Give Your AI Agent Grounded Context: Daigest Skill Integration Guide

When AI agents search the web, they risk hallucination. Daigest takes a different approach: agents subscribe to trusted sources and receive pre-summarized, factual context on demand. Install the Daigest skill in Claude Code and your agent can query, create, and sync feed data directly via API.

Getting Started

You can complete the setup in 2 steps from Account Settings → API Keys.

Step 1: Add the skill to your AI agent

Install the Daigest skill from the Claude Code terminal:

npx skills add daige-st/daigest-skill

Or fetch the skill documentation directly to pass to your agent:

curl -s https://daige.st/skill.md

Once installed, Claude Code understands the Daigest API and can use feed data as context.

Step 2: Create a key and provide it to your agent

Click Create Key to open the key creation modal.

Enter a name for the key and click Create. Your key starting with dk_live_ will be generated.

The API key is only shown once. Copy it somewhere safe immediately.

Provide the key to your agent and you're ready to go.

Agent Usage Patterns

A. Read existing feeds (most common)

If feeds are already set up in the web app, agents can query them immediately:

Get my feed list and show me the latest content from the "Competitor Monitoring" feed

What the agent runs internally:

GET /feeds                              → find the relevant feed
GET /feeds/{id}?since=2026-02-20T00:00Z → check for updates

If has_changes: false, skip. If true, use the content as context.

B. Create a new feed and sync

Agents can create feeds on the fly when they need context on a new topic:

Create a feed for the latest AI news, sync it now, and show me what's in it

What the agent runs:

POST /feeds
{
  "name": "AI News Digest",
  "memory": "Summarize the latest AI news. Highlight new model releases and benchmark results.",
  "connections": [
    { "type": "rss", "url": "https://example.com/ai-feed.xml" }
  ]
}

POST /feeds/{id}/sync   → takes 30s to 5min, consumes AI quota

C. Agent writes content directly

When the AI curates information itself and wants to store it as a feed:

Save my research into a "Research Notes" feed and publish it as a post

POST /feeds { "name": "Research Notes" }

PATCH /feeds/{id}
{ "content": "# Agent-curated content\n\n..." }

POST /feeds/{id}/posts   → publish as a snapshot (once per 15 min)

Practical Examples

Track competitor activity

Sync my competitor monitoring feed now and pick out any product launches
or pricing changes from this week's updates

Daily morning briefing

Get all feeds updated today and give me the 5 most important updates,
one sentence each

Prepare research context

Create a new feed with 2 LLM benchmark RSS sources and set it to
auto-update every day at 9am

Memory Tips

Add AI summarization instructions in memory when creating a feed to get more useful context:

FAQ

Q: Can I use the API without installing the skill?

Yes. The skill is just documentation that teaches the agent how to use the API. With an API key, you can call any endpoint directly from any HTTP client. You can also auto-generate tools from the OpenAPI spec at GET https://daige.st/api/v1/openapi.json.

Q: Does calling sync frequently get expensive?

Each sync consumes an AI quota request. Use GET /feeds/{id}?since=... to check for changes first, and only process when has_changes: true. For regular updates, set a schedule instead.

Q: Can I add OAuth-based sources like Slack or Notion via API?

RSS and URL sources can be added directly via API. OAuth-based sources (Slack, Notion, GitHub, Discord, etc.) need to be connected first in the web app, after which you can read and sync feeds containing them via the API.

Set up your feeds on Daigest and start using them with your AI agent today.