API usage examples

Detailed examples for common NAAS API operations.

API v1 sunset: 2027-01-01

The /v1/* routes and legacy unversioned aliases (/send_command, /send_config) are deprecated and will be removed in NAAS v3.0. All examples below use /v2/ routes. If you have existing integrations on /v1/, see Upgrading to v2.0 for the migration guide.

The /healthcheck endpoint (unversioned) is not deprecated — it remains as a permanent operational endpoint.

Breaking change in v1.3

The delay_factor parameter was replaced with read_timeout (float, seconds). Migrate by converting: delay_factor=2 → read_timeout=60.0 (approximate).

Authentication

NAAS supports two authentication methods: HTTP Basic and API key (Bearer JWT).

Basic Authentication

Deprecation notice

Basic auth will be disabled by default in v3.0. Migrate to API key authentication for programmatic access. Basic auth will remain available as an opt-in configuration option.

The original auth method. Credentials are passed through to the network device.

# Using curl with -u flag
curl -k -u "username:password" https://localhost:8443/v2/healthcheck

# Using Authorization header
curl -k -H "Authorization: Basic $(echo -n 'username:password' | base64)" \
  https://localhost:8443/v2/healthcheck

Important: Always use HTTPS. Credentials are transmitted to the network device.

API Key Authentication

API keys are JWTs that authenticate the caller to NAAS independently of device credentials. Device credentials are provided in the request body instead of the Authorization header.

# Send a command using an API key
curl -k -X POST https://localhost:8443/v2/send-command \
  -H "Authorization: Bearer eyJhbG..." \
  -H "Content-Type: application/json" \
  -d '{
    "host": "192.168.1.1",
    "platform": "cisco_ios",
    "commands": ["show version"],
    "username": "admin",
    "password": "device_password"
  }'

When using API key auth, username and password are required in the request body. enable is optional (defaults to the password value).

Key differences from Basic auth:

API key identifies the caller; device credentials are separate
API key users are not subject to TACACS lockout
Keys embed role and context claims (for future RBAC support)

API Key Management

Create, list, and revoke API keys. Keys are returned once at creation — store them securely.

All key management endpoints require Basic auth — API keys cannot be used to manage other keys. The Basic auth password must match the NAAS_ADMIN_SECRET environment variable. If NAAS_ADMIN_SECRET is not set, key management is disabled.

Create a Key

curl -k -X POST https://localhost:8443/v2/api-keys \
  -u "admin:$NAAS_ADMIN_SECRET" \
  -H "Content-Type: application/json" \
  -d '{"role": "admin", "ttl": 7776000}'

Response:

{
  "key_id": "k-a1b2c3d4e5f6",
  "token": "eyJhbG...",
  "role": "admin",
  "contexts": ["*"],
  "expires_at": "2026-07-04T00:00:00Z"
}

Parameter	Default	Description
`role`	`admin`	Role: `admin`, `operator`, or `viewer`
`contexts`	`["*"]`	Allowed routing contexts
`ttl`	`7776000` (90 days)	Expiration in seconds, `0` for no expiry
`created_by`	`system`	Creator identity (for audit)

List Keys

curl -k -u "admin:$NAAS_ADMIN_SECRET" https://localhost:8443/v2/api-keys

Returns metadata only — tokens are never shown after creation.

Revoke a Key

curl -k -X DELETE -u "admin:$NAAS_ADMIN_SECRET" \
  https://localhost:8443/v2/api-keys/k-a1b2c3d4e5f6

Revoked keys are immediately rejected on subsequent requests.

Send Command

Execute show commands on network devices.

Enqueue Response (202 Accepted)

{
  "job_id": "550e8400-e29b-41d4-a716-446655440000",
  "message": "Job enqueued",
  "queue_position": 1,
  "enqueued_at": "2026-03-20T19:00:00+00:00",
  "timeout": 60
}

queue_position — approximate position in queue (1 = next to run, may change as jobs complete)
enqueued_at — ISO 8601 timestamp when job was accepted
timeout — maximum seconds the job will run before being killed

Basic Example

curl -k -X POST https://localhost:8443/v2/send-command \
  -u "admin:password" \
  -H "Content-Type: application/json" \
  -d '{
    "host": "192.168.1.1",
    "platform": "cisco_ios",
    "commands": ["show version", "show ip interface brief"]
  }'

With Custom Port

curl -k -X POST https://localhost:8443/v2/send-command \
  -u "admin:password" \
  -H "Content-Type: application/json" \
  -d '{
    "host": "192.168.1.1",
    "port": 2222,
    "platform": "cisco_ios",
    "commands": ["show version"]
  }'

With Enable Password

curl -k -X POST https://localhost:8443/v2/send-command \
  -u "admin:password" \
  -H "Content-Type: application/json" \
  -d '{
    "host": "192.168.1.1",
    "platform": "cisco_ios",
    "enable": "enable_password",
    "commands": ["show running-config"]
  }'

With Custom Request ID

Track your requests with custom IDs:

curl -k -X POST https://localhost:8443/v2/send-command \
  -u "admin:password" \
  -H "Content-Type: application/json" \
  -H "X-Request-ID: my-custom-id-12345" \
  -d '{
    "host": "192.168.1.1",
    "platform": "cisco_ios",
    "commands": ["show version"]
  }'

With Custom Prompt Matching

Use expect_string to override automatic prompt detection with a regex pattern:

curl -k -X POST https://localhost:8443/v2/send-command \
  -u "admin:password" \
  -H "Content-Type: application/json" \
  -d '{
    "host": "192.168.1.1",
    "platform": "cisco_ios",
    "commands": ["show version"],
    "expect_string": "router.*#"
  }'

Use cases:

Non-standard prompts that Netmiko doesn't detect
Commands that change the prompt temporarily
Devices with custom prompt formats

Note: This is an advanced feature. Most users should rely on automatic prompt detection.

Connection Timeout

Control the TCP connection timeout with conn_timeout (default: 10.0 seconds):

curl -k -X POST https://localhost:8443/v2/send-command \
  -u "admin:password" \
  -H "Content-Type: application/json" \
  -d '{
    "host": "192.168.1.1",
    "platform": "cisco_ios",
    "commands": ["show version"],
    "conn_timeout": 5.0
  }'

conn_timeout — seconds to wait for the TCP connection to be established before failing
read_timeout — seconds to wait for a command response after connection is established (default: 30.0)

Reduce conn_timeout for faster failure detection on unreachable hosts. Increase it for devices on high-latency links.

Supported Platforms

NAAS supports all Netmiko platforms:

cisco_ios - Cisco IOS
cisco_nxos - Cisco NX-OS
arista_eos - Arista EOS
juniper_junos - Juniper Junos
hp_procurve - HP ProCurve
And many more...

Send Configuration

Push configuration changes to devices.

Basic Configuration

curl -k -X POST https://localhost:8443/v2/send-config \
  -u "admin:password" \
  -H "Content-Type: application/json" \
  -d '{
    "host": "192.168.1.1",
    "platform": "cisco_ios",
    "commands": [
      "interface GigabitEthernet0/1",
      "description Uplink to Core",
      "no shutdown"
    ]
  }'

With Save Config

Automatically save configuration after changes:

curl -k -X POST https://localhost:8443/v2/send-config \
  -u "admin:password" \
  -H "Content-Type: application/json" \
  -d '{
    "host": "192.168.1.1",
    "platform": "cisco_ios",
    "commands": [
      "interface GigabitEthernet0/1",
      "description Configured via NAAS"
    ],
    "save_config": true
  }'

With Commit (Juniper)

For platforms that require commit:

curl -k -X POST https://localhost:8443/v2/send-config \
  -u "admin:password" \
  -H "Content-Type: application/json" \
  -d '{
    "host": "192.168.1.1",
    "platform": "juniper_junos",
    "commands": [
      "set interfaces ge-0/0/1 description \"Configured via NAAS\""
    ],
    "commit": true
  }'

Job Deduplication

NAAS automatically deduplicates in-flight jobs. If you submit the same command to the same device while a previous identical job is still running or queued, NAAS returns the existing job_id instead of enqueuing a new one.

Two jobs are considered duplicates when they share the same host, platform, commands, and username.

Detecting a Deduplicated Response

{
  "job_id": "abc-123",
  "message": "Job enqueued",
  "deduplicated": true,
  "queue_position": 0,
  "enqueued_at": "2026-03-24T18:00:00+00:00",
  "timeout": 60
}

When deduplicated: true, the job_id refers to the existing in-flight job. Poll it normally to get results.

Disabling Deduplication

Set JOB_DEDUP_ENABLED=false to disable deduplication globally. Useful for testing or when you intentionally want parallel identical jobs.

Idempotency Keys

For client-controlled deduplication (e.g., safe retries on network failure), use X-Idempotency-Key:

curl -k -X POST https://localhost:8443/v2/send-command \
  -H "X-Idempotency-Key: my-unique-key-abc123" \
  -u "admin:password" \
  -d '{"host": "192.168.1.1", "platform": "cisco_ios", "commands": ["show version"]}'

Repeat requests with the same key within 24 hours return the original job_id with idempotent: true. Unlike server-side dedup, idempotency keys are opt-in and key-based (not content-based).

Webhooks

Instead of polling for results, you can provide a webhook_url to receive a notification when the job completes.

curl -k -u admin:admin -X POST https://naas.example.com/v2/send-command \
  -H "Content-Type: application/json" \
  -d '{
    "host": "192.168.1.1",
    "platform": "cisco_ios",
    "commands": ["show version"],
    "webhook_url": "https://my-app.example.com/naas-callback"
  }'

When the job finishes (success or failure), NAAS POSTs a notification to your URL:

{
  "job_id": "abc-123",
  "status": "finished",
  "enqueued_at": "2026-03-20T19:00:00+00:00",
  "completed_at": "2026-03-20T19:00:05+00:00"
}

Use job_id to fetch the full results from GET /v2/jobs/{job_id}.

Security

webhook_url must be HTTPS (HTTP is rejected)
The payload contains only job metadata — results and credentials are never included
Your webhook endpoint must be reachable from the NAAS worker network
Webhook delivery is fire-and-forget: failures are logged but do not affect job status
No retries in v1.4 (tracked in #278)

Dead Letter Queue

Failed jobs are retained in RQ's FailedJobRegistry for JOB_TTL_FAILED (7 days default). Use these endpoints to inspect and replay them.

List Failed Jobs

curl -k -u "admin:password" https://naas.example.com/v2/jobs/failed

{
  "jobs": [
    {
      "job_id": "abc-123",
      "host": "192.168.1.1",
      "platform": "cisco_ios",
      "port": 22,
      "failed_at": "2026-03-20T19:00:00+00:00",
      "error": "NetMikoTimeoutException: timed out",
      "func": "naas.library.netmiko_lib.netmiko_send_command"
    }
  ],
  "total": 1
}

Security: Credentials are never included in the response. Error messages have credential values redacted.

Replay a Failed Job

Re-enqueue a failed job using your current credentials (stored credentials are never used):

curl -k -u "admin:password" -X POST \
  https://naas.example.com/v2/jobs/abc-123/replay

Returns a standard 202 Accepted with a new job_id. The replayed job uses the caller's credentials, not the original submitter's.

Note: Only the original submitter can replay a job (same auth check as job results).

Configuration

Variable	Default	Description
`FAILED_JOB_MAX_RETAIN`	`500`	Maximum number of failed jobs to retain in the registry

Job Cancellation

Cancel running or queued jobs using DELETE.

Cancel a Job

curl -k -X DELETE https://localhost:8443/v2/send-command/{job_id} \
  -u "admin:password"

Response codes:

204 No Content - Job cancelled successfully
404 Not Found - Job ID doesn't exist
409 Conflict - Job already finished (cannot cancel)

Example

# Submit a job
JOB_ID=$(curl -k -X POST https://localhost:8443/v2/send-command \
  -u "admin:password" \
  -H "Content-Type: application/json" \
  -d '{"host": "192.168.1.1", "platform": "cisco_ios", "commands": ["show version"]}' \
  | jq -r '.job_id')

# Cancel it
curl -k -X DELETE https://localhost:8443/v2/send-command/$JOB_ID \
  -u "admin:password"

Use cases:

Cancel long-running commands that are no longer needed
Clean up queued jobs during maintenance
Stop jobs targeting unreachable devices

Limitations:

Cannot cancel jobs that have already completed
Cancellation is best-effort (job may complete before cancellation)

Job Status and Results

Check Job Status

curl -k https://localhost:8443/v2/send-command/550e8400-e29b-41d4-a716-446655440000 \
  -u "admin:password"

The X-Request-ID header in the 202 response contains the job ID:

# Capture job ID from response header
JOB_ID=$(curl -k -s -D - -X POST https://localhost:8443/v2/send-command \
  -u "admin:password" -H "Content-Type: application/json" \
  -d '{"host": "192.168.1.1", "platform": "cisco_ios", "commands": ["show version"]}' \
  | grep -i x-request-id | awk '{print $2}' | tr -d '\r')

Job States

Queued:

{
  "job_id": "550e8400-e29b-41d4-a716-446655440000",
  "status": "queued",
  "enqueued_at": "2026-02-22T19:00:00Z"
}

Started:

{
  "job_id": "550e8400-e29b-41d4-a716-446655440000",
  "status": "started",
  "enqueued_at": "2026-02-22T19:00:00Z",
  "started_at": "2026-02-22T19:00:01Z"
}

Finished:

{
  "job_id": "550e8400-e29b-41d4-a716-446655440000",
  "status": "finished",
  "result": "Command output here...",
  "enqueued_at": "2026-02-22T19:00:00Z",
  "started_at": "2026-02-22T19:00:01Z",
  "ended_at": "2026-02-22T19:00:05Z"
}

Failed:

{
  "job_id": "550e8400-e29b-41d4-a716-446655440000",
  "status": "failed",
  "error": "Authentication failed",
  "enqueued_at": "2026-02-22T19:00:00Z",
  "started_at": "2026-02-22T19:00:01Z",
  "ended_at": "2026-02-22T19:00:02Z"
}

SSE Streaming

Stream real-time job status updates via Server-Sent Events instead of polling.

Endpoint: GET /v2/jobs/<job_id>/stream

Auth: Same as job results (basic auth, must own the job)

curl -N -k -u admin:admin https://localhost:8443/v2/jobs/abc-123/stream

Event types:

Event	When	Data
`status`	Job status changes (queued → started)	`{"job_id": "...", "status": "started"}`
`result`	Job reaches terminal state	`{"job_id": "...", "status": "finished", "results": {...}}`
`timeout`	Server-side stream timeout (5 min)	`{"job_id": "...", "message": "Stream timeout"}`

Example event stream:

event: status
data: {"job_id": "abc-123", "status": "queued"}

event: status
data: {"job_id": "abc-123", "status": "started"}

event: result
data: {"job_id": "abc-123", "status": "finished", "results": {"show version": "..."}}

Limits:

Max 100 concurrent SSE connections per API process
Server-side timeout: 5 minutes
Returns 429 if connection limit is reached

When to use SSE vs polling:

Use SSE when you need instant notification of job completion
Use polling (GET /v2/send-command/<id>) for simple scripts or when SSE isn't supported
The Python client library uses polling by default; SSE is for direct API consumers

List Jobs

List all jobs with optional pagination and status filtering.

# All jobs (default: page 1, 20 per page)
curl -k -u "admin:password" https://localhost:8443/v2/jobs

# Filter by status
curl -k -u "admin:password" "https://localhost:8443/v2/jobs?status=failed"

# Paginate
curl -k -u "admin:password" "https://localhost:8443/v2/jobs?page=2&per_page=50"

Valid status values: queued, started, finished, failed.

Response:

{
  "jobs": [
    {
      "job_id": "550e8400-e29b-41d4-a716-446655440000",
      "status": "finished",
      "created_at": "2026-02-22T19:00:00+00:00",
      "ended_at": "2026-02-22T19:00:05+00:00"
    }
  ],
  "pagination": {
    "page": 1,
    "per_page": 20,
    "total": 1,
    "pages": 1
  }
}

Job Tags

Attach key-value metadata to any job for filtering and auditing. Tags are stored in job metadata and returned in job results and list responses.

Submitting Tags

curl -k -X POST https://localhost:8443/v2/send-command \
  -u "admin:password" \
  -d '{
    "host": "192.168.1.1",
    "platform": "cisco_ios",
    "commands": ["show version"],
    "tags": {"team": "network-ops", "env": "prod", "ticket": "CHG-12345"}
  }'

Constraints

Maximum 10 tags per job
Keys and values: alphanumeric, hyphens, underscores, colons (max 64 characters each)
Tags are optional — omit the field entirely if not needed

Filtering by Tag

Use ?tag=key:value on the list jobs endpoint:

# All jobs tagged team:network-ops
curl -k -u "admin:password" "https://localhost:8443/v2/jobs?tag=team:network-ops"

# Combine with status and pagination
curl -k -u "admin:password" "https://localhost:8443/v2/jobs?tag=env:prod&status=failed&per_page=50"

Tags are included in each job object in the response:

{
  "jobs": [
    {
      "job_id": "550e8400-e29b-41d4-a716-446655440000",
      "status": "finished",
      "created_at": "2026-02-22T19:00:00+00:00",
      "ended_at": "2026-02-22T19:00:05+00:00",
      "tags": {"team": "network-ops", "env": "prod", "ticket": "CHG-12345"}
    }
  ]
}

Connection Pooling

NAAS automatically reuses SSH connections to improve performance and reduce load on network devices.

How It Works

Persistent connections: SSH sessions are kept alive between requests
Per-worker pools: Each worker maintains its own connection pool
Automatic cleanup: Idle connections are closed after timeout
Credential isolation: Connections are keyed by (IP, port, username, password hash)

Performance Benefits

Faster response times: Eliminates SSH handshake overhead (~1-2 seconds per request)
Reduced device load: Fewer VTY sessions consumed on network devices
Better throughput: Handle more requests per second

When Pooling is Disabled

Connection pooling is automatically disabled for:

Platform autodetect (platform: "autodetect") - requires clean connection state
Structured commands (/v2/send-command_structured) - TextFSM state makes pooling unreliable

Configuration

Set via environment variables (see Kubernetes deployment):

CONNECTION_POOL_ENABLED: "true"  # Enable pooling (default)
CONNECTION_POOL_MAX_SIZE: "10"   # Max connections per worker (default)
CONNECTION_POOL_TTL: "300"       # Idle timeout in seconds (default)

Troubleshooting

Stale connections: If devices are rebooted, pooled connections may become stale. NAAS detects this and automatically reconnects.

Credential changes: Changing a user's password invalidates pooled connections. New requests will establish fresh connections.

High memory usage: Reduce CONNECTION_POOL_MAX_SIZE if workers consume too much memory.

Python Examples

Using Requests Library

import requests
import time
from requests.auth import HTTPBasicAuth

# Disable SSL warnings for self-signed certs
requests.packages.urllib3.disable_warnings()

# NAAS configuration
NAAS_URL = "https://localhost:8443"
AUTH = HTTPBasicAuth("admin", "password")

# Send command
response = requests.post(
    f"{NAAS_URL}/v2/send-command",
    auth=AUTH,
    verify=False,
    json={
        "host": "192.168.1.1",
        "platform": "cisco_ios",
        "commands": ["show version"]
    }
)

job_id = response.json()["job_id"]
print(f"Job ID: {job_id}")

# Poll for results
while True:
    result = requests.get(
        f"{NAAS_URL}/v2/send-command/{job_id}",
        auth=AUTH,
        verify=False
    ).json()

    status = result["status"]
    print(f"Status: {status}")

    if status == "finished":
        print(f"Result:\n{result['result']}")
        break
    elif status == "failed":
        print(f"Error: {result['error']}")
        break

    time.sleep(1)

Async Python Example

import asyncio
import aiohttp
from aiohttp import BasicAuth

async def send_command(session, device_ip, commands):
    """Send command to device via NAAS."""
    async with session.post(
        "https://localhost:8443/v2/send-command",
        json={
            "host": device_ip,
            "platform": "cisco_ios",
            "commands": commands
        },
        ssl=False
    ) as response:
        data = await response.json()
        return data["job_id"]

async def get_results(session, job_id):
    """Poll for job results."""
    while True:
        async with session.get(
            f"https://localhost:8443/v2/send-command/{job_id}",
            ssl=False
        ) as response:
            data = await response.json()

            if data["status"] in ["finished", "failed"]:
                return data

            await asyncio.sleep(1)

async def main():
    auth = BasicAuth("admin", "password")

    async with aiohttp.ClientSession(auth=auth) as session:
        # Send commands to multiple devices
        devices = ["192.168.1.1", "192.168.1.2", "192.168.1.3"]

        # Submit all jobs
        job_ids = await asyncio.gather(*[
            send_command(session, ip, ["show version"])
            for ip in devices
        ])

        # Get all results
        results = await asyncio.gather(*[
            get_results(session, job_id)
            for job_id in job_ids
        ])

        for device, result in zip(devices, results):
            print(f"\n{device}: {result['status']}")
            if result["status"] == "finished":
                print(result["result"][:100])

if __name__ == "__main__":
    asyncio.run(main())

Error Handling

Common HTTP Status Codes

200 OK - Job status retrieved successfully
202 Accepted - Job queued successfully
401 Unauthorized - Missing or invalid credentials
403 Forbidden - Job belongs to another user, or device is locked out
404 Not Found - Job ID not found
422 Unprocessable Entity - Validation failed (invalid IP, unknown platform, etc.)

Example Error Responses

Validation error (422):

{
  "validation_error": {
    "json": [
      {
        "loc": ["host"],
        "msg": "value is not a valid IP address or hostname",
        "type": "value_error"
      }
    ]
  }
}

Authentication failed (401):

{ "message": "Unauthorized" }

Device locked out (403):

{ "message": "Forbidden" }

Handling Errors in Python

import requests
from requests.auth import HTTPBasicAuth

def send_command_safe(ip, commands):
    """Send command with error handling."""
    try:
        response = requests.post(
            "https://localhost:8443/v2/send-command",
            auth=HTTPBasicAuth("admin", "password"),
            verify=False,
            json={
                "host": ip,
                "platform": "cisco_ios",
                "commands": commands
            },
            timeout=10
        )
        response.raise_for_status()
        return response.json()["job_id"]

    except requests.exceptions.HTTPError as e:
        print(f"HTTP error: {e.response.status_code}")
        print(f"Message: {e.response.json().get('message')}")
    except requests.exceptions.ConnectionError:
        print("Failed to connect to NAAS")
    except requests.exceptions.Timeout:
        print("Request timed out")
    except Exception as e:
        print(f"Unexpected error: {e}")

    return None

API Client Collections

Each NAAS release includes downloadable API client collections on the GitHub Releases page:

naas-vX.Y.Z.postman_collection.json — Postman collection with all endpoints
naas-vX.Y.Z.openapi.json — OpenAPI spec for import into any compatible tool

Postman

Download naas-vX.Y.Z.postman_collection.json from the release
In Postman: Import → select the file
Set environment variables: base_url, username, password

Insomnia

Download naas-vX.Y.Z.openapi.json from the release
In Insomnia: Create → Import → select the file

Bruno / Other Tools

Import naas-vX.Y.Z.openapi.json — any OpenAPI 3.x compatible client works.

Next steps

Troubleshooting Guide - Common issues
Security Best Practices - Secure your deployment
Full API Reference - Complete API documentation

API usage examples

Contents

Authentication

Basic Authentication

API Key Authentication

API Key Management

Create a Key

List Keys

Revoke a Key

Send Command

Enqueue Response (202 Accepted)

Basic Example

With Custom Port

With Enable Password

With Custom Request ID

With Custom Prompt Matching

Connection Timeout

Supported Platforms

Send Configuration

Basic Configuration

With Save Config

With Commit (Juniper)

Job Deduplication

Detecting a Deduplicated Response

Disabling Deduplication

Idempotency Keys

Webhooks

Security

Dead Letter Queue

List Failed Jobs

Replay a Failed Job

Configuration

Job Cancellation

Cancel a Job

Example

Job Status and Results

Check Job Status

Job States

SSE Streaming

List Jobs

Job Tags

Submitting Tags

Constraints

Filtering by Tag

Connection Pooling

How It Works

Performance Benefits

When Pooling is Disabled

Configuration

Troubleshooting

Python Examples

Using Requests Library

Async Python Example

Error Handling

Common HTTP Status Codes

Example Error Responses

Handling Errors in Python

API Client Collections

Postman

Insomnia

Bruno / Other Tools

Next steps