Tencil vs Agentation
Agentation and Tencil both let you leave feedback on the elements of a running page for a coding agent. Agentation turns annotations into structured Markdown you paste into your agent, or that your agent reads over MCP. Tencil starts the agent itself: every pin is a thread in Claude Code or T3 Code, and the agent’s progress streams onto the pin.
npx tencilAt a glance
Checked against Agentation’s own site on 16 October 2026.
| Tencil | Agentation | |
|---|---|---|
| What you do | Press C, click an element, type what should change. | Click an element in the toolbar, write feedback, add it to the list. |
| How it reaches the agent | Tencil starts a thread with your agent, one per pin or one per batch. | Copy the Markdown and paste it, or let the agent read it over MCP. |
| What the agent gets | Component and owners with file and line (React 19), markup, selector, text, size, viewport, an optional screenshot. | Selectors, source file paths, the React component tree, computed styles, and your feedback with intent and priority. |
| Conversation | On the pin: progress streams in, and you reply there. | Over MCP, the agent can ask questions on an annotation and resolve it with a summary. |
| Agents | Claude Code and T3 Code. | Any agent; it names Claude Code, Codex, Gemini and Grok. |
| Integrations | A Next.js component, a Vite plugin or a script tag. | An npm package, an MCP server, webhooks and an API. |
| License | Apache-2.0 | Free for internal use; a commercial license to redistribute it. |
Which should you choose?
They solve overlapping problems in different ways. Pick by how you like to work.
Choose Tencil if
- You want the agent started for you, and the result on the page while you keep reviewing.
- You want every piece of feedback to stay a thread you can continue in your agent.
- You need an Apache-2.0 license, for example to build on it.
Choose Agentation if
- You want feedback as structured Markdown, or served to an agent Tencil does not run.
- You want webhooks or an API to send feedback to other services.
- You prefer to collect a review first and hand it to the agent later.
Tencil vs Agentation: questions
What people ask when they compare the two.
No. Tencil does not wait for the agent to read feedback: it starts the agent with the pin, through T3 Code or the Claude Agent SDK.
Yes. Batch mode queues comments on their elements, and Send hands them to your agent as one thread, numbered like their pins.
More comparisons
How Tencil compares with other ways to point an agent at your UI.
- Tencil vs React GrabReact Grab copies a React element’s file, component and HTML for you to paste into your agent. Tencil sends it to your agent with your prompt and streams the change back onto the page.
- Tencil vs stagewisestagewise is an open-source agentic IDE with its own browser and agents on any model. Tencil adds element pins to the dev site in your own browser and hands them to Claude Code or T3 Code.