# Get a daily LinkedIn feed digest with Kimi Code

Source: https://finchreach.com/blog/kimi-linkedin-feed-digest
Published 2026-10-28 by Finchreach team. Tags: kimi, linkedin, mcp, productivity, use-case.

Have Kimi read your LinkedIn feed and notifications, summarize what matters in five minutes, and queue the reactions you choose for approval.

## The short answer

Scrolling LinkedIn to keep up with your network eats time and attention. With Finchreach connected, Kimi can read your home feed and notifications for you, sort out the posts that actually matter to your work, and give you a short digest. If you want to like or comment on something it found, it queues that action for your approval.

Today the route is Kimi Code, Moonshot AI’s command-line agent, which supports remote MCP servers with an API key header. MCP is a standard plug that lets AI apps use outside tools; Finchreach is a remote MCP server at `https://mcp.finchreach.com/mcp`. As of October 2026 we couldn’t confirm custom MCP support in the kimi.com chat app, so use Kimi Code (or another MCP client running a Kimi model).

## Why read LinkedIn through a digest instead of the app?

The LinkedIn feed mixes the posts you care about with promoted content, viral posts and suggestions. Reading it in the app means a lot of skipping. A digest flips that: you say what you care about once, and the assistant filters for it.

Kimi models offer long context windows, which suits a task where the assistant reads many posts and returns a few lines. Finchreach gives it `linkedin_get_feed` (with a cursor for more pages) and `linkedin_get_notifications` (latest notifications and unseen counts). Both are reads, so they run without approval.

## What does a good feed digest prompt look like?

Tell Kimi who matters and what to ignore:

> Ask your assistant: “Read the first three pages of my LinkedIn feed and my notifications. Give me a digest: (1) posts from people I work with or customers, (2) posts about climate tech funding or carbon accounting, (3) anything where I was mentioned or someone commented on my posts. Skip promoted posts, job ads and generic motivation. One line per item with the author and why it matters.”

Then act on a few items:

> Ask your assistant: “Like items 2 and 5. For item 4, draft a two-sentence comment congratulating her on the Series A and asking what they will hire for first. Show me the comment before you queue it.”

The reactions and the comment go to your approval queue at [/approvals](https://finchreach.com/approvals). You can save your digest instructions in a file in your project and tell Kimi to follow it each morning, so the format stays the same.

> Ask your assistant: “Compared with yesterday’s digest in digest-notes.md, what is new? Append today’s digest to the file.”

## How do you connect Kimi to LinkedIn?

1. Create an account at [finchreach.com/sign-up](https://finchreach.com/sign-up), connect LinkedIn under Accounts, and create an API key under API keys.
2. Add Finchreach to `~/.kimi-code/mcp.json` (or `.kimi-code/mcp.json` in a project), or use `/mcp-config` in the Kimi Code interface.
3. Use the snippet below. Instead of the `headers` block you can set `"bearerTokenEnvVar": "FINCH_KEY"` and export the key as an environment variable.
4. Start Kimi Code and ask it to list your LinkedIn accounts to check the connection.
5. The [Kimi guide](https://finchreach.com/guides/kimi) covers the legacy kimi-cli command too; the [docs](https://finchreach.com/docs) explain keys and approvals.

~/.kimi-code/mcp.json:
```json
{
  "mcpServers": {
    "finchreach": {
      "url": "https://mcp.finchreach.com/mcp",
      "headers": { "Authorization": "Bearer lk_your_key" }
    }
  }
}
```

> Config paths have moved between Kimi Code documentation versions, as of October 2026. If the file isn’t picked up, check the current [Kimi Code MCP docs](https://moonshotai.github.io/kimi-code/en/customization/mcp). With the legacy kimi-cli, the equivalent is `kimi mcp add --transport http finchreach https://mcp.finchreach.com/mcp --header "Authorization: Bearer $KEY"`.

## How do you make a feed digest actually useful?

- Name people and companies, not just topics. “Posts from Acme or anyone at Acme” filters better than “sales posts”.
- Ask for the reason each item made the cut. It shows you quickly when the filter is off.
- Limit the length. Ten lines is a digest; forty is just a shorter feed.
- Keep engagement selective. Reacting to everything the digest finds defeats the point and looks automated.
- Check your notifications section first. Replies to your own posts are usually the most time-sensitive.

## What does a digest look like in practice?

A good digest is short enough to read with your first coffee. Ask Kimi for a fixed shape, for example three headings (people, topics, mentions) with one line per item, and a final line with anything that needs a reply today. Once the shape is stable, you can compare days at a glance and notice when someone important has been quiet or busy.

> Ask your assistant: “Use this format from now on: People, Topics, Mentions, then “Needs a reply today”. Maximum ten lines in total.”

## What can’t the digest do?

Finchreach reads what LinkedIn shows your account: your feed, your notifications, your messages and search results. It doesn’t provide analytics on your own posts, and it doesn’t run on a timer by itself. A daily digest means you (or a script you write) start Kimi Code each day. Finchreach has no scheduling of its own.

## Does reading the feed with an AI put my account at risk?

> Automating a personal LinkedIn account carries risk under LinkedIn’s User Agreement. Finchreach puts every write behind your approval, so actions happen at your pace; that lowers the risk but doesn’t remove it.

Reads are lighter than writes, but they are still activity on your account. Reading a few pages once a day is closer to normal use than paging through hundreds of posts every hour.

## Frequently asked questions

Can I use the kimi.com chat app? We couldn’t verify custom MCP support there as of October 2026. Use Kimi Code, or a Kimi model inside another MCP client.

Can the digest include my messages? Yes. `linkedin_get_conversations` and `linkedin_get_messages` let Kimi summarize your inbox too; replies it drafts wait for approval.

Can I use a different model for the same workflow? Yes; see [Claude Code](https://finchreach.com/integrations/claude-code) or the [Kimi integration](https://finchreach.com/integrations/kimi) page.

Where do I start? [Sign up](https://finchreach.com/sign-up) and follow the [quickstart](https://finchreach.com/docs).