facticity-ai_skill

Overview

This skill provides a complete integration with the Facticity.AI API for automated fact-checking, claim extraction, transcription, link reliability scoring, credit checks, and async task monitoring. It routes user commands to the appropriate API endpoints, returns raw JSON responses, and enforces API key presence and usage rules. Designed for workflows that need fast, reproducible verification and content analysis.

How this skill works

The skill first verifies the FACTICITY_API_KEY is configured; if not, it returns a short onboarding message with setup steps. It maps command prefixes or input patterns to six operations (fact-check, extract claims, transcribe link, check link reliability, get credits, and check task status) and issues HTTP requests to https://api.facticity.ai using the required headers and JSON bodies. Responses are returned verbatim so callers can parse classification, evidence, transcripts, bias/quality metrics, and task IDs. Async tasks return task_id values that can be polled with the check-task-status route.

When to use it

• Verify the truthfulness of a single claim or short statement quickly.
• Extract and list claims from articles or social video URLs before deeper analysis.
• Transcribe videos or audio links (YouTube, TikTok, Instagram) to enable claim extraction.
• Check a news source URL for bias and quality using MediaBias data.
• Monitor long-running fact-check tasks produced in async mode.
• Check remaining API credits and manage usage limits.

Best practices

• Ensure FACTICITY_API_KEY is set as an environment variable or in your OpenClaw config before calling the skill.
• Use /fact-check for short, focused claims; switch to async mode for lengthy or complex checks and poll with /check-task-status.
• Pass video URLs to /extract-claim or /transcribe-link to obtain transcripts before extracting claims.
• Call /get-credits regularly in batch workflows to avoid unexpected rate limits.
• Treat the skill output as machine-readable JSON and implement your own post-processing for UI display and citation formatting.

Example use cases

• Feed a viral tweet into /fact-check to get immediate classification, evidence, and source breakdown.
• Run /extract-claim on a YouTube link to generate a transcript and list of claims for a moderation pipeline.
• Use /link-reliability/check on article URLs to tag content with bias and quality labels in a news aggregator.
• Start a long fact-check in async mode and poll with /check-task-status until completion for background processing.
• Call /get-credits from a scheduler to track remaining tokens and plan bulk extractions.