AGENT_PLATFORM

Give your AI agents document intelligence.

The fz CLI works as an agent skill — your AI agents upload, extract, and search enterprise documents with full citations. Install it as a skill in seconds, or integrate directly with the REST API.

Add the Agent Skill API Reference

AGENT_SKILL_SETUP

Give your agent the fz skill in three steps.

Install the CLI, authenticate, and your AI agent can upload documents, define schemas, and extract structured data.

01INSTALL

brew install fluidzero/tap/fz

Makes fz available as an agent skill

02AUTHENTICATE

fz auth login

Device flow or M2M credentials

03AGENT READY

Your agent can now call fz commands

Upload, define schemas, extract, search

# fz CLI — used by AI agents as a skill
# Install once, then your agent calls these commands.

# Install the CLI (makes fz available as an agent skill)
$ brew install fluidzero/tap/fz

# Authenticate (device flow for setup, M2M for agents)
$ fz auth login

# Agent lists available projects
$ fz projects list -o json

# Agent uploads documents and waits for processing
$ fz documents upload -p <project-id> *.pdf --wait

# Agent defines a schema (required before extraction)
$ fz schemas create -p <project-id> --name "Concrete Specs" \
    --fields "compressive_strength:number" "mix_design:string" "curing_period:string"

# Agent runs extraction — schema is mandatory for structured outputs
$ fz runs create -p <project-id> -s <schema-id> --wait

# Agent searches with citations
$ fz search "concrete strength requirements" -p <project-id>

# Agent pipes JSON output for further reasoning
$ fz search "max wind load" -p <project-id> -o json

DEVELOPER_TOOLS

AVAILABLE

fz CLI (Agent Skill)

brew install fluidzero/tap/fz

The fz CLI is designed to be used by AI agents as a skill. Agents call fz commands to upload documents, define schemas, run extractions, and search with citations.

Agent skillClaude CodeAny LLM agent

AVAILABLE

REST API

OpenAPI 3.1 specification

RESTful endpoints for all platform operations. OAuth 2.0 authentication and JSON responses. Use directly when your agent needs HTTP-level control.

Custom agent backendsDirect integrationWeb applications

PLANNED

MCP Server

Model Context Protocol

Native MCP tools for document search, upload, and extraction. Compatible with Claude Desktop, Cursor, and other MCP-capable agents.

Claude DesktopIDE agentsMCP-compatible tools

PLATFORM_CAPABILITIES

What your agents can do

Upload, define schemas, extract structured data, and search with citations — every capability is available as an fz command your agent can call.

UPLOAD

Document Upload

Multipart S3 uploads with parallel parts and automatic resume. Supports PDF, images, and spreadsheets with server-side processing.

EXTRACT

Schema Extraction

Define a schema before every extraction run — schemas are mandatory for structured outputs. Versioned schema definitions with AI-powered field extraction and quality confidence scores.

Search with Citations

Natural language search across your documents. Every result includes document name, page number, and source excerpt.

EVENTS

Webhooks

Create, test, and monitor webhook endpoints. Signed payloads, delivery history, and configurable retry logic.

AUTH

Authentication

Device flow for interactive use, M2M API keys for AI agents, and OAuth 2.0 token exchange. Scoped permissions per credential.

BATCH

Batch Processing

Process entire directories of documents with configurable batch sizes. JSONL output for downstream pipeline integration.

API Documentation

Complete REST API reference with endpoint docs

CLI on GitHub

Source code, issues, and contribution guide

fluidzero

Document intelligence platform for AI agents. Upload, define schemas, extract structured data, and search enterprise documents with full citations. The fz CLI is designed to be used by AI agents as a skill.

fluidzero gives your AI agents the ability to read, understand, and cite enterprise documents — PDFs, images, and spreadsheets. Agents use the fz CLI as a skill or integrate directly with the REST API.

Quick start

Install the CLI and authenticate:

brew install fluidzero/tap/fz

fz auth login

Your agent can now call fz commands to upload, extract, and search.

Agent skill setup

The fz CLI is designed to be called by AI agents as a skill.

Example SKILL.md for Claude Code:

---

name: document-search

description: Search fluidzero documents with citations

allowed-tools: Bash(fz *)

---

Search documents using the fz CLI:

1. Run: fz search "$ARGUMENTS" -p <project-id> -o json

2. Parse the JSON response for citations

3. Present results with source references

Developer tools

- [fz CLI (Agent Skill)](https://fluidzero.ai/developers#quickstart): brew install fluidzero/tap/fz — AVAILABLE. Agents call fz commands to upload documents, define schemas, run extractions, and search with citations.

- [REST API](https://fluidzero.ai/developers#api): OpenAPI 3.1 specification — AVAILABLE. RESTful endpoints for all platform operations. OAuth 2.0 authentication and JSON responses.

- [MCP Server](https://fluidzero.ai/developers): Model Context Protocol — PLANNED. Native MCP tools for document search, upload, and extraction. Compatible with Claude Desktop, Cursor, and other MCP-capable agents.

CLI commands (agent skill)

# Install

brew install fluidzero/tap/fz

# Authenticate (device flow for setup, M2M for agents)

fz auth login

# List projects

fz projects list -o json

# Upload documents

fz documents upload -p <project-id> *.pdf --wait

# Define a schema (required before extraction)

fz schemas create -p <project-id> --name "Concrete Specs" \

--fields "compressive_strength:number" "mix_design:string" "curing_period:string"

# Run extraction — schema is mandatory for structured outputs

fz runs create -p <project-id> -s <schema-id> --wait

# Search with citations

fz search "concrete strength requirements" -p <project-id>

# JSON output for agent parsing

fz search "max wind load" -p <project-id> -o json

REST API

# Exchange M2M credentials for an access token

curl -X POST https://api.fluidzero.ai/oauth/token \

-d grant_type=client_credentials \

-d client_id=$CLIENT_ID \

-d client_secret=$CLIENT_SECRET

# List projects

curl https://api.fluidzero.ai/v1/projects \

-H "Authorization: Bearer $TOKEN"

# Create a schema (required before running extraction)

curl -X POST https://api.fluidzero.ai/v1/schemas \

-H "Authorization: Bearer $TOKEN" \

-H "Content-Type: application/json" \

-d '{"project_id":"<project-id>","name":"Concrete Specs","fields":[{"name":"compressive_strength","type":"number"},{"name":"mix_design","type":"string"}]}'

# Search with citations

curl -X POST https://api.fluidzero.ai/v1/search \

-H "Authorization: Bearer $TOKEN" \

-H "Content-Type: application/json" \

-d '{"project_id":"<project-id>","query":"concrete strength requirements","include_citations":true}'

Configuration

# ~/.config/fluidzero/config.toml

# Merge order: CLI flags > env vars (FZ_*) > project config > user config

[auth]

method = "device" # "device" | "client-credentials"

[defaults]

project_id = "<project-id>"

output = "text" # "text" | "json" | "table"

[upload]

wait = true

concurrency = 4

[search]

include_citations = true

max_results = 10

Platform capabilities

- Document Upload: Multipart S3 uploads with parallel parts and automatic resume. Supports PDF, images, and spreadsheets with server-side processing.

- Schema Extraction: Define a schema before every extraction run — schemas are mandatory for structured outputs. Versioned schema definitions with AI-powered field extraction and quality confidence scores.

- Search with Citations: Natural language search across your documents. Every result includes document name, page number, and source excerpt.

- Webhooks: Create, test, and monitor webhook endpoints. Signed payloads, delivery history, and configurable retry logic.

- Authentication: Device flow for interactive use, M2M API keys for AI agents, and OAuth 2.0 token exchange. Scoped permissions per credential.

- Batch Processing: Process entire directories of documents with configurable batch sizes. JSONL output for downstream pipeline integration.

Links

- [API Documentation](https://fluidzero.ai/developers#api): Complete REST API reference with endpoint docs

- [Get Started](https://fluidzero.ai/resolve): Create an account and get your API key

- [CLI on GitHub](https://github.com/fluidzero/fz-cli): Source code, issues, and contribution guide

Give your agent the fz skill in three steps.

Install the CLI, authenticate, and your AI agent can upload documents, define schemas, and extract structured data.

01INSTALL

brew install fluidzero/tap/fz

Makes fz available as an agent skill

02AUTHENTICATE

fz auth login

Device flow or M2M credentials

03AGENT READY

Your agent can now call fz commands

Upload, define schemas, extract, search

# fz CLI — used by AI agents as a skill # Install once, then your agent calls these commands. # Install the CLI (makes fz available as an agent skill) $ brew install fluidzero/tap/fz # Authenticate (device flow for setup, M2M for agents) $ fz auth login # Agent lists available projects $ fz projects list -o json # Agent uploads documents and waits for processing $ fz documents upload -p <project-id> *.pdf --wait # Agent defines a schema (required before extraction) $ fz schemas create -p <project-id> --name "Concrete Specs" \ --fields "compressive_strength:number" "mix_design:string" "curing_period:string" # Agent runs extraction — schema is mandatory for structured outputs $ fz runs create -p <project-id> -s <schema-id> --wait # Agent searches with citations $ fz search "concrete strength requirements" -p <project-id> # Agent pipes JSON output for further reasoning $ fz search "max wind load" -p <project-id> -o json

fluidzero

Quick start

Install the CLI and authenticate:

brew install fluidzero/tap/fz

fz auth login

Your agent can now call fz commands to upload, extract, and search.

Agent skill setup

The fz CLI is designed to be called by AI agents as a skill.

Example SKILL.md for Claude Code:

---

name: document-search

description: Search fluidzero documents with citations

allowed-tools: Bash(fz *)

---

Search documents using the fz CLI:

1. Run: fz search "$ARGUMENTS" -p <project-id> -o json

2. Parse the JSON response for citations

3. Present results with source references

Developer tools

- [REST API](https://fluidzero.ai/developers#api): OpenAPI 3.1 specification — AVAILABLE. RESTful endpoints for all platform operations. OAuth 2.0 authentication and JSON responses.

CLI commands (agent skill)

# Install

brew install fluidzero/tap/fz

# Authenticate (device flow for setup, M2M for agents)

fz auth login

# List projects

fz projects list -o json

# Upload documents

fz documents upload -p <project-id> *.pdf --wait

# Define a schema (required before extraction)

fz schemas create -p <project-id> --name "Concrete Specs" \

--fields "compressive_strength:number" "mix_design:string" "curing_period:string"

# Run extraction — schema is mandatory for structured outputs

fz runs create -p <project-id> -s <schema-id> --wait

# Search with citations

fz search "concrete strength requirements" -p <project-id>

# JSON output for agent parsing

fz search "max wind load" -p <project-id> -o json

REST API

# Exchange M2M credentials for an access token

curl -X POST https://api.fluidzero.ai/oauth/token \

-d grant_type=client_credentials \

-d client_id=$CLIENT_ID \

-d client_secret=$CLIENT_SECRET

# List projects

curl https://api.fluidzero.ai/v1/projects \

-H "Authorization: Bearer $TOKEN"

# Create a schema (required before running extraction)

curl -X POST https://api.fluidzero.ai/v1/schemas \

-H "Authorization: Bearer $TOKEN" \

-H "Content-Type: application/json" \

-d '{"project_id":"<project-id>","name":"Concrete Specs","fields":[{"name":"compressive_strength","type":"number"},{"name":"mix_design","type":"string"}]}'

# Search with citations

curl -X POST https://api.fluidzero.ai/v1/search \

-H "Authorization: Bearer $TOKEN" \

-H "Content-Type: application/json" \

-d '{"project_id":"<project-id>","query":"concrete strength requirements","include_citations":true}'

Configuration

# ~/.config/fluidzero/config.toml

# Merge order: CLI flags > env vars (FZ_*) > project config > user config

[auth]

method = "device" # "device" | "client-credentials"

[defaults]

project_id = "<project-id>"

output = "text" # "text" | "json" | "table"

[upload]

wait = true

concurrency = 4

[search]

include_citations = true

max_results = 10

Platform capabilities

- Document Upload: Multipart S3 uploads with parallel parts and automatic resume. Supports PDF, images, and spreadsheets with server-side processing.

- Search with Citations: Natural language search across your documents. Every result includes document name, page number, and source excerpt.

- Webhooks: Create, test, and monitor webhook endpoints. Signed payloads, delivery history, and configurable retry logic.

- Authentication: Device flow for interactive use, M2M API keys for AI agents, and OAuth 2.0 token exchange. Scoped permissions per credential.

- Batch Processing: Process entire directories of documents with configurable batch sizes. JSONL output for downstream pipeline integration.

Links

- [API Documentation](https://fluidzero.ai/developers#api): Complete REST API reference with endpoint docs

- [Get Started](https://fluidzero.ai/resolve): Create an account and get your API key

- [CLI on GitHub](https://github.com/fluidzero/fz-cli): Source code, issues, and contribution guide