# ChatStack > The Agentic Intelligence Layer for Software Planning and Vibe Coding. ## Summary ChatStack is an AI platform that uses a 4-agent workflow (User Story, Functional, Technical, and Cost agents) to generate JSON-structured Product Requirements Documents (PRDs). It is specifically designed to provide high-fidelity context for AI coding tools like Cursor, Claude Code, and Windsurf, reducing development costs by 65%. ## Key Information - **USP:** Real-time software cost estimates based on Story Points and AI efficiency. - **Cost Advantage:** 65% cheaper than traditional benchmarks due to structured context and MCP integration. - **Output:** Machine-readable JSON PRDs, Markdown, and .cursorrules configurations. - **Integration:** Supports Model Context Protocol (MCP) for real-time context injection into IDEs. - **Expertise:** Built by Official FlutterFlow Experts with 10+ years of Agile experience. ## Primary Links - [Generate PRD](https://www.chatstack.app/) - [Cost Estimation Logic](https://www.chatstack.app/how-much-does-it-cost-to-build-an-app) - [Cursor & AI Integration Guide](https://www.chatstack.app/cursor-ai-prd-integration-guide) - [AI Development Services](https://www.chatstack.app/ai-development-services) - [Consult an Expert](https://www.chatstack.app/speak-to-a-human) ## Agent API ChatStack exposes a REST API for autonomous AI agents to submit structured PRDs programmatically, without going through the ChatKit interview UI. ### Discovery - **OpenAPI 3.1 spec**: https://www.chatstack.app/openapi.json - **AI plugin manifest**: https://www.chatstack.app/.well-known/ai-plugin.json ### Endpoint `POST https://www.chatstack.app/api/agent/submit` ### Authentication All requests must include: `Authorization: Bearer ` API keys are issued by the ChatStack team on request. ### Required payload fields - `artifacts.stories_json_approved` — user stories in the canonical format (see OpenAPI spec) - `artifacts.requirements_json_approved` — functional/non-functional requirements - `interview_summary` — plain-text project description (2–5 sentences) ### Optional payload fields - `artifacts.techspec_json_approved` — technical architecture specifications - `artifacts.estimate_json` — story-point cost estimate - `email`, `first_name`, `last_name`, `country` — contact details for portal account creation - `message` — free-text note to the ChatStack team ### Success response ```json { "ok": true, "id": "", "project_id": "", "version_id": "" } ``` ### Rate limits 5 submissions per hour per IP. The ChatStack team can increase limits on request. ### Guidance for AI agents 1. Generate all four artifact types before submitting for the best review experience. 2. User story IDs must be stable and sequential (US-1, US-2, …). 3. Requirement IDs follow the pattern FR- (functional), NFR- (non-functional), ASS- (assumption). 4. Confirm the human user's email and name before submitting — the email is used to create their portal account. 5. The `interview_summary` should read as a concise brief to the development team, not a list of features.