Hooklayer FAQ

Quick answers about Hooklayer.

The 15 questions Claude users, AI engineers, and content marketers ask most. If you need more depth: read the docs or try the playground.

What is Hooklayer?

Hooklayer is an MCP + REST API exposing 7 viral-content intelligence tools — TikTok account analysis, hook scoring, viral remix, trend pulse, voice DNA matching, virality prediction, and template discovery — for AI agents running in Claude Desktop, Cursor, Claude Code, n8n, and the OpenAI Agents SDK.

How is Hooklayer different from Composio's TikTok MCP?

Composio gives raw TikTok actions (post, fetch, analytics). Hooklayer returns intelligence: viral DNA scores, replicability ratings, steal maps, and a recommended_chain field pre-filling next-tool calls so AI agents chain themselves. Wrap, don't replace.

How is Hooklayer different from generic TikTok MCPs like seym0n/tiktok-mcp?

Generic TikTok MCPs are data pipes — search, scrape, return raw video data. Hooklayer is the intelligence layer — viral DNA scoring, hook scoring, viral prediction, voice matching, recommended_chain auto-fire. Different layers. Most production stacks combine both: generic MCP for volume, Hooklayer for intelligence.

Does Hooklayer extract video transcripts?

Yes — Whisper transcription is bundled into analyze_account (5 transcripts per call), viral_remix (URL → transcript → remix), and match_voice (URL reference samples → transcripts → voice profile). Other TikTok MCPs typically require a separate Whisper integration; Hooklayer handles it natively at no extra credit cost.

Can I combine Hooklayer with a generic TikTok scraper MCP?

Yes — and most production stacks do. The pattern: generic MCP (seym0n/tiktok-mcp, Apify TikTok actor) scrapes raw video/comment/hashtag data at scale into a dataset. Hooklayer scores the specific handles or videos an AI agent picks from that dataset. Generic for breadth, Hooklayer for depth.

Do I need OAuth to use Hooklayer?

No. Hooklayer accepts simple Bearer hl_live_ API keys for direct use. OAuth 2.1 with Dynamic Client Registration is supported when Hooklayer is added as a Claude.ai custom connector. Choose whichever matches your client.

What does analyze_account actually cost?

Five credits per call. The call bundles one TikTok scrape, five Whisper transcriptions, one pattern-analysis Sonnet call, and one meta-analysis Sonnet call producing the viral DNA scores and recommended_chain — into a single response.

Can my agent chain Hooklayer tools automatically?

Yes. analyze_account returns a recommended_chain field with pre-filled match_voice, trend_pulse, and viral_remix calls — parameters and reasons included. Claude / Cursor agents read the chain field and fire the next three calls without prompt engineering.

Does Hooklayer work with Cursor and n8n?

Yes. Hooklayer is a standard streamable-HTTP MCP server, so it drops into Claude Desktop, Cursor, n8n MCP nodes, and the OpenAI Agents SDK with the same endpoint URL and the same hl_live_ Bearer key. One credits ledger across all clients.

Is TikTok scraping legal?

Hooklayer only reads publicly visible TikTok data and respects robots.txt. We never extract private user data, DMs, or geofenced content. Use the data for research, content strategy, and remix inspiration — never republish raw scraped content as your own.

What is a viral DNA score?

A 0-100 composite metric measuring how predictably a creator's format produces virality. It rolls up replicability, originality, consistency, hook reuse rate, and audience fatigue into one number you can compare across creators. Every score ships with cited evidence.

Why not just use Claude's web search to analyze TikTok?

Claude's search returns text snippets. Hooklayer returns structured fields — viral_dna, format_fingerprint, pattern_analysis, steal_map, recommended_chain — that agents can branch on without re-parsing prose. The structure is what makes agentic chaining possible.

Can I cache results to save credits?

Hooklayer caches upstream data by handle for one hour and caches AI-meta-analysis results by SHA-256 hash for 24 hours. Same input returns the same response from cache deterministically. Credits are still charged on cache hits — we sell determinism, not cost relief.

What is the difference between MCP and an API?

APIs are stateless and built for deterministic code. MCP adds runtime tool discovery, structured tool catalogs, and stateful sessions designed for probabilistic AI agents. Hooklayer ships both — same backend, two interfaces (REST at /api/v1/* and MCP at /api/mcp).

Can I post TikToks with Hooklayer?

No — Hooklayer v1 is read and analyze only. Every tool carries readOnlyHint: true. Posting is roadmapped. For now, pair Hooklayer (intelligence) with a posting-capable MCP like Composio or TokPortal for full content pipelines.

How does the credit ledger work?

Each hl_live_ key has a credit balance with two buckets: subscription (resets monthly) and pay-as-you-go (never expires). Tool calls deduct atomically; failures don't charge. Free tier: 100 lifetime credits at signup, no card required.

Will Hooklayer slow down Claude Desktop?

No. Hooklayer registers 7 tool definitions (~5K tokens total). The slowdown people see in Claude Desktop comes from running 20+ MCP servers stacked together; Hooklayer's focused 7-tool surface stays well below that ceiling and loads in milliseconds.

How do I try Hooklayer without signing up?

Visit hooklayer.dev/playground — the anonymous playground runs analyze_account, score_hook, and trend_pulse against real data with no key required (5 calls per hour per IP). When you're ready, sign up for 100 free lifetime credits and unlock all 7 tools.

Question not answered?

The full reference + curl examples + per-tool FAQs live in the docs. Or DM us — we read every message.