Glasser lets Openclaw Skills discover, price, and safely run more than 1,000 third-party data API endpoints through one authenticated Workspace.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install glasser
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install glasser using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Glasser CLI?
Glasser is a paid data API broker for Openclaw Skills that provides a unified way to search and execute external capabilities, including person and company enrichment, web and news search, image and video search, maps, places, scholar, shopping, lead lookup, and webpage scraping. Instead of integrating with every vendor separately, agents search available Endpoints, inspect their input contracts and pricing, then run the selected operation through Glasser.
The skill supports MCP tools by default and an optional Node.js CLI for scripting, CI, machine-readable workflows, and large results. It preserves provider response structures while removing private billing fields, reports exact charges and Run status, and enforces safe operating rules such as user approval before paid calls, small initial volumes, idempotent retries, and transparent Run URLs.
Glasser CLI Use Cases
- Search the web, news, images, videos, maps, places, scholarly sources, or shopping data when no existing free capability or user-owned integration is available.
- Enrich person, company, prospect, or lead records through specialized third-party data providers.
- Scrape or retrieve structured webpage data without building and maintaining a custom scraper first.
- Compare multiple providers and Endpoints for the same capability using relevance, schemas, prices, and charge clauses.
- Run paid external data operations from an AI agent while reporting the provider, result, exact charge, status, and Workspace Run URL.
- Automate repeatable data retrieval in scripts, CI pipelines, scheduled jobs, or unattended environments with GLASSER_API_KEY.
- Handle large provider responses efficiently by writing output to a file and reading only the required sections.
How Glasser CLI Works
- Determine whether Glasser is appropriate. User instructions, existing user-owned keys or integrations, and free environment capabilities take precedence over Glasser.
- Authenticate through the MCP client or CLI. MCP authentication opens a browser-based sign-in and Workspace authorization flow; the CLI uses
glasser login, while unattended jobs can useGLASSER_API_KEY. - Confirm access and available funds with the
balancetool orglasser balance. A successful balance check is the completion signal for setup and authentication. - Search for a capability with
glasser search -qor the MCPsearchtool. Results identify the Provider slug, provider name, Endpoint, relevance ranking, price, and pagination cursor. - Inspect the selected Endpoint before spending money. Use
glasser inspectto read the input schema, supported version, run mode, current Price, charge clauses, and volume-related fields such asnum,size,limit, or query arrays. - Explain the per-call price and obtain user approval before a paid, speculative, bulk, or repeated operation. Start with small result volumes unless the user requests more.
- Execute the operation with
glasser runor the MCPruntool. MCP calls require an agent-generated UUID idempotency key; the CLI generates one unless JSON mode requires an explicit key. - Wait for completion when needed. Use
--wait, or retrieve and poll a queued or running operation withglasser runs get -r <runId> --wait. - Interpret both indicators separately: Run status describes execution, while the provider payload describes the actual data outcome. A completed provider not-found response can be a normal result.
- Report the provider response, Run status, exact Charge amount, and Run URL. If a transport failure leaves the outcome uncertain, retry with the same Idempotency-Key to avoid duplicate billing.
- Manage history, balances, and active operations with
runs list,runs get,runs stop, and balance history commands.
Glasser CLI Setup
MCP setup
The skill includes Glasser MCP tools, so no separate installation is required for the default workflow. Use search, inspect, run, runs_get, runs_list, runs_stop, and balance, then authenticate through the MCP client's browser flow. Never ask users to paste an API key into a conversation.
Optional CLI installation
Install the CLI when results are too large for the context window, when scripting, or when running CI jobs. Glasser requires Node.js 22 or newer.
glasser --version 2>/dev/null || npm install -g @glasser-ai/cli@latest
If the CLI reports an available update, finish the current task and upgrade before the next one:
npm install -g @glasser-ai/cli@latest
Interactive authentication
glasser login
glasser balance
glasser login opens browser-based device authorization when possible and prints a fallback URL and code when necessary. Confirm that the resulting balance applies to the environment used by subsequent commands.
Unattended authentication
Set GLASSER_API_KEY to a key created in the Glasser console, or store a key locally:
export GLASSER_API_KEY="your-key"
glasser keys add --label ci --key "$GLASSER_API_KEY"
glasser balance
Optional configuration includes GLASSER_API_BASE_URL, which defaults to https://api.glasser.ai, and NO_COLOR. The Workspace associated with the configured API stack is billed. Do not block work on an upgrade check if the package registry is unreachable.
Glasser CLI Data Schema & Taxonomy
Endpoint and run model
| Object | Important fields or concepts |
|---|---|
| Provider | Provider slug used with -p, plus the human-readable provider name shown in search results |
| Endpoint | Provider operation selected with -e; exposes an input schema, run mode, compatible versions, Price, and charge clauses |
| Input | JSON supplied inline with -i or from a file with -f; must match the inspected Endpoint schema |
| Run | Execution record with an Idempotency-Key, status, provider payload or failure block, charge, and Run URL |
| Workspace | Authenticated billing and access boundary whose available balance funds Runs |
| Balance | Available funds, held funds, top-ups, and charge ledger information |
Run statuses
QUEUED: accepted but not yet dispatched.RUNNING: dispatched while awaiting the provider.COMPLETED: provider responded; the payload may still indicate no result.FAILED: no usable provider response; the failure block explains why.STOPPED: stopped throughruns stop; a request already dispatched may still complete and be charged.
Files and output modes
- Use
-o <file>for large provider responses; the file preserves the provider payload while the Run URL remains in command output rather than the output file. - Use
-jor--jsonfor machine-readable output. Standard output contains data only, while errors are emitted as one JSON object on standard error. - Inline JSON uses
-i '<json>'; file-based input uses-f <file>. - Money is represented as an exact decimal string such as
0.0125, never a floating-point number. - Search results are paginated, normally showing five rows per page, with a cursor for subsequent pages.
- Endpoint versions select a supported compatible contract but do not lock the Price. New Runs use the selected version's admission price, while accepted Runs retain their original price.
Charge taxonomy
Inspect output is authoritative for normal Price and exception clauses such as NO_RESULT. Volume controls can affect charges, so fields like num, size, limit, and arrays of queries must be reviewed before execution.
Glasser CLI Advanced Features
- Unified MCP and CLI transports expose the same seven core verbs, Workspace, Key, and balance state.
- Endpoint discovery spans more than 1,000 paid operations across multiple providers without requiring separate vendor signups.
- Browser-based device authorization supports interactive users, while environment keys and local key management support CI and unattended automation.
- Idempotency keys make ambiguous retries safe: reuse the exact same key after a timeout or uncertain transport failure to retrieve the original Run.
- Synchronous and asynchronous run modes are supported, with
--wait, polling, Run retrieval, and queued-operation cancellation. - Machine-readable JSON mode provides clean stdout and structured stderr for robust scripts and agent orchestration.
- Large-result handling with
-o <file>prevents provider payloads from overwhelming the agent context window. - Run history, filtering, pagination, balance history, and key activation or removal support operational auditing.
- Transparent pricing controls expose normal Prices, exception charge clauses, exact decimal billing, and provider-specific volume rules before execution.
- Structure-preserving redaction removes private vendor billing fields such as usage counters without renaming or reshaping provider data.
- Rate-limit handling uses the supplied
retry_after_msvalue and permits one controlled retry instead of an unsafe retry loop. - Provider attribution, charge reporting, status interpretation, and Run URLs make external data operations auditable for users and Openclaw Skills.
SKILL.md
Loading
Related Openclaw Skills
haircut
Haircut measures, prices, and safely trims the configuration overhead loaded by every coding agent on a machine.
Movie Night Decider
Movie Night Decider finds and ranks sourced movie recommendations that match your preferences while respecting licensing, approval, and verification rules.
XinJianXue Anxiety and Rumination Advisor
An AI advisor that analyzes anxiety and rumination loops through XinJianXue data and converts them into practical, manageable next steps.
XinJianXue Parent–Child Conflict Advisor
A consent-based XinJianXue advisor that analyzes parent–child conflict patterns and provides neutral, practical ways to de-escalate and communicate.
1688 One-Stop Procurement Assistant
A complete 1688 procurement workflow that takes users from product discovery and supplier comparison to cart management, checkout, and user-approved payment.
Crun Media Agent Skills
Crun Media Agent Skills orchestrates safe, model-routed workflows for generating, editing, and transforming image, video, audio, music, and other media.
Featured*
