Getting Started

Quick Start

Releap is read-only codebase intelligence by default. Connect sources, ask a product or engineering question, and use the cited answer to create tickets, epics, KB pages, release notes, or opt-in draft PRs with real implementation context.

The Releap delivery loop

Releap is organized around the software delivery loop: understand the product, plan the change, capture the work, implement it, verify it, document it, and communicate it. Each surface shares the same grounded context instead of forcing teams to re-explain the codebase at every step.

• Chat - Ask product, support, architecture, security, or implementation questions and get cited answers with confidence signals.
• Planning - Turn exploratory Q&A into epics, child stories, acceptance criteria, code references, and review questions engineers are likely to ask.
• Tickets - Capture actionable work with title, description, acceptance criteria, linked code references, suggested fix prompts, status, priority, story points, due dates, and external tracker exports.
• Knowledge Base - Convert durable answers and tickets into searchable internal docs that can be retrieved in future chats and flagged stale when cited code changes.
• Public Docs - Publish selected external-facing documentation from a separate Public KB with explicit snapshots, custom domains, and optional token-restricted access.
• Release Notes - Draft audience-specific release communications from tickets, milestones, code context, and approved KB content while preserving approvals and publication history.
• MCP - Let agentic development tools query Releap, inspect ticket context, update tickets, propose documentation changes, record verification, and coordinate opt-in branch work.
• VS Code extension - Bring the working queue, Releap chat, ticket handoffs, local acceptance checks, and MCP setup into the editor.

Chat and cited answers

Chat is the starting point for most Releap workflows. A user can ask "how does this flow work?", "where is this enforced?", "what changed?", "what would break if we changed this?", or "what should we document?" Releap retrieves from the sources visible to the workspace and answers with citations so the user can inspect the evidence.

• Multi-source retrieval - Code, Knowledge Base articles, release-note artifacts, Jira, Confluence, Aha!, and other enabled product sources can participate in the same answer.
• Application and branch scope - Queries can be constrained to the product surface, service, repository, or branch that matters for the question.
• Thread memory - Follow-up questions keep the investigation together so a thread can become a ticket, KB page, summary, or planning session.
• Action hints - When Releap detects actionable work, it can suggest creating a ticket, planning an epic, updating documentation, or preparing release communication.
• Governance - Workspace plans, role checks, audit history, BYO model settings, sensitivity handling, and monthly AI usage apply the same way in chat, API, and MCP calls.

Tickets, epics, and planning

When a question uncovers work to do, create a Releap ticket draft or switch into Planning Mode for a larger epic. Releap keeps the evidence close to the request: title, description, acceptance criteria, child stories, code references, implementation notes, and a suggested fix prompt that another teammate or coding agent can act on.

• Ticket drafts - Store status, priority, story points, due date, acceptance criteria, code references, comments, and suggested fix prompts.
• Planning Mode - Generate an epic from the current investigation, ask clarifying questions, produce child stories, and run a review against the current implementation.
• Tracker exports - Export to GitHub Issues on all paid plans; Business and above can export to Jira, Linear, and Aha! as well.
• Duplicate-aware work - Compare tickets, search existing ticket drafts, and update exported work instead of creating repeated issues when the same draft is exported again.
• Branch handoff - With explicit workspace admin enablement and the separate GitHub write scope, apply a ticket to an auto-generated or user-designated non-default branch and open a draft PR for review.
• Verification - Record acceptance-criteria checks, linked PR metadata, commit SHAs, and local or CI test results so the ticket reflects what was actually verified.

Knowledge Base

Releap can turn answers and tickets into internal Knowledge Base pages. Internal KB pages stay authenticated to your team. They are searchable, retrievable in chat, versioned, and flagged stale when cited source files change.

• Spaces and pages - Organize product knowledge by workspace and application.
• Generated sections - Start from a cited answer, ticket, or agent proposal, then edit and publish with your team.
• Chat retrieval - Published pages become first-class retrieval sources, so future answers can combine product docs with current code.
• Staleness signals - See which KB pages need review after related source files or product context changes.
• Attachments - Add images, pasted screenshots, and supporting documents to preserve the evidence and visuals behind an article.
• Agent updates - MCP clients can append a new section or replace the first generated section after a user confirms the update.

Public Docs

Public Docs is a separate publish surface for customer-facing documentation. Teams can copy from Internal KB as a starting point, then edit, review, and publish an independent Public KB snapshot that is safe to share outside the workspace.

• Separate public content - Public KB sites, areas, articles, sections, versions, and assets are independent from Internal KB content.
• Snapshot publishing - Public visitors see only the latest published snapshot. Draft edits, internal prompts, private citations, and unpublished assets stay hidden until republished.
• Images and documents - Authors can paste screenshots, upload images, add alt text and captions, and attach supporting files to public articles.
• Search and feedback - Published docs support public search, basic helpfulness feedback, and moderation controls for comments-style discussion.
• Custom domains - Business and Enterprise teams can serve Public Docs from a verified hostname such as docs.example.com after DNS ownership verification.
• Restricted access - Public Docs can be anonymous or token-restricted for customer-only documentation links.
• Release-note links - Release Notes can point readers to Public Docs articles, sections, or areas so a shipped change connects directly to the relevant guide.

Release communications

Use tickets, milestones, code context, and approved KB content to draft audience-specific release notes. Releap preserves approvals, destinations, and publication records so release communication becomes part of the searchable product memory.

• Audience-aware drafts - Generate different versions for customers, internal teams, executives, support, or implementation partners from the same underlying change set.
• Review workflow - Track approval state, change requests, reminders, delivery policy, and publication failure records.
• Source-grounded content - Pull from tickets, milestones, code references, commit summaries, KB pages, and release artifacts instead of starting from a blank document.
• Documentation links - Attach internal KB links for authenticated audiences and Public Docs links for customers, including stable URLs frozen at publish time.
• Retrievable history - Published communication chunks can be searched later, so Releap can answer not just what shipped, but what the team told users.
• Agent-assisted edits - MCP clients can update release-note content after confirmation, making customer-facing copy part of the same agentic delivery loop.

MCP and agentic development tools

Releap exposes a Model Context Protocol endpoint at /mcp so Claude Code, Cursor, Cline, Codex, and other MCP-capable agents can use Releap as a grounded context and workflow layer. The agent can stay in the tool a developer already prefers while Releap supplies the product memory, citations, tickets, and verification records.

• query_codebase - Ask across indexed code, KB articles, release-note artifacts, and visible workspace sources; receive citations, clean answers, proposed updates, and action hints.
• list_repos and read_chunk - Discover visible repositories and retrieve cited chunk content directly when the agent cannot follow a GitHub citation URL.
• releap_create_ticket and releap_compare_tickets - Convert discoveries into ticket drafts and compare related work before creating duplicates.
• generate_epic, update_story, run_review, export_epic - Drive planning workflows from an agent while retaining Releap's structured epic and story model.
• search_tickets and update_ticket - Let an agent find assigned work, refine acceptance criteria, adjust priority, or enrich the suggested fix prompt after user confirmation.
• apply_ticket_to_branch and create_pr - Queue opt-in branch work on non-default branches when the workspace has enabled Releap Code and granted the separate write scope.
• verify_acceptance_criteria and run_tests - Record acceptance checks and agent-reported local or CI test results. Releap does not execute arbitrary repository commands for MCP clients.
• update_kb_article and update_release_note - Promote implementation discoveries into documentation and release communications without leaving the agent workflow.

MCP access uses workspace-scoped API keys or OAuth-connected agent tokens. Calls are audited, plan-limited, and governed by the same repository visibility, application access, sensitivity, BYO model, and monthly AI usage settings as the Releap web app.

VS Code extension

The Releap VS Code extension brings the delivery workflow into the editor: the working queue, Releap Q&A, native VS Code Chat commands, ticket handoffs, local acceptance checks, Releap Code controls, status, and MCP setup.

• Working queue - Add tickets to a personal queue, open ticket details, jump to referenced files, and keep implementation work visible next to the code.
• Native chat - Use @releap and slash commands such as /ask, /ticket, /handoff, and /verify.
• Agent prompts - Copy handoff prompts for Codex, Claude, Copilot, Releap Code Chat, or another agent, including the Releap ticket and MCP context the agent should use.
• Local verification - Run acceptance checks from the editor and record the result back on the Releap ticket or code job.
• MCP setup - Copy the right MCP config snippet for Claude Code, Codex, Cursor, or another client. The extension does not proxy MCP traffic; agents connect directly to Releap.

Monthly AI usage

AI-powered actions draw from one workspace usage pool: answers, planning, ticket and story generation, KB generation, release-note drafting, summaries, verification, and draft-PR workflows. Reading existing content and exporting already-created work does not consume the pool.

• Pro - Standard monthly usage per seat; 1M-token usage blocks are $10.
• Business - Shared team pool; 1M-token usage blocks are $8.
• Growth - Larger shared team pool; 1M-token usage blocks are $6.
• Enterprise - Contracted usage volume and usage-block terms.

Core concepts

• Workspaces - The tenant boundary for members, repos, queries, tickets, settings, and audit rows.
• Applications - Product surfaces or services inside a workspace that help Releap organize sources and answers.
• Sources - Repositories, Confluence spaces, Jira projects, and Aha! products Releap is allowed to index.
• Knowledge Base - Authenticated team knowledge that can be generated, edited, searched, embedded, and checked for staleness.
• Public Docs - Intentional unauthenticated or token-restricted documentation published from Public KB snapshots, optionally on a custom domain.
• Release communications - Audience-specific notes, approvals, destinations, and publication records linked to tickets and milestones.
• MCP clients - Agentic development tools connected to Releap through workspace-scoped API keys or OAuth tokens.
• Releap Code - The opt-in workflow that applies a ticket to a non-default branch and opens draft PR work for engineering review.
• Citations - File, line, and source references attached to answers and tickets.
• Monthly AI usage - The shared workspace pool consumed by LLM-backed actions.
• API keys - Workspace-scoped credentials for MCP access and programmatic clients.

Need Help?

If you are preparing a rollout or security review, start with the security model or contact our team.