Skills API

Use skill endpoints to extend your agents with specialized capabilities.

Base URL

https://agentbot.sh/api/skills

List skills

GET /api/skills

Returns all available skills in the marketplace. No authentication required for basic listing. When a valid session is present and agentId is provided, the response includes which skills are already installed for that user and agent.

Query parameters

Parameter	Type	Description
`category`	string	Filter by category. Use `all` or omit for all skills. Options: `streaming`, `events`, `payments`, `finance`, `productivity`, `communication`, `development`, `channels`, `music`, `creative`, `marketing`, `ai`
`featured`	string	Set to `true` to return only featured skills
`search`	string	Case-insensitive search across skill names and descriptions
`agentId`	string	Agent identifier or managed runtime `openclawInstanceId`. When provided with an authenticated session, the response includes an `installedSkillIds` array indicating which skills are already installed on this agent.

Response

{
  "skills": [
    {
      "id": "dj-streaming",
      "name": "baseFM DJ Streaming",
      "description": "Create baseFM streams, fetch live DJs, and generate ffmpeg broadcaster commands for agent DJs.",
      "category": "streaming",
      "author": "Agentbot",
      "downloads": 150,
      "rating": 4.8,
      "ratingCount": 12,
      "installs": 47,
      "userRating": 5,
      "featured": true,
      "code": "",
      "hasDownload": true,
      "scan": {
        "trustTier": "trusted",
        "riskLevel": "low",
        "installAllowed": true,
        "reasons": [],
        "warnings": [],
        "requiresManualReview": false
      }
    }
  ],
  "categories": [
    "streaming",
    "events",
    "payments",
    "finance",
    "productivity",
    "communication",
    "development",
    "channels",
    "music",
    "creative",
    "marketing",
    "ai"
  ],
  "featured": [],
  "installedSkillIds": ["dj-streaming", "guestlist"]
}

Field	Type	Description
`skills`	array	List of skill objects matching the query filters
`skills[].downloads`	number	Cumulative install counter on the skill record. Maintained for backwards compatibility — prefer `installs` for live engagement counts.
`skills[].rating`	number	Average user-submitted rating, rounded to one decimal place. Derived from real ratings posted via `POST /api/skills/{skillId}/rating`. Returns `0` when no users have rated the skill yet.
`skills[].ratingCount`	number	Number of unique users who have submitted a rating for the skill.
`skills[].installs`	number	Number of currently enabled installations across all users and agents. Computed from live `InstalledSkill` rows rather than the seeded `downloads` counter.
`skills[].userRating`	number \| null	The authenticated caller’s own rating for the skill (1–5), or `null` if they have not rated it. Always `null` when the request is unauthenticated.
`skills[].code`	string	Skill handler code or configuration. Empty string for platform-authored skills without custom code.
`skills[].hasDownload`	boolean	Whether the skill has a downloadable package. `true` when the skill has handler code, an MCP server, or a widget configuration that can be exported via `GET /api/skills/{skillId}/download`. `false` for catalog-only skills that must be installed via the runtime.
`skills[].scan`	object	Marketplace safety scan result for the skill. See scan object for field details.
`categories`	string[]	All distinct skill categories in the marketplace
`featured`	array	Subset of `skills` where `featured` is `true`. Each entry includes the same `rating`, `ratingCount`, `installs`, `userRating`, and `scan` fields.
`installedSkillIds`	string[]	IDs of skills installed and enabled for the authenticated user and the specified `agentId`. Returns an empty array when no session is present or `agentId` is omitted.

The rating, ratingCount, installs, and userRating fields are derived from live engagement records, not the seeded Skill columns. New skills with no user ratings return rating: 0 and ratingCount: 0. If the database is unreachable and the endpoint falls back to the default catalog, all four fields are returned as 0 (or null for userRating).

Install skill

POST /api/skills

Requires session authentication. Installs a skill on an agent.

Request body

Field	Type	Required	Description
`skillId`	string	Yes	ID of the skill to install
`agentId`	string	Yes	ID of the agent to install the skill on. This can be either a database agent ID or a managed runtime’s `openclawInstanceId`.

Managed runtime resolution

When agentId does not match an existing agent row in the database, the endpoint checks whether it corresponds to the authenticated user’s managed OpenClaw runtime (openclawInstanceId). If it does, a backing agent row is automatically created in the database before the skill is installed. This allows you to install skills on managed runtimes that have not yet been explicitly registered as agents. The auto-created agent uses the following defaults:

Field	Value
`name`	`Managed OpenClaw Runtime`
`model`	`openclaw`
`status`	`running`

The agent row is created via an upsert, so subsequent skill installations on the same managed runtime reuse the existing row. The same resolution logic applies to the uninstall endpoint.

Idempotency

Installs are idempotent. Repeated installs of the same skill on the same agent do not surface generic failure errors:

If the skill is already installed and enabled for the agent, the endpoint returns 200 OK with alreadyInstalled: true and does not redeploy.
If a previous install record exists but was disabled (for example, after an earlier uninstall), it is re-enabled in place and the response message reflects that the skill was re-enabled.
If a concurrent install causes a duplicate-key conflict at the database layer, the endpoint returns 409 Conflict with code: "already_installed" and a clear, human-readable message instead of a generic install failure.

Response

When the skill is installed and deployed to the gateway successfully:

{
  "success": true,
  "installed": {
    "id": "inst_123",
    "skillId": "dj-streaming",
    "agentId": "agent_456",
    "installedAt": "2026-03-20T00:00:00Z"
  },
  "runtimeHydrated": false,
  "deployed": true,
  "message": "Skill installed successfully."
}

When a previously disabled install is re-enabled:

{
  "success": true,
  "installed": {
    "id": "inst_123",
    "skillId": "dj-streaming",
    "agentId": "agent_456",
    "installedAt": "2026-03-20T00:00:00Z"
  },
  "runtimeHydrated": false,
  "deployed": true,
  "message": "Skill re-enabled successfully."
}

When the skill is already installed and enabled for the agent:

{
  "success": true,
  "installed": {
    "id": "inst_123",
    "skillId": "dj-streaming",
    "agentId": "agent_456",
    "installedAt": "2026-03-20T00:00:00Z"
  },
  "alreadyInstalled": true,
  "runtimeHydrated": false,
  "deployed": false,
  "message": "This skill is already installed for this agent."
}

When the skill is saved but the agent is offline or the gateway is unreachable:

{
  "success": true,
  "installed": {
    "id": "inst_123",
    "skillId": "dj-streaming",
    "agentId": "agent_456",
    "installedAt": "2026-03-20T00:00:00Z"
  },
  "runtimeHydrated": false,
  "deployed": false,
  "deployWarning": "Gateway unreachable - skill saved to database and will sync automatically",
  "message": "Skill saved to the database and will sync to the runtime automatically."
}

Field	Type	Description
`success`	boolean	Always `true` when the skill is saved to the database
`installed`	object	The installation record
`installed.id`	string	Unique installation identifier
`installed.skillId`	string	ID of the installed skill
`installed.agentId`	string	ID of the agent the skill was installed on
`installed.installedAt`	string	ISO 8601 timestamp of the installation
`alreadyInstalled`	boolean \| undefined	Present and `true` when the skill was already installed and enabled for this agent. The endpoint returns the existing install record without redeploying. Omitted on fresh installs and re-enables.
`runtimeHydrated`	boolean	`true` when the `agentId` resolved to a managed OpenClaw runtime and a backing agent row was auto-created. `false` when the agent already existed in the database. See managed runtime resolution.
`deployed`	boolean	Whether the skill was successfully deployed to the gateway. `false` when the agent is offline, the gateway is unreachable, or the skill was already installed (`alreadyInstalled: true`).
`deployWarning`	string \| undefined	Present only when `deployed` is `false` because of a gateway issue. Describes why the deployment did not complete. The skill is still saved and will sync automatically when the agent comes back online.
`message`	string	Human-readable summary of the install outcome. Varies based on whether the install was new, re-enabled, already present, whether the runtime was hydrated, and whether the gateway deploy succeeded.

Errors

Code	Error code	Description
400	—	Missing `skillId` or `agentId`
400	—	Skill blocked by marketplace safety checks. The response includes a `scan` object with the reasons for the block. See scan object.
401	—	Unauthorized
404	—	Agent not found or skill not found
409	`already_installed`	A duplicate-key conflict was raised when creating the install record. Treat this as a successful install — the skill is already installed for this agent.
500	`install_failed`	Internal error. The response includes a generic install failure message suggesting the user refresh and try again, or open the OpenClaw skills manager for the agent.

Error responses for the 409 and 500 cases include a code field alongside error:

{
  "error": "This skill is already installed for this agent.",
  "code": "already_installed"
}

When a skill is blocked by safety checks, the response looks like this:

{
  "error": "Skill blocked by marketplace safety checks",
  "scan": {
    "trustTier": "blocked",
    "riskLevel": "blocked",
    "installAllowed": false,
    "reasons": ["Dynamic code execution via eval detected"],
    "warnings": [],
    "requiresManualReview": true
  }
}

A skill install is only considered active in the runtime when "deployed": true is returned. A "deployed": false response — whether from a gateway error, an unreachable runtime, or a 2xx body the gateway client rejects (for example, a runtime that returns "success": false) — means the install is saved to the database but has not been accepted by the live OpenClaw runtime. Use the agent sync endpoint to push installed skills to the runtime and confirm they are active.

Create skill

POST /api/skills/create

Requires session authentication. Creates a new custom skill in the marketplace. The skill becomes available to all users once created.

Request body

Field	Type	Required	Description
`name`	string	Yes	Display name of the skill. Maximum 80 characters. Must be unique across the marketplace.
`description`	string	Yes	Summary of the skill’s capabilities. Maximum 600 characters.
`category`	string	No	Category for the skill. Maximum 40 characters. Defaults to `custom` when omitted.
`code`	string	No	Skill handler code or configuration. Maximum 2000 characters.
`sourceUrl`	string	No	URL to the skill’s source code repository or homepage. Maximum 300 characters. Providing a source URL improves the skill’s trust tier during safety scanning.

Response

{
  "success": true,
  "skill": {
    "id": "clx9abc123",
    "name": "My Custom Skill",
    "description": "Analyzes track mix quality",
    "category": "custom",
    "code": "",
    "author": "user@example.com",
    "downloads": 0,
    "rating": 0,
    "featured": false,
    "mcpEnabled": false,
    "mcpConfig": null,
    "createdAt": "2026-04-02T12:00:00.000Z",
    "updatedAt": "2026-04-02T12:00:00.000Z"
  },
  "scan": {
    "trustTier": "review",
    "riskLevel": "medium",
    "installAllowed": true,
    "reasons": [],
    "warnings": ["Source URL missing"],
    "requiresManualReview": true
  }
}

Field	Type	Description
`skill.id`	string	Unique identifier for the created skill
`skill.name`	string	Display name
`skill.description`	string	Skill description
`skill.category`	string	Category (defaults to `custom`)
`skill.code`	string	Skill handler code
`skill.author`	string	Set to the authenticated user’s display name or email
`skill.downloads`	number	Download count, starts at `0`
`skill.rating`	number	Rating score, starts at `0`
`skill.featured`	boolean	Always `false` for user-created skills
`skill.mcpEnabled`	boolean	Whether the skill has a bundled MCP server. Always `false` for user-created skills.
`skill.mcpConfig`	object \| null	MCP server configuration. Always `null` for user-created skills.
`skill.createdAt`	string	ISO 8601 creation timestamp
`skill.updatedAt`	string	ISO 8601 last-updated timestamp
`scan`	object	Marketplace safety scan result. See scan object. Skills that fail the safety scan are rejected before creation.

Errors

Code	Description
400	Missing `name` or `description`, or a skill with the same name already exists
400	Skill blocked by marketplace safety checks. The response includes a `scan` object with the reasons for the block. See scan object.
401	Unauthorized
500	Internal error

Uninstall skill

DELETE /api/skills

Requires session authentication. Removes a skill from an agent.

Request body

Field	Type	Required	Description
`skillId`	string	Yes	ID of the skill to uninstall
`agentId`	string	Yes	ID of the agent to uninstall the skill from. Accepts a managed runtime’s `openclawInstanceId` — see managed runtime resolution above.

Response

{
  "success": true,
  "runtimeHydrated": false,
  "message": "Skill removed successfully."
}

Field	Type	Description
`success`	boolean	Always `true` when the skill is removed from the database
`runtimeHydrated`	boolean	`true` when the `agentId` resolved to a managed OpenClaw runtime. See managed runtime resolution.
`message`	string	Human-readable summary of the uninstall outcome. Varies based on whether the runtime was hydrated.

Errors

Code	Description
400	Missing `skillId` or `agentId`
401	Unauthorized
404	Agent not found
500	Internal error

The uninstall endpoint returns { "success": true } even when no matching installation exists. It does not return a 404 error for missing skill installations, though it does return 404 if the agent itself cannot be found.

When a skill is uninstalled, the updated agent state is automatically synced to the OpenClaw gateway. If the sync fails, the skill is still removed from the database — you can retry the sync manually using the agent sync endpoint.

Download skill

GET /api/skills/{skillId}/download

Requires session authentication. Downloads a skill as a portable JSON manifest. Use this endpoint when a user wants to export a packaged skill (handler code, MCP server config, or widget configuration) instead of installing it directly to a runtime. The response is delivered as a file download with Content-Disposition: attachment and a slugified filename derived from the skill name.

Path parameters

Parameter	Type	Description
`skillId`	string	ID of the skill to download

Response

Returns a JSON manifest with Content-Type: application/json; charset=utf-8 and Cache-Control: no-store:

{
  "schema": "agentbot.skill.v1",
  "exportedAt": "2026-04-29T16:52:02.000Z",
  "skill": {
    "id": "clx9abc123",
    "name": "My Custom Skill",
    "description": "Analyzes track mix quality",
    "category": "custom",
    "author": "user@example.com",
    "code": "// skill handler code",
    "mcpEnabled": false,
    "mcpConfig": null,
    "widgetUrl": null,
    "widgetConfig": null
  }
}

Field	Type	Description
`schema`	string	Manifest schema identifier. Currently always `agentbot.skill.v1`.
`exportedAt`	string	ISO 8601 timestamp when the manifest was generated
`skill`	object	The exported skill payload
`skill.id`	string	Skill identifier
`skill.name`	string	Display name
`skill.description`	string	Skill description
`skill.category`	string	Skill category
`skill.author`	string	Author name
`skill.code`	string	Skill handler code or configuration
`skill.mcpEnabled`	boolean	Whether the skill bundles an MCP server
`skill.mcpConfig`	object \| null	MCP server configuration, or `null` when not present
`skill.widgetUrl`	string \| null	Widget URL, or `null` when not present
`skill.widgetConfig`	object \| null	Widget configuration, or `null` when not present

Response headers

Header	Value
`Content-Type`	`application/json; charset=utf-8`
`Content-Disposition`	`attachment; filename="<slugified-skill-name>.agentbot-skill.json"`
`Cache-Control`	`no-store`

The filename slug is derived from the skill name: lowercased, non-alphanumeric characters replaced with -, leading and trailing - trimmed, and capped at 80 characters. Skills with names that slugify to an empty string fall back to agentbot-skill.

Side effects

Each successful download increments the skill’s downloads counter by 1. The increment is best-effort and does not fail the download if the database update errors.

This endpoint returns the package only when the skill has at least one downloadable component: non-empty handler code, an MCP server (mcpEnabled: true), a widgetUrl, or a widgetConfig. Catalog-only skills without any of these fields return 404 with a message directing the caller to use install instead. Check the hasDownload field on the list skills response to determine whether a skill supports download.

Errors

Code	Description
401	`Sign in to download skills` — no authenticated session
404	Skill not found
404	`This skill does not have a downloadable package yet. Use Install to sync it to your runtime.` — the skill exists but has no exportable code, MCP config, or widget

Rate skill

POST /api/skills/{skillId}/rating

Requires session authentication. Submits or updates the authenticated user’s rating for a skill. Each user has at most one rating per skill — re-posting overwrites the previous rating in place. After the rating is recorded, the skill’s average rating is recomputed from all SkillRating rows and persisted on the skill record so it is reflected immediately in subsequent GET /api/skills responses.

Path parameters

Parameter	Type	Description
`skillId`	string	ID of the skill to rate

Request body

Field	Type	Required	Description
`rating`	number	Yes	Whole number from `1` to `5`. Decimal values, values outside this range, and non-numeric values are rejected.

Response

{
  "success": true,
  "skillId": "clx9abc123",
  "userRating": 5,
  "rating": 4.8,
  "ratingCount": 12
}

Field	Type	Description
`success`	boolean	Always `true` on a successful rating
`skillId`	string	The skill that was rated
`userRating`	number	The rating the authenticated user just submitted (1–5)
`rating`	number	New average rating across all users, rounded to one decimal place. Returns `0` when no ratings exist (this can only happen if the rating was rejected before reaching this point).
`ratingCount`	number	Total number of unique users who have rated the skill, including the current submission

Errors

Code	Description
400	`rating` is missing, not an integer, or outside the range `1`–`5`
401	Unauthorized — sign in to rate skills
404	Skill not found
500	Failed to save rating

Verify skill

POST /api/skills/verify

Runs a marketplace safety scan on a skill without creating or installing it. Use this endpoint to check whether a skill would pass safety checks before submitting it to the marketplace. No authentication required.

Request body

Field	Type	Required	Description
`name`	string	Yes	Display name of the skill
`description`	string	No	Skill description
`code`	string	No	Skill handler code or configuration
`author`	string	No	Author name
`featured`	boolean	No	Whether the skill is featured
`sourceUrl`	string	No	URL to the skill’s source repository

Response

{
  "scan": {
    "trustTier": "verified",
    "riskLevel": "low",
    "installAllowed": true,
    "reasons": [],
    "warnings": ["Network access present"],
    "requiresManualReview": false
  }
}

Field	Type	Description
`scan`	object	The safety scan result. See scan object.

Errors

Code	Description
500	Internal error

Scan object

Every skill listed, created, or verified includes a scan object from the marketplace safety scanner. The scanner performs static analysis on skill code to detect potentially dangerous patterns.

Field	Type	Description
`trustTier`	string	One of `trusted`, `verified`, `review`, or `blocked`. See trust tiers.
`riskLevel`	string	One of `low`, `medium`, `high`, or `blocked`
`installAllowed`	boolean	Whether the skill can be installed. `false` when the skill is blocked.
`reasons`	string[]	List of reasons the skill was blocked. Empty when `trustTier` is not `blocked`.
`warnings`	string[]	Informational warnings about patterns found in the skill code (for example, network access or filesystem access). These do not prevent installation.
`requiresManualReview`	boolean	Whether the skill should be reviewed by a human before being promoted or sold

Trust tiers

Tier	Description
`trusted`	Reserved for platform-authored skills and manually approved partners. Auto-install allowed and eligible for featured placement.
`verified`	User-created skills with a source URL that pass static checks with minimal warnings. Install allowed, visible trust badge, eligible for paid distribution.
`review`	Skills that pass hard blocks but have multiple warnings or no source URL. Discoverable and installable, but not promoted or eligible for paid distribution until reviewed.
`blocked`	Skills that match dangerous code patterns (shell injection, eval, process execution, environment variable scraping, destructive commands, or direct IP endpoints). Cannot be created or installed.

Blocked patterns

The following patterns in skill code cause a skill to be blocked:

Piped shell execution (curl ... | bash)
Dynamic code execution (eval(), Function())
Process execution imports or calls (child_process, exec, spawn)
Destructive filesystem commands (rm -rf)
Direct environment variable access (process.env)
Direct IP-based remote endpoints

Available skills

Streaming

Skill	Endpoint	Description
baseFM DJ Streaming	`/dj-streaming`	Create baseFM streams, fetch live DJs, and generate ffmpeg broadcaster commands for autonomous agent DJs

Events

Skill	Endpoint	Description
Guestlist Manager	`/guestlist`	Manage event guestlists, RSVPs, check-ins, and capacity limits
Event Ticketing	`/event-ticketing`	Sell tickets with USDC payments on Base via x402 protocol
Event Scheduler	`/event-scheduler`	Schedule events across Telegram, Discord, WhatsApp, and email with recurring support
Venue Finder	`/venue-finder`	Find venues worldwide with capacity and price filters
Festival Finder	`/festival-finder`	Discover festivals globally, compare lineups, and get recommendations

Payments

Skill	Endpoint	Description
USDC Payments	`/usdc-payments`	Accept USDC payments on Base, generate payment links, and track transactions
Booking Settlement	`/booking-settlement`	Escrow and split execution for booking payments with auto-release on gig completion
Instant Split	`/instant-split`	Execute royalty splits instantly in USDC on Base with configurable thresholds

Finance

Skill	Endpoint	Description
Community Treasury	`/treasury`	Track spending, reimbursements, and multi-sig treasury management
Royalty Tracker	`/royalty-tracker`	Track streaming royalties across platforms in USDC
Crypto Price Alerts	`/crypto-price-alerts`	Monitor crypto prices and send alerts via Telegram or Discord when thresholds are hit
DeFi Portfolio	`/defi-portfolio`	Track wallet holdings, LP positions, and yield farming across Base and Ethereum
Invoice Generator	`/invoice-generator`	Create and send invoices in USDC with payment tracking and reminders

Productivity

Skill	Endpoint	Description
Google Calendar	`/google-calendar`	Schedule events, manage availability, and set reminders with full Google Calendar sync. See the Calendar API for detailed endpoint documentation.
File Manager	`/file-manager`	Upload, download, and organize files with local storage integration
Google Workspace	`/google-workspace`	Gmail, Calendar, Drive, and Sheets integration
Notion	`/notion`	Sync with Notion databases, pages, and workflows
CRM Helper	`/crm-helper`	Track leads, follow-ups, and customer interactions across channels
Meeting Notes	`/meeting-notes`	Auto-generate meeting notes from Zoom and Google Meet transcripts

Communication

Skill	Endpoint	Description
Email	`/email`	Send and receive emails with newsletter support

Development

Skill	Endpoint	Description
Webhooks	`/webhook`	Connect to any API with HTTP requests, webhooks, and integrations
Browser Automation	`/browser`	Browse websites, fill forms, and scrape data autonomously
Chat SDK	`/chat-sdk`	Multi-platform bot SDK for building on Slack, Teams, Discord, Google Chat, GitHub, and Linear with a single TypeScript SDK
Sentry CLI	`/sentry-cli`	Production error monitoring with issue management, log streaming, distributed tracing, and Sentry API access
Docker Containers	`/docker-containers`	Best practices for agent container isolation, persistent state, secret management, health checks, and resource limits
Stateful Agents	`/stateful-agents`	Persistent state management with Prisma, real-time agent coordination, scheduled tasks, and Drizzle ORM migrations
Deploy CLI	`/deploy-cli`	CLI reference for agent provisioning, secrets management, log streaming, and deployment troubleshooting
Code Review	`/code-review`	Review agent code against production best practices covering security, state management, error handling, and anti-pattern detection

Channels

Skill	Endpoint	Description
Telegram	`/telegram`	Connect via Telegram with bot commands, messages, and groups
Discord	`/discord`	Connect via Discord with slash commands, embeds, and voice channels
WhatsApp	`/whatsapp`	Connect via WhatsApp with message templates, media, and status updates
WhatsApp Business	`/whatsapp-business`	Full WhatsApp Business API with automated replies, labels, and catalogs
Slack	`/slack`	Post to channels, create threads, and handle slash commands

Music

Skill	Endpoint	Description
Track Archaeologist	`/track-archaeologist`	Deep catalog search via BlockDB similarity search
Setlist Oracle	`/setlist-oracle`	BPM, key, and energy analysis to build DJ sets with Camelot mixing
Demo Submitter	`/demo-submitter`	Submit demos to Base FM for airplay consideration
Spotify Analytics	`/spotify-analytics`	Track streams, followers, and playlist placements with a cross-platform analytics dashboard
SoundCloud Manager	`/soundcloud-manager`	Upload tracks, manage likes, track reposts, and analyze audience demographics
Bandcamp Sync	`/bandcamp-sync`	Sync releases, track sales, and manage merchandise across Bandcamp

Creative

Skill	Endpoint	Description
Visual Synthesizer	`/visual-synthesizer`	Generate release artwork and social media assets using Stable Diffusion XL
AI Image Generator	`/ai-image-generator`	Generate images via Stable Diffusion, DALL-E, or Midjourney with batch processing
Video Editor	`/video-editor`	Auto-edit video clips with transitions, captions, and background music
Podcast Producer	`/podcast-producer`	Auto-edit podcasts, remove filler words, and add intro/outro music
Video Generator	`/video-generator`	Generate videos via xAI Grok, Runway, or Wan for social media and marketing
Music Generator	`/music-generator`	Create original music with Google Lyria or MiniMax with async generation and follow-up delivery
ComfyUI Workflows	`/comfyui-workflows`	Run ComfyUI workflows locally or on Comfy Cloud for image, video, and music generation

Marketing

Skill	Endpoint	Description
Groupie Manager	`/groupie-manager`	Fan segmentation, lifecycle tracking, and automated merch drop campaigns
Content Calendar	`/content-calendar`	Plan and schedule social media posts across all platforms
SEO Analyzer	`/seo-analyzer`	Analyze website SEO, suggest keywords, and audit backlinks
Affiliate Tracker	`/affiliate-tracker`	Track affiliate links, clicks, and conversions across networks

AI

Skill	Endpoint	Description
Qwen AI	`/qwen-ai`	Use Qwen models for chat, reasoning, and tool calling with fast inference
Fireworks AI	`/fireworks-ai`	Access Fireworks AI models for generation and reasoning with high-throughput inference
Bedrock Mantle	`/bedrock-mantle`	Use Amazon Bedrock Mantle models with automatic inference profile discovery

Production availability

The following skill routes are demo-only and are disabled in production by default:

Booking Settlement (/booking-settlement)
Demo Submitter (/demo-submitter)
Event Scheduler (/event-scheduler)
Event Ticketing (/event-ticketing)
Festival Finder (/festival-finder)
Groupie Manager (/groupie-manager)
Instant Split (/instant-split)
Royalty Tracker (/royalty-tracker)
Setlist Oracle (/setlist-oracle)
Track Archaeologist (/track-archaeologist)
Venue Finder (/venue-finder)

When called in production without the ENABLE_DEMO_SKILLS environment variable set to true, these routes return 501 with the following response:

{
  "error": "disabled_in_production",
  "message": "{skill-name} is a demo route and is disabled in production."
}

To enable demo skill routes in production, set the ENABLE_DEMO_SKILLS=true environment variable. In non-production environments, all skill routes are available without this flag.

Use a skill

POST /api/skills/{skill-name}

Visual Synthesizer

{
  "prompt": "dark techno album cover",
  "style": "industrial"
}

Track Archaeologist

{
  "action": "search",
  "bpm": 128,
  "genre": "techno",
  "mood": "dark"
}

Setlist Oracle

{
  "action": "build",
  "genre": "house",
  "duration": 120
}

Groupie Manager

{
  "action": "segment"
}

Royalty Tracker

{
  "action": "total"
}

Demo Submitter

{
  "action": "submit",
  "title": "My Track",
  "artist": "My Name"
}

Event Ticketing

{
  "action": "purchase",
  "eventId": "e1",
  "email": "fan@example.com",
  "tier": "vip"
}

Event Scheduler

{
  "action": "schedule",
  "title": "Newsletter",
  "date": "2026-03-20",
  "time": "18:00",
  "channels": ["telegram", "email"]
}

Venue Finder

{
  "action": "search",
  "city": "London",
  "type": "underground",
  "capacity": 200,
  "maxPrice": 500
}

Festival Finder

{
  "action": "search",
  "genre": "techno",
  "country": "UK",
  "maxPrice": 350
}

Booking Settlement

Manage booking escrow, fund releases, and settlement simulations. Actions: list, get, create_escrow, release_funds, simulate_settlement

{
  "action": "create_escrow",
  "bookingId": "bk_1"
}

{
  "action": "release_funds",
  "bookingId": "bk_1"
}

{
  "action": "simulate_settlement",
  "booking": {
    "guarantee": 500,
    "backend": 200,
    "deposit": 100
  }
}

Instant Split

Execute royalty splits instantly in USDC on Base. Actions: list_pending, create_split_rule, execute_split, get_balance, simulate

{
  "action": "execute_split",
  "splitId": "split_1"
}

{
  "action": "create_split_rule",
  "splits": {
    "recipientAddress": "0x...",
    "percentage": 50,
    "role": "artist",
    "name": "Artist"
  },
  "threshold": 50
}

{
  "action": "simulate",
  "streams": 100000,
  "rate": 0.003
}

Response format

{
  "success": true
}

Each skill returns additional data specific to its function alongside the success field.

Errors

Code	Description
400	Invalid request parameters
404	Skill not found
500	Internal error
501	Demo skill route is disabled in production. Set `ENABLE_DEMO_SKILLS=true` to enable. See production availability.

​Skills API

​Base URL

​List skills

​Query parameters

​Response

​Install skill

​Request body

​Managed runtime resolution

​Idempotency

​Response

​Errors

​Create skill

​Request body

​Response

​Errors

​Uninstall skill

​Request body

​Response

​Errors

​Download skill

​Path parameters

​Response

​Response headers

​Side effects

​Errors

​Rate skill

​Path parameters

​Request body

​Response

​Errors

​Verify skill

​Request body

​Response

​Errors

​Scan object

​Trust tiers

​Blocked patterns

​Available skills

​Streaming

​Events

​Payments

​Finance

​Productivity

​Communication

​Development

​Channels

​Music

​Creative

​Marketing

​AI

​Production availability

​Use a skill

​Visual Synthesizer

​Track Archaeologist

​Setlist Oracle

​Groupie Manager

​Royalty Tracker

​Demo Submitter

​Event Ticketing

​Event Scheduler

​Venue Finder

​Festival Finder

​Booking Settlement

​Instant Split

​Response format

​Errors

Skills API

Base URL

List skills

Query parameters

Response

Install skill

Request body

Managed runtime resolution

Idempotency

Response

Errors

Create skill

Request body

Response

Errors

Uninstall skill

Request body

Response

Errors

Download skill

Path parameters

Response

Response headers

Side effects

Errors

Rate skill

Path parameters

Request body

Response

Errors

Verify skill

Request body

Response

Errors

Scan object

Trust tiers

Blocked patterns

Available skills

Streaming

Events

Payments

Finance

Productivity

Communication

Development

Channels

Music

Creative

Marketing

AI

Production availability

Use a skill

Visual Synthesizer

Track Archaeologist

Setlist Oracle

Groupie Manager

Royalty Tracker

Demo Submitter

Event Ticketing

Event Scheduler

Venue Finder

Festival Finder

Booking Settlement

Instant Split

Response format

Errors