linkitin 0.1.3

LinkedIn automation library - post, fetch, and search via Voyager API

linkitin

Python library for LinkedIn automation via the Voyager API. Fetch posts, search content, and publish — all through an async client.

macOS only — login_from_browser() requires macOS and Chrome. The login_with_cookies() path works cross-platform.

Install

pip install linkitin

Requires Python 3.10+.

Development

python -m venv .venv
source .venv/bin/activate
pip install -e ".[dev]"

Authentication

Chrome proxy mode (recommended — requires macOS and Chrome logged into LinkedIn):

async with LinkitinClient() as client:
    await client.login_from_browser()  # routes API requests through Chrome via AppleScript

Chrome 145+ uses App-Bound Encryption for cookies, making them inaccessible to external tools. login_from_browser() works around this by executing API requests inside Chrome's LinkedIn tab, which automatically includes all session cookies. Requires "Allow JavaScript from Apple Events" enabled in Chrome (View > Developer).

Manual cookie input (if you have raw li_at and JSESSIONID values):

async with LinkitinClient() as client:
    await client.login_with_cookies(li_at="AQED...", jsessionid="ajax:123...")

Cookies from manual input are saved to disk and reused on subsequent runs via login_from_saved().

Usage

import asyncio
from linkitin import LinkitinClient

async def main():
    async with LinkitinClient() as client:
        # Authenticate
        if not await client.login_from_saved():
            await client.login_from_browser()

        # Fetch your posts
        posts = await client.get_my_posts(limit=10)
        for p in posts:
            print(p.text[:80], f"({p.likes} likes)")

        # Search posts
        results = await client.search_posts("AI startups", limit=5)

        # Get home feed
        feed = await client.get_feed(limit=20)

        # Get trending posts
        trending = await client.get_trending_posts(topic="AI", period="past-week", limit=5)

        # Create a text post
        urn = await client.create_post("Hello LinkedIn!", visibility="PUBLIC")

        # Create a post with image
        with open("chart.png", "rb") as f:
            urn = await client.create_post_with_image(
                text="Check out these metrics!",
                image_data=f.read(),
                filename="chart.png",
            )

        # Schedule a post (automatically rounded to next 15-min slot)
        from datetime import datetime, timezone, timedelta
        urn = await client.create_scheduled_post(
            text="Scheduled post — hello future!",
            scheduled_at=datetime.now(timezone.utc) + timedelta(hours=2),
        )

        # Comment on a post (feed posts always have thread_urn)
        feed = await client.get_feed(limit=5)
        comment_urn = await client.comment_post(
            feed[0].urn, "Great post!",
            thread_urn=feed[0].thread_urn or "",
        )

        # Reply to a comment (threaded)
        reply_urn = await client.comment_post(
            feed[0].urn, "Thanks!",
            parent_comment_urn=comment_urn,
            thread_urn=feed[0].thread_urn or "",
        )

        # Repost a post from the feed (requires share_urn)
        repost_urn = await client.repost(feed[0].share_urn)

        # Repost with your thoughts
        repost_urn = await client.repost(feed[0].share_urn, text="Great insights!")

        # Comment on a trending post (only works for posts with real URNs)
        trending = await client.get_trending_posts(topic="AI", period="past-week")
        for post in trending:
            if "dom:post" in post.urn:
                continue  # synthetic URN — read-only
            await client.comment_post(
                post.urn, "Interesting!",
                thread_urn=post.thread_urn or "",
            )

asyncio.run(main())

Configuration

LinkitinClient accepts optional parameters to override defaults:

client = LinkitinClient(
    cookies_path="~/.myapp/linkedin_cookies.json",  # default: ~/.linkitin/cookies.json
    timezone="America/New_York",                     # default: detected from system
    timezone_offset=-5.0,                            # default: computed from system
    display_width=1440,                              # default: 1920
    display_height=900,                              # default: 1080
    user_agent="Mozilla/5.0 ...",                    # default: modern Chrome/macOS UA
)

Rate Limiting

All requests go through a token-bucket rate limiter: 10 requests per minute with 1-5 second random jitter between requests. On 429/403 responses, exponential backoff kicks in starting at 30 seconds.

API Reference

Method	Description
login_from_browser()	Authenticate via Chrome proxy (AppleScript, macOS only)
login_with_cookies(li_at, jsessionid)	Authenticate with manual cookies
login_from_saved()	Load previously saved cookies
get_my_posts(limit=20)	Fetch your posts
search_posts(keywords, limit=20)	Search posts by keywords
get_feed(limit=20)	Fetch home feed
get_trending_posts(topic, period, limit, from_followed, scrolls)	Fetch trending posts sorted by engagement. Some posts may have synthetic URNs (read-only).
create_post(text, visibility="PUBLIC")	Create a text post
upload_image(image_data, filename)	Upload an image, returns media URN
create_post_with_image(text, image_data, filename, visibility="PUBLIC")	Create post with image
create_scheduled_post(text, scheduled_at, visibility="PUBLIC")	Schedule a text post; scheduled_at rounded to next 15-min slot
create_scheduled_post_with_image(text, image_data, filename, scheduled_at, visibility="PUBLIC")	Schedule a post with image; scheduled_at rounded to next 15-min slot
comment_post(post_urn, text, parent_comment_urn="", thread_urn="")	Comment on a post (or reply to a comment). Requires a real URN.
repost(share_urn, text="")	Repost (reshare) an existing post. Requires share_urn (feed/my_posts only).
delete_post(post_urn)	Delete a post by URN
close()	Close the HTTP session

Contributing

See CONTRIBUTING.md for guidelines. PRs welcome!

• Open issues for bugs or feature requests
• Keep PRs focused — one change per PR
• Run python -m pytest tests/ before submitting