Capture fast
Drop the idea into the inbox before it gets edited out of existence.
capture --type audio --project api-gatewayCapture raw ideas. Turn them into builder-ready outputs.
Voice notes, text jots, screenshots, and links - saved instantly, source-preserved, and distilled into specs, prompts, plans, issues, and exports.
Capture firstSource preservedAI separatedExport ready
Recording idea
Saved locally every few seconds
AudioTextLinksImages
RecordSaved locallyDistilledExported
Raw transcript remains accessible.
Explicit, inferred, open questions, risks, next actions.
How it works
Messy input goes in. Useful working artifacts come out.
The landing workflow mirrors the product contract: capture now, organize later, preserve source always, then generate or export when the intent is clear.
Preserve source
Keep transcripts, screenshots, links, and notes as first-class records.
source.lock({ mode: 'preserve' })Distill later
Generate structured output only after the messy input is safely stored.
ai.distill({ separateInference: true })Export or connect
Ship Markdown, workspace exports, issue drafts, or connector-ready artifacts.
export.markdown({ includeSource: true })Idea InboxQuick capture ready
Audio: auth redirect edge caseScreenshot: billing table layoutLink: Supabase RLS docs
Source + AIseparate records
Raw source preserved exactly as captured.
AI output highlights decisions, risks, and next actions.
Export Consoleconnector-ready
markdown.ziplinear.issue.draftmcp.search.readonlyFeatures
A capture loop tuned for technical context.
QuillCaddie is not another meeting-notes page. It is a builder-focused place to keep raw thinking, technical vocabulary, and reusable output connected.
Capture without ceremony
Start with audio, text, links, images, or screenshots. Folder, model, connector, and template choices can wait.
Source-first by design
Raw transcript and source text stay preserved. AI output is derived, inspectable, and never silently replaces the original.
Developer vocabulary aware
Keep RLS, OIDC, JWT, Supabase, Kubernetes, product names, and internal terminology intact during capture and cleanup.
Project context containers
Group captures by product, workspace, folder, or product context so later AI output has the right working memory.
Export-ready outputs
Move source and derived output into Markdown, folder exports, vault-shaped bundles, or future connector actions.
Connector-ready architecture
QuillCaddie is shaped for action/export connectors first, with full sync and enterprise permissioning kept behind stronger controls.
Read-only MCP direction
The roadmap includes safe read-only search, get, export, and generated-prompt tools for AI coding environments.
Trust controls in the loop
Retention, export, deletion, provider isolation, and workspace boundaries are visible product concerns, not hidden footnotes.
Developer forward
Designed to become the connective tissue between ideas and tools.
The product is architected around export/action connectors first: GitHub issue drafts, Linear work items, Slack posts, Drive or Notion exports, and read-only MCP surfaces for AI tools. Live availability should remain explicit per connector as each integration ships.
Provider-agnostic processing interfaces Workspace and product context boundaries Export and deletion flows built into the model
GitHub
Linear
Slack
Drive
Notion
MCP
QuillCaddiesource + output graph
Legal
Trust pages are part of the product surface.
These draft policy pages reflect the current V1 posture: preserve user source, separate AI output, keep secrets out of user-visible context, and make retention, export, and deletion visible.