Issue enrichment

Pullfrog enriches new issues by default on newly added repositories. Existing repositories keep their settings. Configure enrichment under Automations → Enrich issues in the console.

Enrich new issues

Enrichment checks for duplicates, surfaces useful related issues, and investigates concrete bugs or feature requests filed by collaborators with repository write access. It posts a concise plan for actionable collaborator requests; other issues receive analysis or a request for missing information. Enrichment does not implement changes automatically.

Each run uses the repository’s configured model. Choose the behavior:

• Enrich — use the default enrichment prompt or replace it with a custom prompt
• Plan — analyze the issue and post an implementation plan as a comment
• Build — jump straight into implementation (creates a PR)

Customize the prompt

Click View default prompt, then Copy prompt. Paste it into Enrichment prompt and edit it. An empty field uses the default; entered text replaces it. Clear the field to restore the default. In Plan and Build modes, custom instructions supplement the selected mode’s prompt instead.

Implement from an issue

Click Implement a fix on an enrichment reply to start a separate Build run. Plan comments offer Implement plan. Both require sign-in and repository write access. Pullfrog reads the issue discussion, implements the requested work, and opens a PR; the enrichment reply remains on the issue.

Options

• Include issues from non-collaborators (public repos) — auto-respond to community-filed issues without giving non-collaborators access to other triggers

If you’d rather let authors decide when to involve Pullfrog, set issues to Quick Links mode instead. Pullfrog will post a comment with one-click Make a plan and Build this buttons instead of auto-triggering.

Issue indexing

Issue indexing is a Pro feature.

Enrichment can search repository issues by keyword without a Pro subscription. Pro adds retrieval from Pullfrog’s background issue index. Pullfrog inspects likely matches before deciding whether an issue is a duplicate, and links a confirmed duplicate instead of producing another plan. Neither search is an exhaustive duplicate check.

Pullfrog never closes or labels an issue on similarity alone; the decision stays with a maintainer.

Indexing itself runs from the moment a repository is set up, on every plan, and Pullfrog backfills the existing issues at that point. Pro is what lets Pullfrog use the index. That ordering is deliberate: a duplicate is only catchable if the index already covers the original, so building the history early is what makes the feature useful on the day you upgrade rather than a year later.

Indexing reads issue titles, bodies, and comments, and may read a bounded amount of repository source when an issue references specific files. It never writes to your repository.

Auto-label issues

Independently of issue enrichment, Pullfrog can automatically label new issues based on their content. It reads the title and body, picks from your repository’s existing label set, and applies the best matches.

You can customize labeling behavior with Labeling instructions in the console (e.g., “Prefer specific labels over generic ones. Use at most 3 labels.”).

Plan instructions

The Plan instructions field in the Modes section of the console guides how Pullfrog writes implementation plans. These apply whenever Pullfrog runs in Plan mode — whether triggered by issue enrichment, Quick Links, or a direct @pullfrog mention.