Scribble Studio Docs

Scribble Studio turns messy inputs (typed notes, audio recordings, and uploaded files) into structured outputs: summaries, action items, diagrams, and GitHub-ready issues. The product is intentionally opinionated: everything is optimized for engineering throughput.

The interface is designed around a single flow: capture → analyze → review → export. The docs below explain what each step produces, what controls you have, and what to expect when connecting external services.

A session is a container for a single meeting, voice note, brainstorm, or document. It stores input, outputs, and any integrations used for that work.

Sessions are designed to be revisited. You can refine output, export multiple times, and keep a clear trace from raw input to shipped work.

Tokens are a usage unit for AI analysis. Each analysis request consumes tokens based on how much content you submit, so shorter, cleaner inputs are both cheaper and higher quality.

If you hit a limit, it’s usually because your plan restricts daily analysis volume or you’ve exhausted your token balance.

Projects represent a team boundary. They group sessions, permissions, and integrations so the right people can access the right work.

If you collaborate across repos or business units, use multiple projects to keep exports and access clean.

Record standups, voice notes, or meeting segments directly in the app. Once you stop recording, Scribble turns the audio into text and uses that transcript as analysis input.

Best results come from clear speakers, a short recap of context, and explicit decisions (“we will”, “we won’t”, “blocked by”, “definition of done”).

Upload a file when the source of truth is already written: meeting notes, incident writeups, PRDs, or QA checklists. Scribble reads the content, extracts structure, and proposes actionable tasks.

Use file upload for dense contexts; use typed notes for quick iteration; use audio when you want speed and natural recall.

Analysis converts your input into structured artifacts: a readable summary, a prioritized task list, and optional diagrams. You can use the output as-is, or refine by editing your input and re-running.

The task list aims to be shippable: each task includes a title, description, priority, and type (bug/feature/chore). The goal is a PR-sized unit of work, not an epic.

Linking a repository lets Scribble create issues in the correct place and attach context (labels, assignees, and summaries). Repository access is scoped; you can disconnect at any time.

If you’re working across repos, link the repo that will ultimately own the work. You can still keep the session content generic.

After analysis, select the tasks you want and create issues. The generated issue body is designed to be developer friendly: context, expected behavior, and acceptance criteria are prioritized.

Review issue titles for specificity. “Fix checkout 500 on submit” ships better than “Fix checkout bug”.

Bulk export creates multiple issues in one pass. Use it after a meeting where you’ve captured many action items and want to seed a sprint backlog quickly.

When exporting in bulk, keep tasks short and independent. If a task depends on another, note that dependency in the description.

Scribble supports a two-step PR workflow so you can review diffs before opening a pull request.

Use Initiate PR to generate the Scribble branch and get a GitHub “Comparing changes” link. Share that link for early feedback without creating a PR yet.

When you’re ready, use Open PR to create a draft PR from the initiated branch. This targets develop when available, otherwise the repository’s default branch.

To work in bulk, click Select in the issue list, choose the issues you want, then initiate or open PRs in one pass.

Access control determines who can view sessions, export issues, and manage integrations. Keep write access limited to people who are responsible for backlog hygiene.

If you’re evaluating the product, start with a small project and expand once your workflow is stable.

Analytics helps you understand usage patterns: how often the team analyzes, which flows they use, and where they drop off. This is a feedback loop for improving the workflow, not monitoring individuals.

If your org has strict compliance requirements, align analytics collection with consent settings and internal policy.

Use the in-app Support center to report issues, request features, and get help with integrations. Support threads keep context and attachments together so resolution is faster.

Include: steps to reproduce, expected behavior, and what you saw instead. Screenshots are helpful; logs are optional.

Keep sensitive data out of inputs whenever possible. Treat recordings and uploads like internal documents: only share what’s needed to generate work.

If you need stricter guarantees, use project boundaries and keep exports limited to approved repositories.