Agent skill marketplace

Find, install, review, and monetize agent skills.

SkillHub is becoming the operating layer for agent capabilities: searchable skills, explicit permissions, install commands, usage metering, review queues, pricing, and payout operations.

Developer install journey

The developer path proves SkillHub is more than a directory: discovery becomes project install state, runtime governance, logs, cost, and updates.

CurrentMarketplaceNextSkill detail
  1. MarketplaceSearch, filter, and compare trust, runtime, pricing, and publisher signals.
  2. Skill detailReview manifest, permissions, examples, feedback, incidents, and alternatives.
  3. Developer workspaceCreate projects, manage team, billing, webhooks, and buyer requests.
  4. Project commandPin versions, approve policy, create keys, and manage updates.
  5. Runtime testUse REST or MCP through the same policy, budget, subscription, and log path.
Inspect a skill
Live catalog1
Public publishers1
Verified publishers0
Review gatesSchema + Runtime + Human
Money flowLedger before payout
Marketplace evidence chain

The catalog points to the operating system behind it.

Discovery is credible only when a public listing connects to publisher trust, install readiness, runtime governance, and ledger-backed operations.

Live catalog1
Public publishers1
Recorded calls0
Payout-ready publishers0/1
Live rows stay live: empty production data is shown as empty, not replaced by demo supply.
Operating overview

The marketplace already has separate loops for buyers, publishers, and operators.

These signals come from the platform overview API, so the public marketplace can show the working surfaces behind discovery instead of only presenting a catalog.

Developers

Projects return for policies, version changes, budgets, keys, and runtime evidence.

Projects0
Installed skills0
Update inbox0
browser-research-proEvent: new versionSeverity: info
codebase-risk-scannerEvent: owner approval requiredSeverity: high
crm-enrichmentEvent: data policy updateSeverity: medium
Developers

Publishers

Authors return for review feedback, runtime checks, demand, revenue, and payout readiness.

Submitted versions0
Failed checks0
Available balance$0
browser-research-proStage: Pricing approvalNext step: Confirm per-call cap
crm-enrichmentStage: Data policyNext step: Review CRM token scope
codebase-risk-scannerStage: Restricted launchNext step: Owner approval required
Publishers

Platform operators

Review, money, notification, and incident queues protect the marketplace.

Review queue0
Payout review0
Queued notifications0
High-risk filesystem skillScope: codebase-risk-scannerAction: Require owner approval
Unusual payout requestScope: $4,800 requestAction: Hold for finance transfer review
Runtime error spikeScope: browser-research-proAction: Throttle and notify publisher
Platform operators
Why teams come back
Approve and update installed skills per project.Watch cost, failures, latency, incidents, and budget state.Compare better verified alternatives as the catalog grows.Repair review and runtime issues before losing distribution.Turn buyer requests and feedback into new paid versions.Track ledger state and payout readiness after usage lands.
Supplier trust is part of discovery

Every marketplace card now links to the supplier behind the skill. The public directory lets teams compare verified listings, payout readiness, runtime evidence, and active paid inventory before installing.

Browse publishers
Installed skill operations

The marketplace keeps a live path after install.

A useful skill store is not finished when an agent copies a command. Buyers need runtime evidence and spend controls; publishers need feedback, changelog pressure, and payout-ready usage records.

Schema + Runtime + Human
Callable skills1
Recorded calls0
Feedback signals0
Payout-ready publishers0/1
Public contract
Inspect

Manifest, permissions, runtime, price, and publisher profile are visible before install.

Project gate
Install

Teams attach the skill to a project key or MCP server with explicit policy context.

Usage ledger
Invoke

Runtime calls carry typed input, budget checks, success signals, and reviewable output.

Retention loop
Return

Feedback, incidents, changelog updates, and payout review create the next publisher action.

SkillHub Catalog
1 results
Discovery filters
Category
Pricing
Permission risk
Runtime
Verification
Sort
Research

Browser Research

by SkillHub
medium

Research a web topic and return concise findings with source URLs.

verifiedn/a successn/a latency
Recommendation reasons
Verified reviewReviewed permissions
researchbrowsercitations
skillhub install browser-research
After install
Project installPolicy gateRuntime logUsage ledger
FreeVerified
Details
Requested skills

Buyers can request missing skills. Publishers can claim specs, submit builds, and earn bounties or recurring revenue.

Figma change request to Linear issueOpen | 7 days
$600
Shopify product operations skillSpec review | 12 days
$900
Slack incident summarizerMatched | 5 days
$450
Publisher operating flow
01Draft manifest
02Runtime test
03Human review
04Pricing approval
05Public listing
06Usage ledger
07Payout review
Launch requirements
ManifestTyped input/output, runtime, permissions, version, author.
SecurityPermission classification, runtime checks, secret handling, data retention.
MoneyVersioned commission rule, immutable transaction split, payout audit trail.
SupportChangelog, deprecation policy, issue channel, response expectations.