Enter your organization and monitored email to generate the command.The identity is sent only to sec.gov as the required request User-Agent.
Catalyst Edge Research · Free Local Beta · v0.1.4
Ask your AI research agent what changed for a ticker, why it matters, what contradicts it—and show every source and missing-data warning.
Fixed, sanitized AAPL example
This 90-second walkthrough highlights SEC evidence from the approved August 3, 2026 example. The current default also adds neutral GDELT publisher discovery with a mandatory citation and link. It is a demonstration of provenance, not performance.
catalyst_edge_score gathers recent evidence without portfolio access or trading permissions.
The fixed example found a July 31 Form 10-Q and July 30 Form 8-K, preserving their SEC provenance.
Insider activity returned no qualifying observation; options and technical data require licensed sources; social is off by default.
The neutral 50 is deterministic and unbacktested—not a signal. Open the filings and inspect the facts.
Local and self-serve
Python 3.10+ and uv are required. Use a real monitored SEC identity; the local evidence store uses a private default path unless you override it.
Published on PyPI and the Official MCP Registry.
Enter your organization and monitored email to generate the command.The identity is sent only to sec.gov as the required request User-Agent.
Use the same pinned PyPI package through Claude Desktop’s manual stdio configuration.
.mcpb is not distributed until publisher trust can be verified in production.Report a successful local activation. It links only this browser’s random ID to paid-intent measurement; no ticker, prompt, portfolio, IP address, or local evidence is stored.
Measurement only · No payment
Would you definitely pay $29/month for zero-install hosted access and managed updates? This exact offer is a demand test, not a product account, preorder, or promise to build.
Truthful by construction
rules_v1 is deterministic; it has no demonstrated predictive edge.
Missing, disabled, unavailable, and rights-gated evidence stays explicit and contributes zero direction.
No recommendations, execution, brokerage access, portfolio data, or promise of investment performance.
Tool inputs and evidence remain in your local process and SQLite store. Do not submit personal financial information.