Envelope Open Schema Spec
An open JSON schema (Apache 2.0) for defining AI agent teams — portable, versionable, and deployable across any conforming runtime.
At a Glance
For individuals exploring what AI agents can do.
14-day free trial of the Pro plan.
Engagement
Available On
Alternatives
Listed Jun 2026
About Envelope Open Schema Spec
The Envelope Open Schema Spec is an open JSON schema published under the Apache 2.0 license that defines how AI agent teams are structured, configured, and deployed. It is the portable, runtime-agnostic foundation underlying the Envelope platform, allowing teams to be validated in an IDE, stored in Git, and deployed from CI without vendor lock-in. The schema lives at schema.openenvelope.org and is publicly accessible with no account or subscription required.
What It Is
The Envelope Open Schema Spec is a machine-readable JSON schema that describes the structure of an AI agent team definition — including agents, roles, tools, credentials, access policies, triggers, and memory configuration. Any conforming runtime can read and execute a team defined in this schema, making it a portable standard rather than a proprietary format. Envelope positions it as "the open standard for AI agent teams," analogous to how OpenAPI standardized REST API descriptions.
How It Fits Into the Envelope Platform
The schema is the underlying contract for everything built on the Envelope platform. When a user describes a workflow in plain language on openenvelope.org, Envelope generates a team definition that conforms to this schema. That definition can then be:
- Validated locally in an IDE using standard JSON Schema tooling
- Stored and versioned in a Git repository
- Deployed via CI/CD pipelines to Envelope's managed runtime or to self-hosted infrastructure (AWS Bedrock, Docker, a local Mac mini, etc.)
- Read and executed by any third-party runtime that implements the spec
This separation between the schema (open, portable) and the runtime (Envelope's managed service or self-hosted) is a deliberate architectural choice to avoid lock-in.
Key Schema Concepts
The schema captures the full operational definition of an agent team, including:
- Agent roles and structure — how individual agents are composed into a team
- Tool and credential bindings — which external services each agent can access
- Access policies — per-agent declarations of permitted hosts, methods, and fallback behavior
- Triggers — manual, scheduled (cron), webhook, or condition-based start conditions
- Human review gates — checkpoints requiring sign-off before an action proceeds
- Team memory — read/write memory shared across runs for self-improving behavior
Deployment and Runtime Flexibility
Because the schema is Apache 2.0 licensed and fully public, developers can build validators, adapters, and integrations against it without any Envelope subscription. The Envelope platform itself supports deploying schema-conforming teams to its own managed runtime, to AWS Bedrock, to Paperclip, via Docker, or locally. The schema travels with the team definition, meaning security policies such as SSRF protection and static access policies are enforced at the definition level regardless of where the team runs.
Developer Access and Tooling
The schema reference is documented at openenvelope.org/docs/schema. The GitHub organization at github.com/openenvelope hosts the open-source components. API access for managing teams, triggering runs, reading audit logs, and configuring workspaces programmatically is available on all Envelope plans including the free tier. The schema itself requires no account or plan to use in external tooling.
Community Discussions
Be the first to start a conversation about Envelope Open Schema Spec
Share your experience with Envelope Open Schema Spec, ask questions, or help others learn from your insights.
Pricing
Free
For individuals exploring what AI agents can do.
- 3 active installs
- 5,000 runs / month
- 1M Envelope compute tokens / month
- Scheduled runs — manual and cron
- Human review checkpoints
Pro Trial
14-day free trial of the Pro plan.
- 20 active installs
- 25,000 runs / month
- 5M Envelope compute tokens / month
- Two-person approval for workspace changes
- Role-based access (owner / admin / member)
Pro
For teams running AI agents as part of their operations.
- 20 active installs
- 25,000 runs / month
- 5M Envelope compute tokens / month
- Two-person approval for workspace changes
- Role-based access (owner / admin / member)
- Security audit log — filtering, date ranges, member attribution
- Security dashboard — network map, real-time traffic
- Extended analytics — custom date ranges, exports
- Priority support
Scale
For organisations running AI operations at volume. Coming soon.
- 100 active installs
- 150,000 runs / month
- 30M Envelope compute tokens / month
- Everything in Pro
- Multi-workspace monitoring
- SAML SSO — enterprise identity provider support
- 12-month audit log retention
- Portfolio analytics — org-wide aggregate views
Enterprise
For enterprises needing verified compliance and gateway control.
- Unlimited active installs
- Custom run volume
- Everything in Scale
- Private cloud deployment — run Envelope in your own VPC
- Runtime gateway — Envelope in the call path
- Verified compliance audit trail
- Signed log of every prompt and response
- Dedicated support & SLA
Capabilities
Key Features
- Open JSON schema for AI agent team definitions
- Apache 2.0 license — fully public, no account required
- Portable and versionable — store in Git, deploy from CI
- Per-agent access policies (permitted hosts, methods, fallback)
- Human review gate definitions
- Trigger types: manual, cron, webhook, condition-based
- Team memory configuration for self-improving agents
- Compatible with Envelope managed runtime, AWS Bedrock, Docker, local deployment
- IDE validation via standard JSON Schema tooling
- No vendor lock-in — any conforming runtime can execute
