Envelope

Envelope (also referred to as Open Envelope) is a web-based AI agent generator at openenvelope.org ¹. Users describe a workflow in plain language, and the tool generates a structured multi-agent specification covering agent roles, skills, model assignments, tool access, reporting lines, approval gates, and handoff logic ¹. The output is a .envelope.json file that follows an open JSON schema published under Apache 2.0 ². The project launched publicly on Hacker News on June 2, 2026 as a Show HN post, earning 52 points and 13 comments ³. The GitHub organization lists a United States location ⁴.

Product

Envelope operates as a design-layer tool rather than an execution runtime. The workflow has three steps ¹:

1. Describe the workflow in natural language, or enter a company URL for department-specific suggestions.
2. Design the sub-agent structure: roles, skills, models, tools, and reporting lines, refined through conversation.
3. Export or share the resulting spec as JSON, via a shareable link, or through GitHub integration.

The generated spec captures agent definitions (role, prompt, model, access policy), supervisor/sub-agent hierarchy, human-in-the-loop gates, pipelines, and tool requirements ³. The schema is registered in SchemaStore, so dropping a .envelope.json file in VS Code provides autocomplete and validation without installing anything ³. It is also available on npm as @openenvelope/schema (currently at version 1.1.1) for programmatic validation ⁵.

Integration options include an MCP server (compatible with Claude.ai and ChatGPT), a REST API, or a bring-your-own runtime approach where teams parse the spec and provision agents in their own stack ⁶. A Claude Code skill is available for designing multi-agent workflows directly from the terminal ⁴.

Pricing

All plans are free during the current beta period. The listed post-beta pricing ⁷:

Plan	Price	Key features
Free	$0	5 saved designs, full design canvas, JSON export, read-only share link, schema validator
Pro	$9/mo	Unlimited designs, version history, rendered schema doc export, REST API access
Team	$39/mo	Up to 10 collaborator seats, commenting, GitHub publish, org-wide design library
Enterprise	Custom	Unlimited collaborators, private cloud, SAML SSO, audit logs, SLA
Founding Member	$99 one-time	Lifetime Team access (limited to 200 spots during beta)

JSON export is included on every plan, including Free ⁷.

Open schema

The Envelope Team Definition Schema is the core differentiator. Published under Apache 2.0, the $id is https://schema.openenvelope.org/team/v1.json ². Any platform can implement a conforming runtime without permission. The schema covers agent definitions, supervisor/sub-agent hierarchy, human-in-the-loop gates, pipelines, schedules, access policies (declaring which hosts each agent can call, enforced at the network level), and secrets/variables injected at deploy time ³.

The project draws an analogy to Dockerfiles: define a team once, any compatible runtime can execute it ³.

Who it’s for

• Product managers and team leads who need to translate “we want AI agents” into a concrete, reviewable specification before engineering begins.
• Engineering teams who want a structured handoff document rather than a vague brief, with a spec that can be stored in Git and reviewed in a PR.
• Organizations exploring multi-agent AI who need to map out roles, tools, permissions, and handoffs before committing to a particular execution framework.

How it compares

Envelope occupies the design layer: it produces specifications, not running agents. Most competitors focus on execution (building and running agents) rather than structured design output.

	Envelope 7	CrewAI 8	LangSmith/LangGraph 9	AutoGen/AG2 10
Focus	Agent design and spec generation	Agent building and execution	Agent framework + observability	Multi-agent framework
Free tier	5 designs, JSON export	50 executions/mo	5,000 traces/mo	Fully open-source
Paid from	$9/mo (Pro)	Custom (Enterprise)	$39/seat/mo (Plus)	Free (MIT/Apache 2.0)
Output	Open JSON schema	Running agents	Running agents + traces	Running agents
No-code	Yes	Visual editor	No	No
Open standard	Apache 2.0 schema	No	No	MIT / Apache 2.0 code
MCP support	Yes 6	Yes 8	No	No

Alternatives

• Choose CrewAI if you want a managed platform to build and run multi-agent workflows end-to-end, with a visual editor and enterprise governance ⁸.
• Choose LangGraph if you need a code-first Python framework for building stateful agent graphs with LangSmith observability ⁹.
• Choose AutoGen or AG2 if you want a free, open-source multi-agent framework and are comfortable writing Python to orchestrate agent conversations ¹⁰.
• Choose Envelope if you need to design and communicate an agent system before choosing an execution framework, and you want a portable, versionable spec that any runtime can implement.

Verified performance

Lighthouse audit of openenvelope.org (July 2026) [^11]:

Metric	Score
Performance	96
SEO	100
Accessibility	96
Best Practices	96
LCP	1.1 s
CLS	0.002
TBT	0 ms

Company

Envelope is in active development and currently in beta ². The GitHub organization is registered in the United States ⁴. The Hacker News launch was posted by user “ashconway” ³, but the team has not publicly disclosed individual founders or roles on its website ². The organization maintains five public repositories on GitHub, including the core schema (Apache 2.0) and a Claude Code plugin ⁴. The about page states that the design tool and Team plan are commercial products that fund development, while the schema format itself belongs to the community ².

External links

• openenvelope.org (official site)
• About page
• Pricing
• Developer docs
• Schema reference
• GitHub organization
• @openenvelope/schema on npm
• Show HN launch thread

Footnotes

1. Envelope homepage. Retrieved July 31, 2026. https://openenvelope.org ↩ ↩² ↩³

2. Envelope about page. Retrieved July 31, 2026. https://openenvelope.org/about ↩ ↩² ↩³ ↩⁴ ↩⁵

3. Show HN: Open Envelope, Hacker News. Retrieved July 31, 2026. https://news.ycombinator.com/item?id=48315016 ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶

4. Envelope GitHub organization. Retrieved July 31, 2026. https://github.com/openenvelope ↩ ↩² ↩³ ↩⁴

5. @openenvelope/schema, npm. Retrieved July 31, 2026. https://www.npmjs.com/package/@openenvelope/schema ↩

6. Envelope developer page. Retrieved July 31, 2026. https://openenvelope.org/developer ↩ ↩²

7. Envelope pricing page. Retrieved July 31, 2026. https://openenvelope.org/pricing ↩ ↩² ↩³

8. CrewAI pricing page. Retrieved July 31, 2026. https://www.crewai.com/pricing ↩ ↩² ↩³

9. LangSmith pricing (via pecollective.com). Retrieved July 31, 2026. https://pecollective.com/blog/langsmith-pricing/ ↩ ↩²

10. AutoGen multi-agent framework overview (via sanj.dev). Retrieved July 31, 2026. https://sanj.dev/post/autogen-microsoft-multi-agent-framework ↩ ↩²