x-research-but-cheaper

X/Twitter research agent for OpenClaw and Claude Code. Search, filter, monitor — all from the terminal.

Powered by TwitterAPI.io. No X Developer Portal account needed. Pay-per-use pricing (~$0.15/1k tweets) instead of $100+/month.

What it does

Wraps the TwitterAPI.io API into a fast CLI so your AI agent (or you) can search tweets, pull threads, monitor accounts, and get sourced research without writing curl commands.

• Search with engagement sorting, time filtering, noise removal
• Quick mode for cheap, targeted lookups
• Profile analysis — user info + recent tweets
• Thread reading — full conversation threads
• Watchlists for monitoring accounts
• Cache to avoid repeat API charges
• Cost transparency — every command shows what it cost

Install

ClawHub (recommended)

clawhub install x-research-but-cheaper

OpenClaw (manual)

cd ~/.openclaw/workspace/skills
git clone https://github.com/minilozio/x-research-but-cheaper.git

Claude Code

mkdir -p .claude/skills
cd .claude/skills
git clone https://github.com/minilozio/x-research-but-cheaper.git

Setup

1. API key — Get one from twitterapi.io
2. Set the env var:

   export TWITTERAPI_KEY="your-key-here"
   

   Or save it to your .env file.
3. Node.js 18+ required (for native fetch). Zero npm dependencies.

Usage

Natural language (just talk to Claude)

• "What are people saying about Opus 4.6?"
• "Search X for the latest on Seedance 2.0"
• "What's the sentiment around remote work this week?"
• "Check what @minilozio posted recently"
• "Research the best takes on AI coding assistants"
• "Find viral tweets about startups this week"

CLI commands

cd skills/x-research-but-cheaper/scripts

# Search (sorted by likes, auto-filters retweets)
npx tsx x-search.ts search "Seedance 2.0" --sort likes --limit 10

# Quick pulse check
npx tsx x-search.ts search "remote work" --quick

# Profile — user info + recent tweets
npx tsx x-search.ts profile minilozio --count 5

# Read a thread
npx tsx x-search.ts thread 1234567890123456789

# Single tweet
npx tsx x-search.ts tweet 1234567890123456789

# Replies and quotes
npx tsx x-search.ts replies 1234567890123456789 --sort likes
npx tsx x-search.ts quotes 1234567890123456789

# Mentions — who's talking about someone
npx tsx x-search.ts mentions ycombinator --since 1d

# Followers / following
npx tsx x-search.ts followers minilozio --limit 20
npx tsx x-search.ts following minilozio --limit 20

# Search for users
npx tsx x-search.ts users "machine learning engineer"

# Trending topics
npx tsx x-search.ts trending
npx tsx x-search.ts trending --woeid 23424977  # US trends

# Community research
npx tsx x-search.ts community 1234567890
npx tsx x-search.ts community tweets 1234567890

# Watchlist
npx tsx x-search.ts watchlist add minilozio "AI research"
npx tsx x-search.ts watchlist check

# Save research to file
npx tsx x-search.ts search "query" --save --markdown

Search options

--sort likes|retweets|impressions|recent    Sort order (default: likes)
--since 1h|3h|12h|1d|7d                    Time filter
--min-likes N                               Minimum likes
--min-impressions N                         Minimum views
--pages N                                   Pages to fetch, 1-5 (default: 3)
--limit N                                   Max results (default: 15)
--quick                                     Quick mode (see below)
--from <username>                           Shorthand for from:username
--quality                                   Min 10 likes filter
--no-replies                                Exclude replies
--type Top|Latest                           Search mode (default: Latest)
--save                                      Save results to file
--json                                      Raw JSON output
--markdown                                  Markdown research doc

X search operators work too: from:user, OR, -is:reply, has:links, min_faves:N, lang:en, $TICKER, and more.

Quick Mode

--quick is designed for fast, cheap lookups when you just need a pulse check on a topic.

What it does:

• Forces single page (max 10 results) — reduces API cost
• Auto-appends -is:retweet -is:reply noise filters
• Minimum 5 likes filter (removes spam/zero-engagement)
• Uses 1-hour cache TTL instead of the default 15 minutes
• Shows cost summary after results

Examples:

# Quick pulse check on a topic
npx tsx x-search.ts search "cursor IDE" --quick

# Quick check what someone is saying
npx tsx x-search.ts search "agents" --from minilozio --quick

# Quick quality-only results
npx tsx x-search.ts search "startup funding" --quality --quick

Why it's cheaper:

• Prevents multi-page fetches (biggest cost saver)
• 1hr cache means repeat searches are free
• Noise filters mean fewer junk results
• Cost displayed after every search — no surprises

--from Shorthand

Adds from:username to your query without typing the full operator syntax.

# These are equivalent:
npx tsx x-search.ts search "React from:dan_abramov"
npx tsx x-search.ts search "React" --from dan_abramov

# Works with other flags
npx tsx x-search.ts search "LLM" --from minilozio --quick --quality

If your query already contains from:, the flag won't double-add it.

--quality Flag

Filters out low-engagement tweets (≥10 likes required).

npx tsx x-search.ts search "open source AI" --quality

Examples

# What are the best takes on a trending topic
npx tsx x-search.ts search "Seedance 2.0" --sort likes --since 1d --type Top

# Expert opinions with engagement filter
npx tsx x-search.ts search "AI coding assistants" --min-likes 50 --quality

# What did someone post recently
npx tsx x-search.ts profile paulg --count 10

# Deep research with multiple pages saved to markdown
npx tsx x-search.ts search "startup advice" --pages 3 --save --markdown

# Quick check on breaking news
npx tsx x-search.ts search "GPT-5" --quick

# Filter to a specific expert's takes
npx tsx x-search.ts search "transformers" --from minilozio --quality

Research Loop

For deep research (not just a quick search):

1. Broad search — search "topic" --sort likes to find key voices
2. Drill into profiles — profile username for context on interesting people
3. Read threads — thread <id> for detailed takes
4. Track ongoing — watchlist add key accounts, watchlist check periodically
5. Refine — Narrow with --since, --min-likes, --from, --no-replies
6. Save — Use --save --markdown for research docs

Why TwitterAPI.io?

Cost

TwitterAPI.io uses pay-per-use pricing with prepaid credits. No subscriptions, no monthly caps.

Per-resource costs:

Estimated costs per operation:

How x-research saves money:

• Cache (15min default, 1hr in quick mode) — repeat queries are free
• Quick mode prevents accidental multi-page fetches
• Watchlist checks run in parallel (faster, same cost)
• Cost displayed after every command so you know what you're spending
• --from targets specific users instead of broad searches

File Structure

x-research/
├── SKILL.md              # Agent instructions (Claude reads this)
├── README.md
├── package.json          # Zero dependencies
├── scripts/
│   ├── x-search.ts       # CLI entry point
│   └── lib/
│       ├── api.ts        # TwitterAPI.io wrapper
│       ├── cache.ts      # File-based cache
│       └── format.ts     # Terminal + markdown formatters
├── data/
│   ├── watchlist.json    # Accounts to monitor
│   └── cache/            # Auto-managed
└── references/
    └── twitterapi-io.md  # API reference

Security

API key handling: x-research reads your key from the TWITTERAPI_KEY env var. The key is never printed to stdout, but be aware:

• AI coding agents (Claude Code, OpenClaw, etc.) may log tool calls in session transcripts. If you're running x-research inside an agent session, your API key could appear in those logs.
• Recommendations:
  • Set TWITTERAPI_KEY as a system env var (not inline in commands)
  • Review your agent's session log settings
  • Use environment files (.env) and make sure they're in .gitignore
  • Rotate your key if you suspect exposure

Limitations

• Read-only — never posts or interacts with X
• Search quality depends on TwitterAPI.io's index (may include promoted/irrelevant content)
• min_likes uses X's native min_faves: operator for server-side filtering (efficient, no wasted API quota)
• Thread fetching uses replies endpoint + conversation_id search (may miss some replies in very large threads)
• Zero dependencies means no fancy CLI framework — basic arg parsing only
• Requires Node.js 18+ for native fetch

License

MIT