Access & onboarding
Apply for production access - what we look for, what you provide, turnaround.
Swiggy MCP is invite-based in v1. We review every application. Good integrations get production credentials fast; unclear ones get follow-up questions.
Build locally first, then send us a video
You don't need credentials to start. Wire Swiggy MCP into your agent against a local dev stub (or use staging once you have access), build the experience end-to-end on http://localhost, and record a short screen capture of it working. Walking us through a real flow video is the fastest way to get from application to allowlist - it tells us more in a minute than a paragraph of intent does.
When you apply at /access, include a link to the video (Loom, Drive, YouTube unlisted - anything we can play). Send to builders@swiggy.in if your application form doesn't have a video field.
Who can apply
- Developers building an agent or product that will serve real users on Swiggy tools.
- Platform operators (voice assistants, ambient commerce surfaces) brokering Swiggy for their end users - see Enterprise for the white-glove path.
- Skill authors contributing to Real World Skills with Swiggy as a target provider.
What we look for
- A concrete use case with real end users (not a sandbox demo).
- Alignment with Swiggy's consumer experience - agents that respect the user, confirm orders, and don't surprise them.
- Technical readiness - you can complete OAuth, handle 401/429, retry safely.
- Responsible traffic patterns - you estimate your QPS and agree to honour rate limits.
- Security baseline - HTTPS redirect URIs, no PII storage beyond what you need.
For enterprise partners, additional review covers commercial terms, security posture (SOC 2 preferred), compliance attestations, and platform-specific requirements.
What you provide
When you apply at /access, we ask for:
- Integration name & organization - short, human-readable.
- Redirect URIs - HTTPS, exact-match.
http://localhostallowed for local dev; platform schemes (e.g.alexa://,googleassistant://) allowed case-by-case. - Servers requested - any of
food,instamart,dineout. Access is granted per-server by binding yourclient_idto the server allowlist; v1 scopes (mcp:tools,mcp:resources,mcp:prompts) apply uniformly across whatever you're allowlisted for. - Expected volume - rough estimate of orders/day and tool calls/day.
- Use case - one paragraph describing what you're building.
- Primary technical contact - email that reaches an engineer.
- For enterprise: security attestations available (SOC 2, ISO 27001), expected peak QPS, surfaces (voice/chat/app), end-user geography.
Turnaround
- Developers: staging credentials are issued during application review; production follows after a working staging integration.
- Platform operators / enterprise: variable - typically 4+ weeks because commercial terms are negotiated per-partner.
What you get
- A
client_id- public identifier used in your OAuth flow. - Staging access at
mcp-staging.swiggy.com/{server}- same shape as production, backed by seeded data (no real orders). - Production access once staging has been green for ≥ 48 hours.
- Contact details for your designated engineering point of contact.
- Access to the
builders@swiggy.insupport channel. - Enterprise only: dedicated Slack channel, per-partner dashboards, co-branding assets.
What we ask of you
- Complete the technical checklist in Ship to production before your first real-user call.
- Keep your security contact up-to-date.
- Notify us 7 days ahead of any major traffic event (launch, campaign) so we can provision capacity.
- Treat Swiggy session ids as a support identifier, not a business identifier - don't key your internal records on them.
Revocation
Credentials can be revoked without notice for:
- Security incidents involving your integration.
- Traffic patterns indicative of abuse (scraping, credential stuffing, bulk catalogue export).
- Material violations of the partner terms (covered separately at contract sign).
We'll always try to reach you first if there's ambiguity.