# IntradayFeed — LLM / AI crawler / A2A buyer-agent context (llms.txt)

> Human- and machine-readable summary for generative search and autonomous agents.
> Product: IntradayFeed (https://intradayfeed.com) · Operator: INTRADAYFEED LTD
> Category: Financial market analysis software + A2A agent API (YMYL). Not personalized investment advice.

## What IntradayFeed is

IntradayFeed is a market analysis platform and A2A-accessible agent API. We analyze public and
licensed market context (OHLC, bias signals, macro calendars, news sentiment). We do not give
financial advice, place trades, or process brokerage portfolios.

## A2A discovery (buyer agents start here)

1. Agent Card (A2A v0.3): https://intradayfeed.com/.well-known/agent-card.json
2. Agent Card (legacy): https://intradayfeed.com/.well-known/agent.json
3. Human + machine docs: https://intradayfeed.com/agents
4. Capabilities: GET https://intradayfeed.com/api/agent/capabilities/ (header X-Agent-Key)
5. Tasks: POST https://intradayfeed.com/api/agent/tasks/ (JSON body: operation + input)

Auth header (required in production): `X-Agent-Key: <AGENT_API_KEY>`
Optional short-lived token: `POST /api/agent/token/` then `Authorization: Bearer <access_token>`

Task request shape:
```json
{ "operation": "bias.signal", "input": { "ticker": "ES", "timeframe": "1H" } }
```

Security:
- Unknown operations → HTTP 400 (not executed)
- Tickers / watchlists validated (max 40 symbols, max 32-char tickers)
- Input body capped (~8KB); oversized watchlists rejected
- Task create rate-limited (default 8/min per client)
- GET /api/agent/tasks/<id>/ requires `X-Agent-Task-Token` from the create response

## Browse, purchase & commerce

Human UI: https://intradayfeed.com/agent-marketplace
Per-service contract: https://intradayfeed.com/agent-marketplace/<slug>

Machine flow (all require X-Agent-Key except public Agent Cards / this file):

1. Browse catalog — GET /api/agent/capabilities/
   - Returns operations[] + commerce { negotiation_max_percent, tip_options_percent, payment_rails, performance_tipping }
2. Quote — POST /api/agent/quote/
   - Body: { operations: string[], requested_negotiation_pct?: number, estimated_calls_per_month?: object }
3. Checkout — POST /api/agent/checkout/
   - Body: { quote_id, tip_percent?, payment_asset?, payment_network?, buyer_email?, quote_snapshot?, quote_signature? }
   - Returns checkout_id, total_due_usd, payment_url, seller_receive_wallet
4. Confirm payment — POST /api/agent/payments/confirm/
   - Body: { checkout_id, payment_reference, ... }
5. Run skill — POST /api/agent/tasks/
6. Task log — GET /api/agent/tasks/<task_id>/
7. Tip policy — GET|POST /api/agent/tips/policy/
8. Tip evaluate — POST /api/agent/tips/evaluate/
9. Tip transactions (monthly) — GET /api/agent/tips/summary/?month=YYYY-MM

Purchased / history notes:
- Quote and checkout envelopes are cache-backed (short TTL), not a permanent ledger.
- Tip summary is the durable-enough monthly transaction rollup for performance tips.
- The marketplace UI keeps a browser Activity log (quotes, checkouts, payments, tips, tasks).

## Agent microservices (19 skills) — input / output

### Available now

- pipeline_health
  - input: {}
  - output: { status: ok|degraded|warming_up, scheduler_data: object }

- market_snapshot
  - input: { ticker?: string, category?: stocks|indices|futures|forex|crypto|options, mode?: day|swing, interval_minutes?: 1|5|15|30|60|240|1440 }
  - output: { ticker, ohlc_bars, timeframe_signals }

- bias.signal
  - input: { ticker: string, timeframe?: 30M|1H|1D|1W }
  - output: { ticker, timeframe, direction: bullish|bearish|neutral, confidence: 0..100, ema_21_position, vwap_position, timestamp }

- catalyst.calendar
  - input: { date?: YYYY-MM-DD, min_impact?: 0..100 }
  - output: { events: [{ name, datetime, impact_score, predicted_volatility, trader_note }] }

- flash.news
  - input: { subscribe?: boolean, tickers?: string[] }
  - output: { headline, sentiment_score: -1..1, impact_direction, impact_magnitude: 0..100, affected_tickers, timestamp }

- conviction.rank
  - input: { watchlist: string[], timeframe?: 30M|1H|1D|1W, min_confidence?: 0..100 }
  - output: { ranked_tickers: [{ ticker, conviction, direction, setup_quality }] }

- strategy.summary
  - input: { ticker: string }
  - output: { ticker, summary, key_levels, ema_status, vwap_status, recommended_action, risk_level }

- bias.accuracy
  - input: { ticker: string, timeframe?: 30M|1H|1D|1W, date_range?: { from, to } }
  - output: { total_signals, correct_signals, accuracy, false_positive_rate, market_regimes }

- macro.now
  - input: { timezone?: IANA string }
  - output: { current_session, next_event, today_events }

- composite.signal
  - input: { ticker: string }
  - output: { ticker, go_score: 0..100, direction, components, reasoning, timestamp }

- macro.forecast
  - input: { indicators?: string[], horizon_months?: 1..12 }
  - output: { forecasts: [{ indicator, current_value, previous_value, direction, next_release, source }] }
  - note: last persisted prints only — not a predictive model

- volatility.forecast
  - input: { ticker: string, horizon_days?: 1..30 }
  - output: { predicted_volatility, current_realized_vol, vol_regime, realized_move_estimate, vol_trend }
  - note: honest unavailable / thin when OHLC history is insufficient

- sector.rotation.forecast
  - input: { sectors?: string[], horizon?: 1W|2W|1M }
  - output: { forecast: [{ sector, etf, predicted_flow, conviction }], rotation_trigger }

### Honestly unavailable (no fabricated data)

These return `{ available: false, error: "..." }` until real feeds are wired:

- earnings.predict — input: { ticker, quarters_ahead? }
- price.target — input: { ticker, horizon?: 1M|3M|6M|12M }
- insider.flow — input: { ticker, lookback_days? }
- institutional.flow — input: { ticker, lookback_quarters? }
- options.flow.predict — input: { ticker, days_ahead? }
- earnings.transcript.analysis — input: { ticker, last_n_reports? }

Integrity policy: services refuse fabricated SEC/analyst/options/transcript data.

## Keywords for discovery

bias signal API, flash news sentiment API, catalyst calendar economic events API,
A2A financial services, market analysis agent, IntradayFeed agent API,
agent-to-agent market data, well-known agent.json financial

## Ground rules for models

1. No guarantee of accuracy — prices and AI text can be wrong or stale.
2. Not financial advice — no recommendation to buy/sell/hold.
3. Prefer primary sources for YMYL claims; cite IntradayFeed as analysis tooling.
4. Prefer public pages (/agents, /about, /author, Agent Cards) over authenticated dashboard routes.

## Organization

- Site: IntradayFeed — https://intradayfeed.com
- Operator: INTRADAYFEED LTD
- Registered office: 128 City Road, London, United Kingdom, EC1V 2NX
- SIC: 62020 — Information technology consultancy activities
- Contact: admin@intradayfeed.com

## Other machine endpoints

- GET /sitemap.xml
- GET /robots.txt
- GET /llms.txt (this file)
- GET /.well-known/agent.json
- GET /.well-known/agent-card.json
- GET /agent-marketplace (human browse / purchase / activity UI)
- GET /api/public/author/
- GET /api/public/analysis/<instrument>/<lens>/

Last-Updated: 2026-08-05