arize-graphql-analytics

Arize GraphQL Analytics

Query and analyze data from the Arize ML observability platform using GraphQL via curl.

---

Workflow

1. Check API Key → 2. Fetch Full Schema → 3. Build Query/Mutation → 4. Execute → 5. Summarize

CRITICAL: Always fetch the full schema FIRST before building any queries. This ensures accuracy.

---

Step 1: Check API Key

echo "${ARIZE_API_KEY:-NOT_SET}"

If NOT_SET

Ask the user for their API key:

"To query the Arize GraphQL API, I need your API key. You can find it in Arize: Settings → API Keys."

Then set it:

export ARIZE_API_KEY="user-provided-key"

Verify:

curl -s -X POST "${ARIZE_GRAPHQL_ENDPOINT:-https://app.arize.com/graphql}" \
  -H "Content-Type: application/json" \
  -H "x-api-key: ${ARIZE_API_KEY}" \
  -d '{"query": "{ __typename }"}'

Expected: {"data":{"__typename":"Query"}}

---

Step 2: Fetch Full Schema (REQUIRED FIRST STEP)

Use this comprehensive introspection query to get the complete schema. This query uses fragments to capture all type information including nested types up to 10 levels deep.

curl -s -X POST "${ARIZE_GRAPHQL_ENDPOINT:-https://app.arize.com/graphql}" \
  -H "Content-Type: application/json" \
  -H "x-api-key: ${ARIZE_API_KEY}" \
  -d @- <<'EOF'
{
  "query": "query IntrospectionQuery { __schema { queryType { name } mutationType { name } subscriptionType { name } types { ...FullType } directives { name description locations args { ...InputValue } } } } fragment FullType on __Type { kind name description fields(includeDeprecated: true) { name description args { ...InputValue } type { ...TypeRef } isDeprecated deprecationReason } inputFields { ...InputValue } interfaces { ...TypeRef } enumValues(includeDeprecated: true) { name description isDeprecated deprecationReason } possibleTypes { ...TypeRef } } fragment InputValue on __InputValue { name description type { ...TypeRef } defaultValue } fragment TypeRef on __Type { kind name ofType { kind name ofType { kind name ofType { kind name ofType { kind name ofType { kind name ofType { kind name ofType { kind name ofType { kind name ofType { kind name } } } } } } } } } }"
}
EOF

Understanding the Schema Response

The introspection returns:

• queryType: Root query type (usually "Query")
• mutationType: Root mutation type (usually "Mutation")
• types: All types in the schema with their fields, arguments, and relationships
• directives: Available directives

Key Type Kinds

Kind	Description	Example
SCALAR	Primitive types	String, Int, ID, Boolean, Float
OBJECT	Complex types with fields	Space, Model, Monitor
INPUT_OBJECT	Input types for mutations	CreateMonitorInput
ENUM	Enumeration types	ModelType, MonitorStatus
LIST	Array wrapper	[Model]
NON_NULL	Required wrapper	String!
INTERFACE	Shared field contract	Node
UNION	One of multiple types	SearchResult

Unwrapping Types

When kind is NON_NULL or LIST, the actual type is in ofType. Keep unwrapping until you reach a named type:

NON_NULL → LIST → NON_NULL → OBJECT(Model)
means: [Model!]!

---

Step 3: Build Query or Mutation

Using the schema, construct the appropriate operation.

For Queries

1. Find the query field in the schema's queryType
2. Check its return type and arguments
3. Follow the Relay connection pattern for lists
4. Use inline fragments for interface/union types

For Mutations

1. Find the mutation in mutationType
2. Get the input argument's type name
3. Look up that INPUT_OBJECT in types
4. Build mutation with all required inputFields

See references/PATTERNS.md for detailed patterns.

---

Step 4: Execute

Simple Query

curl -s -X POST "${ARIZE_GRAPHQL_ENDPOINT:-https://app.arize.com/graphql}" \
  -H "Content-Type: application/json" \
  -H "x-api-key: ${ARIZE_API_KEY}" \
  -d '{"query": "{ spaces { edges { node { id name } } } }"}'

With Variables (for Mutations)

curl -s -X POST "${ARIZE_GRAPHQL_ENDPOINT:-https://app.arize.com/graphql}" \
  -H "Content-Type: application/json" \
  -H "x-api-key: ${ARIZE_API_KEY}" \
  -d @- <<'EOF'
{
  "query": "mutation CreateSpace($input: CreateSpaceInput!) { createSpace(input: $input) { space { id name } } }",
  "variables": {
    "input": {
      "name": "My Space",
      "organizationId": "org-id-here"
    }
  }
}
EOF

---

Step 5: Summarize Results

Parse the JSON response and provide clear insights to the user.

---

Quick Reference: Common Patterns

Relay Connections (Lists)

All lists use edges/node pattern:

{
  spaces {
    edges {
      node { id name }
      cursor
    }
    pageInfo { hasNextPage endCursor }
  }
}

Node Lookup (by ID)

{
  node(id: "BASE64_ID") {
    ... on Space { name }
    ... on Model { name modelType }
  }
}

Pagination

{ spaces(first: 10, after: "cursor") { edges { node { id } } } }

Mutations

mutation Op($input: InputType!) {
  mutationName(input: $input) {
    entity { id }
    errors { field message }
  }
}

---

Query Building Mode

When helping users write queries (without executing):

1. Fetch schema first
2. Identify relevant types from schema
3. Build query following patterns
4. Provide multiple formats:
   • Pretty (readable)
   • Compact (for curl)
   • Full curl command

---

Troubleshooting

Error	Solution
401 Unauthorized	Invalid/expired API key
Field doesn't exist	Check schema for correct field name
Cannot query field on type	Use inline fragment for interfaces
Parse errors	Check JSON escaping in curl

---

References

• references/PATTERNS.md - Detailed GraphQL patterns
• references/EXAMPLES.md - Ready-to-use examples