Structured market intelligence for AI trading agents

Give your trading agent something trustworthy to act on.

SignalPack turns messy market research into source-backed strategy packs, policy checks, and audit trails that agents can read and humans can inspect.

Request early access See how it works

SignalsSource-backed

MCPAgent-ready

PolicyApproval-first

SignalPack workspace

Political Trades pack

Demo artifact

Latest reference strategy

Pelosi Tracker

v0.3

updated today

NVDA NVIDIA

Repeat exposure + AI infrastructure overlap

12.0%

High

Source: House PTR, May 29

AVGO Broadcom

Recent buy signal, semiconductor cluster strength

10.5%

Med

Source: House PTR, May 24

PANW Palo Alto Networks

Cybersecurity theme + fresh filing proximity

8.5%

Med

Source: House PTR, May 17

Execution disabled: SignalPack provides research artifacts and order drafts, not autonomous trading.

What it is

A research layer your agent can trust.

Trading agents can already connect to brokers and exchanges. SignalPack helps with the harder part: clean inputs, source disclosure, methodology, guardrails, and reviewable outputs before anything reaches an execution workflow.

Strategy packs

Curated research artifacts that include signals, reference weights, source context, and methodology notes.

Guardrails

Approval-first policies that check concentration, asset type, stale data, missing sources, and turnover.

Audit trails

Versioned outputs that show what changed, why it changed, and which evidence supports the result.

Strategy packs

Start with Political Trades. Expand to any data-driven theme.

Political Trades

First pack

Congressional disclosure signals with source links, filing lag, methodology notes, and risk flags.

Pelosi TrackerCongress ClusterFresh Disclosures

Insider Buying

Planned

Open-market executive buying translated into explainable, agent-readable signal artifacts.

Cluster BuysRepeat BuyersSize Anomalies

13F Copycat

Planned

Institutional filing changes packaged with freshness, confidence, and position-change context.

New PositionsAddsConviction Changes

Workflow

From source data to approval-ready output.

The goal is not to let an agent improvise a portfolio from scraped pages. The goal is to give it structured research it can cite, compare, and validate.

Choose a strategy pack

Start with a source-backed research pack like Political Trades, then pick the signal or reference strategy your agent should inspect.

Review the evidence

Every position includes source disclosures, freshness, methodology notes, confidence, and what changed since the last version.

Apply your policy

SignalPack checks limits like max position size, no options, turnover, stale data, and required human approval before any handoff.

Hand off safely

Your agent can draft orders for Robinhood, Alpaca, or another broker workflow, but execution remains separate and approval-based.

Safety layer

Guardrails before handoff.

SignalPack does not execute trades. It validates the research artifact against your policy so your agent can show what is allowed, what is blocked, and why.

Human approvalRequired

Max position12% cap

OptionsBlocked

Source disclosurePresent

Data lagShown

Example result: order draft created, but execution requires user approval and broker-side confirmation.

API + MCP

Built for agents. Still easy to inspect.

Use SignalPack through a normal API, MCP tools, webhooks, or file exports. The output is structured enough for agents and clear enough for humans reviewing the decision.

get_latest_strategy_pack

explain_position

get_recent_signals

compare_versions

validate_policy

export_audit_trail

{
  "pack": "political-trades",
  "strategy": "pelosi-tracker",
  "version": "2026.06.02-v0.3",
  "as_of": "2026-06-02T12:00:00Z",
  "positions": [
    {
      "ticker": "NVDA",
      "reference_weight": 0.12,
      "confidence": "high",
      "rationale": "Repeat exposure + AI infrastructure overlap",
      "source_count": 3,
      "freshness": "recent filing, delayed transaction"
    }
  ],
  "policy": {
    "human_approval_required": true,
    "execution_enabled": false,
    "blocked_reasons": []
  }
}

Source-backed

Signals link back to the filings, records, or datasets that produced them.

Agent-readable

Clean JSON, MCP tools, webhooks, and CSV exports instead of scraped pages.

Human-inspectable

Every artifact has a model card, version history, assumptions, and plain-English rationale.

Policy-checked

Guardrails catch concentration, stale data, missing sources, unsupported assets, and approval requirements.

Versioned

Agents can ask what changed, when it changed, and why the current output differs from the last one.

Broker-neutral

SignalPack does not custody assets or execute trades. It prepares safer research inputs for your chosen workflow.

Early access

Join the SignalPack waitlist.

Get notified when the first Political Trades pack, API preview, and MCP server are ready for private alpha.

Prototype form only. No investment advice, custody, or execution.