← Home

Feed2Context

Completed

One-click social post → research report for LinkedIn and X. Capture a post, auto-shape a research query, search with reasoning, get a detailed report.

Website ↗ GitHub ↗

Overview

Feed2Context addresses a personal pain point: I bookmark a lot of LinkedIn and X posts but rarely revisit them. The tool turns any social post into a structured research report.

Click the “Save + Analyze” button on a post, add a short note on why it matters, and the system extracts the post text, builds an optimized search query, runs streamed research with reasoning, and delivers a structured report in 30-90 seconds.

Built around platform anti-scraping measures while staying fast and reliable. Uses Groq Compound for extraction and research, Kimi-K2 for query building, and Browser Use for X where static scraping fails.

Feed2Context

Architecture

The system runs two parallel pipelines serving both at different ports for clean separation:

  • Chrome Extension (Manifest V3): Injects a “Save + Analyze” button on LinkedIn and X posts. Captures post URLs, extracts visible text where possible, and prompts for user notes. Sends data to the appropriate local API endpoint.
  • LinkedIn Pipeline: Uses groq/compound-mini to visit the post URL and extract core text. Compound Mini handles JavaScript rendering and “see more” expansion reliably without browser automation.
  • X Pipeline: X employs aggressive anti-scraping that breaks static HTTP fetches. This pipeline uses browser_use.Agent with ChatGroq to control a real browser, load the tweet, and extract author + text reliably.
  • Query Building: Both pipelines feed extracted text into moonshotai/kimi-k2-instruct, which compresses the user’s intent into a single, effective search query (≤50 words). Kimi-K2 excels at understanding nuanced goals from brief context.
  • Research: groq/compound performs an agentic research against the query. Compound’s speed and reasoning capabilities produce detailed Markdown reports in 15-30 seconds.
  • Storage: All results append to a local JSONL, creating a personal research database. Composio MCP integration optionally emails formatted reports.

Status

Launched September 2025. Processing live captures from LinkedIn and X via the Chrome extension.