FalsifyLab — concise second‑pass review

Intro

FalsifyLab presents itself as a compact, developer‑oriented MCP (memory+connector+plugin) data layer that packages finance signals and curated feeds for AI agents. The site emphasizes a single operator running trading bots and publishes demonstrations and comparison materials for agent integrations.

What it does

According to the site, FalsifyLab provides 13 finance tools that AI agents can query for live or near‑live observational signals. The product is shipped as a Python package (pip install falsifylab-alpha-mcp) and is described as having zero runtime dependencies (Python standard library only). Example recipes and MCP client wiring are provided for agent clients such as Claude, Cursor, Cline and Windsurf.

Key features (site claims)

• 13 finance tools for AI agents; the site lists 3 tools as free and 10 as Pro‑only.
• Curated feeds that the site names (including confluence_today and feeds referenced as earnings_drift, token_unlocks, fed_comm).
• Pulse: a cross‑feed snapshot the site describes as updating every minute.
• Public demo agent: a public demo that the site says polls feeds on roughly a 15‑minute cadence and writes observations.
• One‑command install and example recipes for MCP clients.

Pricing and practical value

• Free tier: 3 tools available with no signup required (presented for light experimentation and demos).
• Pro: listed at $19/month and described on the site as unlocking all 13 tools with real‑time access.
• Pro Plus and Teams: higher priced tiers are listed on the pricing page (site snippets reference Pro Plus and Teams with higher costs and added features).

The site positions Pro as the advertised entry point for full access. Whether the feeds and cadence meet production needs requires direct testing (see evaluation checklist below).

Who this is for / not for

Who this is for
- Developers building MCP‑aware agent integrations who want prepackaged finance feeds and example recipes.
- Finance practitioners and researchers seeking agent‑friendly observational signals and working examples.
- Teams evaluating alternatives to other crypto/finance data APIs (the site includes a comparison page).

Who this is not for
- Organizations that require a fully enumerated, auditable table of source provenance for every feed — the site does not publish a complete detailed provenance table.
- Organizations that need formal enterprise SLAs, detailed security audits, or comprehensive compliance documentation — those operational details are not fully enumerated on public pages.

Risks and limitations (observed or missing on the site)

• Access limits: the free tier is limited to three tools and the site indicates restricted result counts; full access requires a paid tier for most tools.
• Cadence and latency: Pulse is presented as updating every minute, while the public demo agent is described as polling roughly every 15 minutes — these are snapshot and polling models, not continuous streaming guarantees.
• Privacy and logging: the public site is stated to collect no personal data and use no analytics cookies, but the privacy policy notes Cloudflare edge logs (IP and user‑agent) are recorded for DDoS protection.
• Missing operational detail: the site does not publish precise API rate limits, retention periods, detailed authentication mechanics, explicit per‑feed provenance, or formal security/audit disclosures.

How to evaluate before production

• Validate feed coverage and freshness by exercising the specific feeds you need (use the public demo where possible).
• Confirm practical rate limits and expected result shapes under your expected query patterns.
• Ask the operator for explicit provenance, licensing/redistribution terms, retention policies for logs and data, and security controls if those matter for your use case.
• Test integration with your target MCP client and validate cost/limits against your usage profile.

FAQ (short)

Q: What does the free tier include?
A: The site states three tools are free with no signup required, intended for quick experiments and demos.

Q: How is it installed and wired into an agent?
A: The site documents a single pip install (pip install falsifylab-alpha-mcp) and provides example recipes for MCP clients.

Q: What is the difference between Pulse and the demo agent?
A: Pulse is described as a minute‑level cross‑feed snapshot; the public demo agent is described separately and is said to poll feeds on roughly a 15‑minute cadence.

Q: What privacy information is published?
A: The site states public pages collect no personal data and use no analytics cookies; the published privacy policy notes Cloudflare edge logging of IP and user‑agent for DDoS protection and lists an effective date and contact address.

Conclusion

FalsifyLab is presented as a lightweight MCP data layer aimed at developers and finance practitioners who want agent‑oriented finance feeds and ready‑to‑use recipes. The public materials support quick experimentation but omit several operational and provenance details that matter for production. Validate feed behavior, rate limits, retention, licensing, and security with the operator before committing to production use.