X-Scout

X/Twitter intelligence scraper. Search tweets, scrape profiles, pull comments, auto-transcribe videos. Classify tweets as replicable methods vs content. CLI...

Security Scan

The skill's requested resources and runtime behavior match its stated purpose (Twitter scraping, optional classification and transcription); it does phone-home and stores API keys locally, but those behaviors are disclosed in the README and consistent with the feature set.

Purpose & Capability

Name/description (Twitter/X scraping, optional method classification and transcription) align with required binaries (python3, curl), the required TWITTERAPI_KEY, and optional keys (OpenRouter, Deepgram). The included Python script and setup script implement scraping, classification and transcription paths that match the description.

Instruction Scope

SKILL.md and setup.sh instruct the agent/user to run setup.sh, create a .env and a config file in ~/.x-scout, and then run x_scout.py. The runtime instructions do not attempt to read unrelated system files, but they do persist keys and an install_id to disk and the runtime code performs telemetry (silently POSTs usage data including an install_id and a hashed query to clawagents.dev). These behaviors are disclosed in the SKILL.md, but they are privacy-relevant and worth the user's attention.

Install Mechanism

No remote archive downloads or obscure install hosts. setup.sh creates a local venv and runs pip install -r requirements.txt (requests, python-dotenv). No high-risk download URLs or extract-from-arbitrary-URL steps are used. yt-dlp is optional and not automatically downloaded.

Credentials

Only TWITTERAPI_KEY is required (declared as primary). Other API keys (OpenRouter, Cerebras, Deepgram) are optional and used only for optional features (method detection, query optimization, transcription). The script does store these keys to .env and ~/.x-scout/config.json in plaintext, which is reasonable for a CLI tool but is a privacy/security consideration for secret management.

Persistence & Privilege

always:false (no forced global inclusion). The setup writes files to $SCRIPT_DIR/.env and ~/.x-scout/config.json and registers an install_id; the runtime reports usage on each run to the analytics endpoint. The skill does not modify other skills or agent-wide settings. The persistent telemetry + stored install_id means activity can be correlated over time; this is disclosed but worth user consideration.

Assessment

This skill is internally consistent with its stated purpose, but review the privacy implications before installing. Setup.sh will: create a virtualenv, write your API keys to a .env in the script directory and to ~/.x-scout/config.json (stored in plaintext), and POST a registration to https://clawagents.dev. The runtime script will silently POST usage telemetry (install_id, a short hash of queries, result counts, errors) to the same analytics endpoint on every run. If you want to reduce risk: (1) avoid entering highly privileged/long-lived credentials if not necessary (use separate limited API keys), (2) inspect or modify setup.sh to skip the analytics registration or block outbound calls to clawagents.dev at the network level, (3) consider running the tool in an isolated environment (container or throwaway VM), and (4) rotate or revoke keys stored by the tool when you stop using it. If you need a different behavior (no telemetry, encrypted key storage), ask the author for a build that omits telemetry or implement those changes locally before using.

X/Twitter intelligence scraper. Search by keyword, scrape profiles, pull comments, auto-transcribe videos. Classifies tweets as replicable methods vs general content.

Setup

Before first use, run the setup script to configure your API keys:

bash setup.sh

This prompts for your TwitterAPI.io key (required) and optional keys for method detection and video transcription. Your install is registered with ClawAgents for usage tracking.

If you already have keys configured, set them as environment variables:

export TWITTERAPI_KEY=your_key_here

Modes

Search tweets by keyword:

python3 x_scout.py --search "ai agent" --limit 20

Scrape profile posts:

python3 x_scout.py --profile @elonmusk --limit 10

Pull comments/replies on a tweet:

python3 x_scout.py --comments "https://x.com/user/status/123456"

Full intel (tweet + video + comments + transcription):

python3 x_scout.py --intel "https://x.com/user/status/123456"

Options

Flag	Description	Default
--search "query"	Search tweets by keyword	--
--profile @handle	Scrape profile posts	--
--comments <url>	Pull replies to a tweet	--
--intel <url>	Full intel on a tweet	--
--limit N	Max results	20
--since YYYY-MM-DD	Date filter	180 days ago
--no-methods	Skip method detection	methods on by default
--no-transcribe	Skip video transcription	transcribes if key set
--json	Output as JSON	table view

Method Detection

When enabled (default), X-Scout classifies each tweet as:

• METHOD: Describes a specific tool, technique, or workflow (replicable)
• CONTENT: General commentary, results showcase, promotional

For METHOD tweets, it extracts: method name, tools required, category, complexity, and a summary. Requires an OpenRouter API key.

Auto-Transcription

Any tweet with an embedded video is automatically:

1. Downloaded via yt-dlp
2. Transcribed via Deepgram
3. Transcript included in output and method classification

Requires a Deepgram API key (set during setup).

Output

Table view (default) shows: author, likes, views, retweets, media type, transcript status, and tweet preview.

JSON view (--json) outputs full structured data for piping to other tools.

Required Keys

Key	Required?	What It Does	Cost
TWITTERAPI_KEY	Yes	Tweet search, profile scrape, replies	~$50/mo
OPENROUTER_API_KEY	Optional	Method detection via Grok	Pay-per-use
CEREBRAS_API_KEYS	Optional	Query optimization	Free tier
DEEPGRAM_API_KEY	Optional	Video transcription	Free tier