ahrefs-python（ahrefs Python 搜索）

Ahrefs Python SDK Skill

Overview

The Ahrefs API provides programmatic access to Ahrefs SEO data. The official Python SDK (ahrefs-python) provides typed request and response models for all endpoints, auto-generated from the OpenAPI spec.

Key capabilities:

• Site Explorer - Backlinks, organic keywords, domain rating, traffic, referring domains
• Keywords Explorer - Keyword research, volumes, difficulty, related terms
• Management — Rank Tracker project, keyword list, competitor, and Brand Radar prompt management (create, update, delete)
• Rank Tracker - SERP monitoring, competitor tracking
• Site Audit - Technical SEO issues, page content, page explorer
• Brand Radar - AI brand mentions, share of voice, impressions
• SERP Overview - Search result analysis
• Batch Analysis - Bulk domain/URL metrics via POST
• Web Analytics - Website visitor analytics (traffic, browsers, devices, sources, pages)
• Public — Ahrefs crawler IP addresses and CIDR ranges (no API key required)
• Subscription Info — API usage limits, billing period, key expiration

Installation

pip3 install git+https://github.com/ahrefs/ahrefs-python.git

Requires Python 3.11+. Dependencies: httpx, pydantic.

API Method Discovery

The SDK has 105 methods across 11 API sections. The built-in search tool is the fastest way to find the right method — it returns matching method signatures, parameters, and return types directly, so there's no need to scan through a large reference.

Python (preferred when already in a Python context):

from ahrefs.search import search_api_methods

# Returns formatted text with method signatures, parameters, and return types
print(search_api_methods("domain rating"))

# Filter by API section and limit results
print(search_api_methods("backlinks", section="site-explorer", limit=3))

CLI (preferred when exploring from the terminal):

# Ensure python3 points to the interpreter where ahrefs-python is installed:
#   which python3
#   python3 -c "import ahrefs"
python3 -m ahrefs.api_search "domain rating"
python3 -m ahrefs.api_search "backlinks" --section site-explorer --limit 3
python3 -m ahrefs.api_search "batch" --json
python3 -m ahrefs.api_search --sections  # list all API sections

IMPORTANT RULES

• ALWAYS use the ahrefs-python SDK. DO NOT make raw httpx/requests calls to the Ahrefs API.
• ALWAYS pass dates as strings in YYYY-MM-DD format (e.g. "2025-01-15").
• ALWAYS use select on list endpoints to request only the columns you need. List endpoints return all columns by default, which wastes API units and increases response size.
• USE context managers (with / async with) for client lifecycle management.
• PREFER AsyncAhrefsClient with asyncio.gather() when making multiple independent API calls — parallelism saves wall-clock time and API round-trips.
• NEVER hardcode API keys in source code. Use the AHREFS_API_KEY environment variable or your preferred secrets mechanism.
• The client handles retries (429, 5xx, connection errors) automatically. DO NOT implement your own retry logic on top of the SDK.

Quick Start

import os
from ahrefs import AhrefsClient

with AhrefsClient(api_key=os.environ["AHREFS_API_KEY"]) as client:
    data = client.site_explorer_domain_rating(target="ahrefs.com", date="2025-01-15")
    print(data.domain_rating)  # 91.0
    print(data.ahrefs_rank)    # 3

Public endpoints work without an API key — just omit the api_key parameter:

from ahrefs import AhrefsClient

with AhrefsClient() as client:
    ips = client.public_crawler_ips()
    ranges = client.public_crawler_ip_ranges()

Only public_* methods work without a key. All other methods require authentication. An authenticated client can also call public methods — there is no need for a separate client or raw HTTP calls.

SDK Patterns

Client Setup

import os
import ahrefs

with ahrefs.AhrefsClient(
    api_key=os.environ["AHREFS_API_KEY"],  # omit for public-only access
    base_url="...",          # override API base URL (default: https://api.ahrefs.com/v3)
    timeout=30.0,            # request timeout in seconds (default: 60)
    max_retries=3,           # retries on transient errors (default: 2)
) as client:
    ...

Async client:

import os
from ahrefs import AsyncAhrefsClient

async with AsyncAhrefsClient(api_key=os.environ["AHREFS_API_KEY"]) as client:
    data = await client.site_explorer_domain_rating(target="ahrefs.com", date="2025-01-15")

For parallel calls, use asyncio.gather:

import asyncio

async with AsyncAhrefsClient(api_key=os.environ["AHREFS_API_KEY"]) as client:
    dr_ahrefs, dr_moz = await asyncio.gather(
        client.site_explorer_domain_rating(target="ahrefs.com", date="2025-01-15"),
        client.site_explorer_domain_rating(target="moz.com", date="2025-01-15"),
    )

Calling Methods

Two calling styles -- both are equivalent:

# Keyword arguments (recommended)
data = client.site_explorer_domain_rating(target="ahrefs.com", date="2025-01-15")

# Request objects (full type safety)
from ahrefs.types import SiteExplorerDomainRatingRequest
request = SiteExplorerDomainRatingRequest(target="ahrefs.com", date="2025-01-15")
data = client.site_explorer_domain_rating(request)

Method names follow {api_section}_{endpoint}, e.g. site_explorer_organic_keywords, keywords_explorer_overview.

Responses

Methods return typed Data objects directly.

Scalar endpoints return a single data object (or None):

data = client.site_explorer_domain_rating(target="ahrefs.com", date="2025-01-15")
print(data.domain_rating)

List endpoints return a list of data objects. There is no pagination — set limit to the number of results you need. Use select to request only the columns you need:

items = client.site_explorer_organic_keywords(
    target="ahrefs.com",
    date="2025-01-15",
    select="keyword,volume,best_position",
    order_by="volume:desc",
    limit=10,
)
for item in items:
    print(item.keyword, item.volume, item.best_position)

Error Handling

import ahrefs

try:
    data = client.site_explorer_domain_rating(target="example.com", date="2025-01-15")
except ahrefs.AuthenticationError:    # 401
    ...
except ahrefs.RateLimitError as e:    # 429 -- e.retry_after has the delay
    ...
except ahrefs.NotFoundError:          # 404
    ...
except ahrefs.APIError as e:          # other 4xx/5xx -- e.status_code, e.response_body
    ...
except ahrefs.APIConnectionError:     # network / timeout
    ...

All exceptions inherit from ahrefs.AhrefsError.

When errors occur and the API key is valid, call subscription_info_limits_and_usage() to diagnose — it returns the subscription plan, key expiration date, and usage vs limits. For 403 errors specifically, also read e.response_body for the exact reason (units limit, export rows limit, domains-per-week limit, etc.).

Common Parameters

Most list endpoints share these parameters:

Parameters typed as enums in the API reference (CountryEnum, VolumeModeEnum, etc.) accept plain strings — pass country="us" not CountryEnum("us").

The where parameter takes a JSON string. Use json.dumps() to build it:

import json
where = json.dumps({"field": "volume", "is": ["gte", 1000]})
items = client.site_explorer_organic_keywords(
    target="ahrefs.com", date="2025-01-15",
    select="keyword,volume", where=where,
)

For full filter syntax (boolean combinators, operators, nested fields), see references/filter-syntax.md.

Recommended Defaults

Unless the user requests otherwise:

• Backlink endpoints (all_backlinks, anchors, pages_by_backlinks, refdomains): Use history="live" for current live data. The API defaults to all_time which includes lost backlinks.
• Traffic/keyword endpoints (organic_keywords, top_pages, metrics, etc.): Use today's date for the date parameter.

API Methods

Use search_api_methods("query") or python3 -m ahrefs.api_search "query" to find methods by keyword. Search covers all 105 methods across 11 API sections and returns complete signatures, parameters, and response fields. Results with very large field lists (e.g. site_audit_page_explorer with 608 fields) are truncated at 9K chars — if you see ... [truncated], use select to request only the columns you need rather than relying on the full field list.

Web Analytics

Web Analytics has 34 endpoints across many dimensions — not all method names match the dimension name directly (e.g. pages data uses web_analytics_top_pages, not web_analytics_pages). Always use search_api_methods("web analytics <topic>") to find the correct method.

These endpoints require a project_id (not a target domain). Use from_ and to for datetime ranges — from is a Python reserved word, so the SDK uses from_ with automatic serialization to the correct API name.

# Overall traffic stats (scalar endpoint)
stats = client.web_analytics_stats(project_id=123)
print(stats.pageviews, stats.visitors, stats.visits)

# Dimension breakdown (list endpoint — use select)
browsers = client.web_analytics_browsers(
    project_id=123,
    from_="2025-01-01T00:00:00Z",
    to="2025-01-31T23:59:59Z",
    select="browser,visitors",
    limit=10,
)

# Time-series chart (list endpoint — requires granularity)
chart = client.web_analytics_chart(
    project_id=123,
    granularity="daily",
    from_="2025-01-01T00:00:00Z",
    to="2025-01-31T23:59:59Z",
)

Management

Management endpoints handle Rank Tracker project configuration and Brand Radar prompt management — creating projects, managing keyword lists, adding competitors, and managing custom Brand Radar prompts. Unlike other API sections, this section includes endpoints that modify state via PUT/POST/PATCH/DELETE. None of these endpoints use select, where, order_by, or limit.

• Use management_projects() to list all projects and find project_id values.
• Use management_locations() to get valid country codes, language codes, and location IDs before adding keywords.
• keyword_list_id is NOT discoverable via the API — the user must provide it from the Ahrefs web UI. Keyword list methods: management_keyword_list_keywords (list), management_set_keyword_list_keywords (add via PUT), management_keyword_list_keywords_delete (remove).
• Brand Radar prompt endpoints require a report_id (from the Brand Radar report URL). Methods: management_brand_radar_prompts (list), management_create_brand_radar_prompts (create), management_brand_radar_prompts_delete (remove).
• Delete endpoints are destructive and irreversible. ALWAYS confirm with the user before calling management_project_keywords_delete, management_project_competitors_delete, management_keyword_list_keywords_delete, or management_brand_radar_prompts_delete.

# List all Rank Tracker projects
projects = client.management_projects()
for p in projects:
    print(p.project_id, p.project_name, p.url)

# Add keywords to a project (requires nested request types)
from ahrefs.types import ManagementSetProjectKeywordsLocation, ManagementSetProjectKeywordsKeyword

client.management_set_project_keywords(
    project_id=12345,
    locations=[ManagementSetProjectKeywordsLocation(
        country="us",
        language="en",
    )],
    keywords=[ManagementSetProjectKeywordsKeyword(keyword="ahrefs seo")],
)

# Add competitors to a project
from ahrefs.types import ManagementCreateProjectCompetitorsCompetitor

client.management_create_project_competitors(
    project_id=12345,
    competitors=[ManagementCreateProjectCompetitorsCompetitor(
        url="moz.com",
        mode="domain",
    )],
)

# List Brand Radar custom prompts
prompts = client.management_brand_radar_prompts(report_id="abc123")
for p in prompts:
    print(p.prompt, p.country, p.created_at)

# Create custom prompts for specific countries
client.management_create_brand_radar_prompts(
    report_id="abc123",
    countries=["us", "gb"],
    prompts=["What is the best SEO tool?"],
)