Post-pipelinePost Pipeline: Substack Distribution

Post Pipeline: Substack Distribution

Post Pipeline: Substack Distribution

Research and options for automating the publishing of content to Substack.

Job Story

When I have a polished post ready to publish, I want to push it to my Substack newsletter programmatically — so I don't have to log in, copy-paste, and manually publish.

The Problem

Substack has no official public API for posting content.

There are unofficial approaches, but they all require:

  • Extracting session cookies from your browser
  • Accepting the risk that things may break when Substack changes their internal APIs
  • Potential ToS concerns (though widely used without issues)

Content Types on Substack

Important distinction:

TypeDescriptionAutomation Status
PostsFull newsletter articlesHard — no library support, browser automation only
NotesShort-form, Twitter-like updatesEasier — library and n8n support available

Options Analysis

Option 1: TypeScript Library (substack-api) + n8n

What it is: An unofficial TypeScript client that wraps Substack's internal API, integrated with n8n for no-code automation.

Capabilities:

  • Publish Notes (short-form)
  • Read posts, profiles, comments
  • Full post publishing: not yet supported (on roadmap)

Setup:

  1. Get connect.sid cookie from browser DevTools
  2. Install n8n and the n8n-nodes-substack community node
  3. Configure with your publication URL and cookie

Pros:

  • No-code automation via n8n
  • Notes work reliably
  • Active development

Cons:

  • Can't publish full posts yet
  • Cookie expires, needs periodic refresh
  • Unofficial, could break

Best for: Automating Substack Notes as a social/engagement channel

Sources:

Option 2: Browser Automation (BrowserMCP)

What it is: Use an MCP server + Chrome extension to have Claude (or another agent) control a browser and perform the posting actions like a human would.

Capabilities:

  • Full posts: Yes (browser can do anything you can)
  • Notes: Yes
  • Any Substack action: formatting, scheduling, images, etc.

Setup:

  1. Install Claude Desktop
  2. Install Node.js
  3. Install BrowserMCP extension on a dedicated Chromium browser
  4. Configure claude_desktop_config.json to connect the MCP server
  5. Keep browser logged into Substack

Pros:

  • Can publish full posts, not just Notes
  • Works with existing login session
  • Can handle any Substack feature

Cons:

  • Requires running a browser (heavier than API calls)
  • More fragile — UI changes can break automation
  • Slower than direct API calls
  • Requires Claude Desktop environment

Best for: Full newsletter posts when API options don't exist

Source: MCP to Schedule Substack Notes

Option 3: Python Library (nhagar/substack_api)

What it is: A Python wrapper for Substack's internal API.

Capabilities:

  • Read posts, profiles, podcasts
  • Search within newsletters
  • Access paywalled content you've paid for

Limitation: Read-only. Cannot publish posts or notes.

Best for: Analytics, content retrieval, backups — not publishing.

Option 4: Manual Reverse-Engineering

What it is: Use browser DevTools to capture the exact requests Substack makes when you publish, then replicate them with curl/fetch.

Approach:

  1. Open DevTools → Network tab
  2. Publish a post manually
  3. Capture the request: URL, headers, payload
  4. Recreate as a script

Pros:

  • Can publish full posts
  • No dependencies on third-party libraries
  • Full control

Cons:

  • Requires technical effort to set up
  • Brittle — any Substack change breaks it
  • Must maintain yourself

Best for: If you want full control and are comfortable maintaining custom scripts

Recommendation

Given your goal (publish full newsletter posts, not just Notes):

┌─────────────────────────────────────────────────────────┐
│  SHORT-TERM: Browser Automation (Option 2)              │
│  - Works for full posts                                 │
│  - BrowserMCP + Claude Desktop                          │
│  - Heavier but reliable                                 │
└─────────────────────────────────────────────────────────┘
                          OR
┌─────────────────────────────────────────────────────────┐
│  SHORT-TERM: Manual Reverse-Engineering (Option 4)      │
│  - Lighter weight if you can figure out the endpoints   │
│  - More effort upfront, less dependencies               │
└─────────────────────────────────────────────────────────┘

┌─────────────────────────────────────────────────────────┐
│  WATCH: TypeScript library (Option 1)                   │
│  - Full post support is on their roadmap                │
│  - Would be cleanest solution when available            │
└─────────────────────────────────────────────────────────┘

Key Decision Point

For "AI Tried Today" specifically:

ApproachProsCons
Notes-onlyEasy to automate (library support), fits "field notes" concept, quick frequent updatesLess depth, no email delivery to subscribers
Full PostsEmail delivery, long-form content, what Substack is designed forHarder to automate (browser automation or reverse-engineering required)

Consideration: Notes could work well for the "field notes" framing — short, frequent observations. Full posts could be reserved for weekly roundups or deeper pieces, posted manually or via browser automation.

Open Questions

  • Is Notes-only sufficient for "AI Tried Today"? (shorter, more frequent updates)
  • If full posts needed, tolerance for browser automation setup?
  • Worth exploring the reverse-engineering approach for full posts?
  • Hybrid approach: automate Notes, manual/semi-manual for full posts?

Next Steps

  • Decide: Notes-only workflow vs. full posts required
  • If Notes: Set up n8n + TypeScript library
  • If full posts: Either BrowserMCP setup or reverse-engineer the publish endpoint
Edit this page