maintainers to absorb. Where it fits: solo developers and small technical teams who already run multiple CLI agents, want agent state on their own machine, and would rather inspect MIT-licensed source than trust a black box. Where it doesn't: anyone who wants an AI assistant rather than an agent runtime, teams needing shared sessions, and organisations requiring a vendor contract.
Researching Wesight? Get your full AI stack in 60 seconds.
Free, no signup — tell us your goal and get tools matched to your budget & existing stack.
Real-world workflow fit
Concrete scenarios for the personas Wesight actually fits — and what changes day-one when you adopt it.
Connect both CLI agents to WeSight, then run a refactor in Claude Code while Codex takes a separate test-writing task in parallel — watching tool calls and permission events for each in the workroom panels rather than two terminal tabs.
Outcome: Both agents supervised from one window with a shared file explorer, and a permission log you can review before either touches a file or shell command.
Configure Ollama, OpenRouter, and a frontier provider in the unified model settings, then run the same debugging task three times, switching the engine from the task header each time.
Outcome: A side-by-side read on whether a cheap GLM-5.3 route handles the task before you commit frontier API budget to it.
Install the WeSight Obsidian plugin, start a cowork session from the vault sidebar, and invoke web-search and docx skills while drafting notes — memory and scheduled tasks stay connected to the desktop app.
Outcome: Agent chat, skills, and task management available next to your writing without leaving the vault.
Use Cases
- Run Claude Code and Codex side by side in one window instead of four terminal tabs
- Route a routine refactor to GLM-5.3 via Ollama and a hard debugging problem to a frontier model, in the same session
- Compare how two different models attack the same coding task before committing to one
- Keep sensitive code processing on your own machine with local SQLite persistence and no vendor cloud
- Review file, shell, and network permission events before an agent acts on them
- Turn recurring research or report generation into a scheduled task with memory carried across runs
- Chat with your agents and invoke skills from inside an Obsidian vault via the plugin
- Chain web-search, pdf, and playwright skills into a reusable automation for document and browser work
Models Under the Hood
as of 2026-09-01
Limitations
- WeSight is a local-first desktop workspace: sessions, settings, and memory live in a local SQLite persistence layer, and public desktop builds prioritise macOS Apple Silicon, so there is no browser-based or mobile access and no shared cloud sessions for a team.
- Support comes from community channels around an MIT-licensed codebase rather than a commercial vendor relationship.
- Because WeSight orchestrates other vendors' CLI runtimes, changes those projects make to their CLIs can affect how a runtime behaves inside the workspace, even though OpenClaw and Hermes runtimes are prepared and maintained by WeSight itself.
- Getting value assumes terminal fluency and willingness to configure your own model providers and API keys.
as of 2026-10-04
Verification history
We have re-verified Wesight 9 times since . Each pass re-reads the vendor's own pages and re-checks every listed field against that evidence; passes where nothing had changed are marked as such.
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-checked, vendor evidence unchanged
- — re-verified summary, description, our verdict, our analysis, pricing model, pricing tiers, features, integrations, who it suits, who should skip it
- — re-checked, vendor evidence unchanged
- — re-checked, vendor evidence unchanged
Showing the 6 most recent of 9 verification passes.
Free to cite with attribution — this page re-verifies continuously.
Where the pricing makes sense
The company stage and team size where Wesight's pricing actually pencils out — and where peers do it cheaper.
No pricing page was reachable for this refresh, so treat cost here as $0 for the MIT-licensed WeSight app plus whatever your chosen providers charge. That makes WeSight cheaper than per-seat orchestrators or cloud agent platforms at any team size, and roughly equal to a plain terminal — the real budget line is API spend, especially if you route everything to frontier models instead of sending routine work to GLM-5.3 via Ollama or OpenRouter.
Setup time & first value
How long it actually takes to get something useful out of Wesight — broken out by persona, not the marketing-page minute.
Developers already running CLI agents: roughly 15-30 minutes to connect existing runtimes (they reuse local CLI configuration and accounts) and point model providers at your API keys. Anyone new to CLI agents should budget an afternoon, since you're configuring runtimes and providers from scratch rather than installing a finished assistant.
Switching to or from Wesight
How to bring data in from common predecessors and how to get it back out — written for the switcher, not the buyer.
- →From bare terminal tabs: connect the CLI agents you already run — WeSight reuses your local CLI sessions, accounts, and configuration rather than re-authenticating.
- →From a single-agent IDE: keep your existing runtime and add others alongside it, switching engines per task from the task header.
- →From a cloud agent platform: recreate your provider list in the unified model settings, then rely on local SQLite persistence for session state.
- →From Obsidian-only workflows: add the plugin so vault sessions and desktop agent sessions share memory and scheduled tasks.
- ↗To a plain terminal: your CLI agents are unchanged underneath, so you can drop back to running them directly at any time.
- ↗To another orchestrator: export or re-enter your provider, base URL, and model list configuration, since that is the piece worth preserving.
- ↗To a cloud agent platform: you would give up local SQLite persistence and gain shared sessions, so expect to re-establish model routing there.
- ↗To a single-agent IDE: keep the one runtime you use most and drop the multi-engine routing and permission-review surface.
Integrations
Resources & Guides
Tutorials & Learning
YouTube returned 6 videos for “Wesight”, and we withheld 6: 6 could not be judged, because “Wesight” is a single word that other videos use for other things. We are showing none, because we could not prove any of them are about Wesight.
Official links
Tools that pair well with Wesight
Common stack mates teams adopt alongside Wesight, with the specific reason each pairing earns its keep.
GitHub Copilot
GitHub Copilot is an AI coding agent that completes code, reviews pull requests, and runs Copilot, Claude, and Codex agents inside GitHub
Ellipsis
Ellipsis Agent Cloud runs Claude Code and Codex coding agents in managed cloud sandboxes defined by environment.yaml.
Refact.ai
Open-source autonomous AI coding agent that plans, executes, and deploys tasks inside VS Code and JetBrains IDEs.
Featured Head-to-Head Comparisons
Wesight vs Locus Robotics
Locus Robotics and Wesight serve entirely different audiences. Locus is a physical automation solution for high-volume warehouses, requiring significant investment and space. Wesight is a free, open-source tool for developers to run AI coding agents locally. Choose Locus if you operate a warehouse and need to boost productivity 2-3x; choose Wesight if you're a developer experimenting with multiple AI agents and want privacy.
Wesight vs Presto Voice
Presto Voice and Wesight serve completely different buyers. Presto is an enterprise-grade drive-thru AI for QSR chains, delivering measurable revenue lift via upselling and high automation rates. Wesight is a free, open-source desktop tool for developers who want to run multiple coding agents locally with custom LLM routing. Choose Presto if you operate a drive-thru chain seeking ROI; choose Wesight if you're a developer managing AI agent workflows.
Wesight vs Truleo
Choose Truleo if you are in law enforcement and need an all-in-one intelligence platform that automates lead generation, jail call analysis, and report writing from siloed systems. Choose Wesight if you are a developer or researcher looking for a free, open-source desktop workspace to run multiple AI coding agents locally with custom LLM routing. They serve completely different use cases.
Alternatives to Wesight
View allGitHub Copilot
GitHub Copilot is an AI coding agent that completes code, reviews pull requests, and runs Copilot, Claude, and Codex agents inside GitHub
Frequently Asked Questions
Best-of guides
Used Wesight? Help shape our editorial sentiment research.