Features

Everything you need to code with AI.

Quill puts AI in the places you already work: the line you are typing, the code you selected, and the task you would rather hand off, with any model you like.

Download free Read the docs

01 · Write

Write faster, in the file you're already in.

Tab completions

Multi-line suggestions from your cursor, the file and your recent edits. Tab to accept, keep typing to ignore.

Quick edit · Ctrl K

Select code, press Ctrl K and describe the change. Review it as an inline diff, accept per line or all at once.

Reviewable diffs

Every AI change lands as a diff. Nothing is written to disk until you say so: per file, per block or everything.

Ask about your code, not code in general

Open chat with Ctrl L. Quill sees the file you are in and the files you mention with @, so the answer uses your names, your types and your conventions.

@users.ts @http.ts Why does fetchUser throw on a 404 instead of returning null? claude-sonnet-5-5 · $$$ http.get rejects on any non-2xx. Wrap it: try { return await http.get(`/users/${id}`); } catch (e) { if (e.status === 404) return null; throw e; } [ Apply as diff ] [ Copy ]

02 · Delegate

Hand off the whole task.

Agent mode

Give it a task. The agent reads, searches, edits across files and checks its own work before it reports back.

How agent mode works

Terminal access

Runs builds, tests and scripts in your terminal. Asks before anything risky; auto-approves the safe stuff if you let it.

Self-check and verify

After editing, the agent runs your verify command (tests, lint, build) and fixes what it broke. Pro

Agent browser

The agent can open your dev server, click through a flow and read console errors to debug UI. Pro

MCP servers

Connect databases, issue trackers and internal tools so the agent can act on them, not just talk about them. Pro

Project rules

Tell Quill how your project works once in .quill/rules.md. Every chat and agent run follows it.

Task Add rate limiting to the public API Read .quill/rules.md "use fastify plugins, no new deps" Search "fastify.register(" in src 3 results Create src/server/plugins/rate-limit.ts Edit src/server/app.ts +4 −0 Run npm run verify ✓ lint · ✓ 48 tests · ✓ build Done 2 files changed · review diff

03 · Plan and test

Know what you're building. Check that it works.

Feature Tree

A map of your app by feature: modules, pages, actions and fields, kept in .quill/feature-tree.yaml next to your code. The agent reads it in every chat. Before it changes code, it names the branch it is about to work in and waits for you to confirm. When it is done, it proposes the tree changes as a diff you approve.

See the tree as a graph, a list or YAML, with the code links and history behind every node, and the branch the current chat is working on lit up.

Clean 7 nodes ├─ Dashboard module │ └─ Dashboard Home page done └─ Administration module ├─ Users Management page done ├─ Roles Management page done └─ Assign Permissions page building └─ Save Role Permissions action planned Working on Administration › Assign Permissions [ Approve tree change ] [ Reject ]

QA in a real browser Pro

Pick QA mode, describe what to test in plain words, and Quill opens your app in its own browser. It clicks and types like a user, reads console errors and failed requests, and reports each step as pass or fail, with a screenshot when something breaks.

Test on a desktop, phone or tablet screen; on phone and tablet it also checks for cut-off text, overlaps and sideways scrolling. Turn on Fix what fails and it fixes the code and runs the same check again. The browser uses a fresh profile, so it never sees your own sign-ins.

QA Sign up, then log in Phone · 390×844 Open localhost:3000/signup ✓ pass Type email, password ✓ pass Click Create account ✓ pass Click Log in ✗ POST /api/login 500 Layout menu ✗ scrolls sideways Done 3 passed · 2 failed · screenshots

04 · Organize

Find any chat again.

Chat folders

Drag a chat into a folder to keep a feature's work together. Folders belong to the project, like its chats, so every project keeps its own.

Labels

Bug fix, Feature, Refactor, Question, Testing and Docs come ready, or make your own. A chat can carry several labels; click one to filter the list.

Bug fixFeatureRefactorQuestion

Auto-label

When a chat finishes, Quill adds the right labels by itself, marked “auto”. It picks only from the labels you tick and reads keywords from your first message, so it costs no credits.

05 · Choose

Any model. Including the ones on your machine.

This is the part other editors won't give you. Pick the model for each job, see what it costs before you send, and bring your own keys or your Claude Code plan for free.

Any model

Claude, GPT, Gemini and Grok hosted by us. Anthropic, OpenAI and Gemini with your own keys. Llama and more locally through Ollama or LM Studio.

ClaudeGPTGeminiGrokOllamaLM Studio

Model guide

Choose the model for each job

A fast, cheap model for completions. A strong reasoning model for the agent. Auto if you'd rather not think about it. The picker shows how capable each model is, from 1 to 5 dots.

CompletionsHaiku 5.5 $

ChatAuto $

AgentSonnet 5.5 $$$

Set models per feature