bookmark-research-engine

bookmark-research-engine

Turn X bookmarks into ranked, analyzed research briefs. Self-contained Claude Code plugin — no external CLI dependencies.

Install

As a Claude Code plugin (recommended)

Add the marketplace, then install:

claude plugin install k3nnethfrancis/bookmark-research-engine

The plugin auto-installs the bre CLI on first session. Then run setup:

bre setup

Standalone

pip install git+https://github.com/k3nnethfrancis/bookmark-research-engine.git
bre setup

Usage

CLI

bre fetch              # pull bookmarks from X into pending queue
bre fetch -n 50        # fetch 50 bookmarks
bre status             # pipeline health and pending count
bre setup              # interactive setup (cookies, vault, config)
bre refresh-ids        # refresh X GraphQL query IDs
bre enrich input.json  # re-run enrichment on raw tweet JSON

In Claude Code

> check bookmarks     # status
> triage bookmarks    # rank pending items against your research interests
> deep dive           # spawn parallel research agents for tier-1 items
> review bookmarks    # summarize findings, connect to projects

What It Does

1. Fetches bookmarks from X using session cookies (no bird/smaug dependency)
2. Enriches them — expands t.co links, fetches articles, GitHub READMEs, resolves quote tweets
3. Triages against your research interests (interactive or automated)
4. Deep dives — spawns parallel agents that investigate tier-1 items using WebSearch, WebFetch, and arxiv MCP
5. Produces structured research briefs with frontmatter, findings, methodology, implications, and actionable takeaways

How It Works

The bre CLI handles the mechanical fetch and enrichment. The Claude Code skill handles the intelligence — triage ranking, deep dive research, report writing.

bre fetch → pending.json → skill triage → approval file → skill deep dive → reports/

The vault structure (where reports, triage, and inbox files live) is fully configurable. Default: ~/.config/bre/vault/. Point vault_root at an existing Obsidian vault to integrate.

Requirements

• Python 3.11+
• Claude Code (for the skill — triage, deep dive, review)
• X account with bookmarks (cookies for auth)