Write specs that reflect how your product works

Overview

Codebase analysis connects Productboard Spark to your codebase so it can answer plain-language questions about how your product works today. Every PM can look up limits, business rules, feature flags, and permission logic directly.

Supported hosts are GitHub, GitLab.com, Bitbucket Cloud, and Azure DevOps Services.

Understand any part of the product

Ask Spark questions like "What are the file upload limits?" or "Which feature flags control the new onboarding flow?" and get answers sourced from your connected repositories. Spark draws on the codebase automatically whenever it's relevant.

Write specs grounded in the implementation

While you write a spec, Spark surfaces the existing constraints, APIs, edge cases, and hidden requirements that affect it, well before you meet with engineering counterparts.

Onboard to new product areas

Explore an unfamiliar domain by asking Spark how it works — billing, permissions, a legacy subsystem — and get an overview drawn from the code. New PMs get up to speed without scheduling engineering walkthroughs.

Look up implementation facts

Check feature flag status, analytics instrumentation, permission logic, rate limits, and tier-specific configuration straight from the source. Answers cite the current state of the code, so they hold up in customer calls and engineering discussions.

Setup

A workspace admin connects an account once. Codebase analysis is then available to everyone in the workspace. The admin needs access to the organization or workspace being connected, and an admin on that host completes or approves the installation.

1. From the integration settings in Productboard, connect your GitHub, GitLab.com, Bitbucket Cloud, or Azure DevOps Services account and approve the installation.
2. Choose which repositories to index. Initial indexing can take up to an hour.
3. Manage repositories any time — toggle indexing per repository, trigger a manual reindex, or synchronize newly added repositories. Everything re-indexes automatically every night.

Access is read-only: Spark reads and analyzes code but cannot create branches, open pull requests, or commit changes. Indexing is supported for TypeScript, JavaScript, Python, Ruby, Kotlin, GraphQL, and YAML.