Marketplace · Phase 35 · Open Beta

Don't build the team.
Hire it.

The first marketplace structured the way orgs actually compose: plugins snap into workspaces, agents drop into roles, and entire bundles ship as pre-debugged organizations. Three layers, runtime-neutral, hot-swappable.

L1 / CAPABILITY
Plugins
Snap-in tools, guardrails, compliance
17 listed
L2 / WORKSPACE
Agents
Pre-trained, hire-ready roles
14 listed
L3 / TOPOLOGY
Bundles
Whole orgs, debugged and exported
4 listed
Runtime-neutral · 8 supported, infinitely extensible
LangGraphClaude CodeCrewAIAutoGenDeepAgentsHermesGemini CLIOpenClawLangChainPydantic AIBedrock AgentsVercel AILangGraphClaude CodeCrewAIAutoGenDeepAgentsHermesGemini CLIOpenClawLangChainPydantic AIBedrock AgentsVercel AI
L1Capability layer

Plugins. Plug, configure, ship.

A tool, a guardrail, a compliance check, an evaluator. Drops into any workspace in any runtime. The smallest unit of capability you can buy.

12first-party · 5 community
Docs

Doc Sync on Diff

by @molecule

Watches code diffs, regenerates the corresponding docs section, opens a PR. Keeps docs.* and src/* in lockstep.

Claude CodeDeepAgents
★4.73.3k installs
FREE
L2Workspace layer

Agents. Pre-trained, ready to hire.

A whole role, debugged. System prompt, tool stack, memory schema, runtime — all packaged. You drop the agent into your org and it starts working the next morning.

14vetted listings · avg setup 4m
CT
Claude Code

Cross-Border Tax Accountant

by @taxlab.io · Finance & Tax

CA/US/HK personal & corporate tax. Knows treaty positions, FBAR thresholds, departure tax timing. Trained on 4,200+ filed returns and reviewed by CPAs.

Tools
4
Memory
3 scopes
Hires
312
★4.9
$129/mo
ML
LangGraph

Medical Literature Researcher

by @medsynth · Research

PubMed, Cochrane, ClinicalTrials.gov — synthesizes evidence with GRADE methodology. Cites primary sources only; refuses speculation.

Tools
4
Memory
2 scopes
Hires
189
★4.8
$89/mo
S2
CrewAI

SOC 2 Auditor

by @compliance.dev · Legal & Compliance

Continuous SOC 2 Type II monitoring. Runs the 64-control checklist nightly, drafts evidence requests, escalates gaps to your compliance lead.

Tools
4
Memory
2 scopes
Hires
94
★4.9
$199/mo
FP
DeepAgents

Frontend Pull-Request Reviewer

by @hongming · Engineering

React 19 + TS + Tailwind 4. Reviews against your design tokens, flags accessibility regressions, suggests refactors. Configurable rule pack.

Tools
4
Memory
2 scopes
Hires
421
★4.7
$59/mo
OS
AutoGen

Outbound SDR · Cold Call Edition

by @gtmlab · Sales & GTM

Drafts personalized cold sequences from a prospect URL. Knows your ICP, refuses to spam. HITL-gated before any send.

Tools
4
Memory
2 scopes
Hires
267
★4.5
$79/mo
IC
Hermes

Incident Commander

by @reliability.ai · Operations

PagerDuty + Datadog + Sentry. Triages, assembles a war room channel, drafts the postmortem. Knows your runbooks.

Tools
5
Memory
2 scopes
Hires
156
★4.8
$149/mo
View all 14 agents →
L3Topology layer

Bundles. The whole org, debugged.

A complete, version-controlled organization. Every agent trained, every relationship wired, every memory scope tuned. Import the YAML, get a working company by lunch.

4curated · 2 enterprise-grade
L3 · Org Bundle

Indie Engineering Org

by @molecule

A complete software org for a 1–3 person founding team. Frontend, backend, devops, QA, docs, code review, incident response, release management — all wired through HMA and HITL gates.

11 Agents18 Plugins4 memory scopes
~12m setup
$890/mo
L3 · Org Bundle

Regulated Research Lab

by @medsynth

Built for biotech & medtech. Lit review, study design, statistical reviewer, regulatory submission drafting. EU AI Act 12/13/17 baked in. Private deployment ready.

14 Agents23 Plugins6 memory scopes
~25m setup
$2,400/mo
L3 · Org Bundle

Founder's GTM Team

by @gtmlab

A 7-agent team that runs your top-of-funnel: research, outbound, content, demand-gen analytics, CRM hygiene, lifecycle email, social. Founder = single point of approval.

7 Agents9 Plugins3 memory scopes
~6m setup
$390/mo
L3 · Org Bundle

Solo Creator Studio

by @hongming

For YouTube + Xiaohongshu + newsletter operators. Topic research, script drafting, thumbnail copy, bilingual translation, scheduling, analytics. Optimized for one-person ops.

9 Agents12 Plugins3 memory scopes
~8m setup
$220/mo
Why now

Four shifts converging on multi-agent.

01

Agent count is exploding

Teams have 1 agent today, 10+ next year. The bottleneck is no longer the model — it's how agents coordinate.

02

A2A & MCP became standards

Cross-agent protocols are stabilizing. Multi-agent stops being an engineering problem and becomes a product problem.

03

Governance is now a buying gate

EU AI Act, OWASP Agentic Top 10. Audit, HITL, RBAC are no longer optional.

04

Runtimes are commoditizing

Every quarter, a new "best" runtime ships. The platform layer wins — runtime-portable is the only safe bet.

Use cases

One platform, many shapes.

9 agents · 14 plugins

Fintech · Series A

Defined as code. Versionable. Forkable. Marketplace-publishable. The same workspace.yml ships to dev, staging, and prod with the topology intact.

$ molecule deploy ./fintech.yml
topology validated
agents provisioned
A2A routes live
workspace.yml
Platform

Eight pillars we wrote into the PRD.

F1

Role > Task

Model the org chart, not the workflow. Roles inherit tools, memory scope, HITL ladder.

F2

Recursive teams

Agents can spin up sub-teams within their permission envelope. Org grows like a living tree.

F3

Org = Topology

Every workspace is a versioned graph. Diff it, branch it, share it, sell it.

F4

Distributed A2A

Agents speak JSON-RPC 2.0 directly. Platform routes; never a single bottleneck.

F5

4-layer security

Identity · permission · audit · encryption. AES-256-GCM envelopes, WorkOS OAuth.

F6

Tiered HITL

No-approval, single-reviewer, dual-reviewer. Risk-graded, configured per role.

F7

Cross-workspace traces

OpenTelemetry + Langfuse — every call across every agent, in one timeline.

F8

3-layer Marketplace

Plugin · Agent · Org. Two layers more than GitHub Actions, two more than GPT Store.

Where we sit

Not a framework. Not Ops. Not Zapier.

Capability
Molecule
Agent frameworks
LLM Ops
Workflow tools
Multi-agent topology, native
✓
—
✓
partial
Runtime-portable
✓
runtime-locked
—
—
A2A protocol
✓
—
—
—
Three-tier marketplace
✓
—
—
—
Audit trail · EU AI Act ready
✓
partial
—
partial
HITL tiers
✓
—
—
partial
Org template versioning
✓
—
—
—
FAQ

Common questions.

Context pollution and hallucination compound. Specialized agents in their own context windows is how every large organization in human history has worked — and how language models stay sharp.

Swap it. Roles, memory, and topology are abstract. Adapters connect runtimes; nothing in your config changes when you migrate.

Three tiers: Plugin, Agent, Org. Set your price, ship a signed manifest, get paid via Stripe Connect. We take 20%.

Yes — full private deployment with the same Control Plane. Common for regulated industries; data never leaves your tenant.

A2A (JSON-RPC 2.0) for inter-agent calls, MCP for tool exposure, OpenTelemetry for traces.

You built a working agent. Now sell it.

List a plugin, ship an agent, package a whole bundle. Set your price, keep 80%. Molecule handles billing, distribution, runtime swaps, and the long tail of compatibility.

Revenue split (creator)80%
Active buyers14.2k
Avg time to first sale9 days
Top earner / month$11,400

Stop wiring agents.
Start running an organization.

Free during public beta. SaaS or self-hosted. Bring your runtime, bring your team — we'll bring the org.