❯ is GOOGL still cheap after this run?
Not cheap now: up +120% 1Y, with P/E at 28.4 vs a 24.0 five-year median. The multiple sits in the 88th percentile.
Clean US equity data for AI agents. Skip raw feeds and give your agent structured, token-efficient data that is ready for decisions.
01 / What your agents can do
Built for agents, token-efficient by default. Computed metrics and market context arrive in compact responses, ready for reasoning instead of parsing.
❯ is GOOGL still cheap after this run?
Not cheap now: up +120% 1Y, with P/E at 28.4 vs a 24.0 five-year median. The multiple sits in the 88th percentile.
❯ is NVDA extended or still trending?
Trend intact, momentum cooled: +19.4% above its 200-day, RSI 60 neutral, volume 1.3× average. ADX says no strong trend.
❯ what's the risk into NVDA earnings?
High bar: last EPS ran +95% above its trailing 4Q baseline, revenue +51%. Next estimated Aug 12, with no guidance flagged.
❯ are insiders bailing on AAPL?
Worth flagging: 13 open-market sells, 0 buys, net −$111.7M over 90d. 0% were 10b5-1 plans, not routine plan noise.
02 / Endpoints & coverage
One compact REST + MCP surface for valuation, fundamentals, technicals, price action, earnings, dividends, filings, insiders, calendars, history, search, and coverage across US stocks + ETFs.
From broad ETFs to single-name filers, agents can answer the questions users actually ask: what changed, what it's worth, how it's trading, and what comes next.
stockcontext_valuation("GOOGL")
{
"data": {
"symbol": "GOOGL",
"as_of": "2026-06-02T09:48:05Z",
"freshness": "end_of_day",
"valuation_context": {
"anchor": "pe_ttm",
"value": 28.42,
"median_5y": 24.04,
"percentile_5y": 88,
"position": "premium"
},
"yields": {
"fcf_yield_pct": 1.42,
"earnings_yield_pct": 3.52
}
}
}Built for the context window
Your agent gets the analyst layer first, not a pile of raw feeds it has to clean, join, and recalculate.
Computed context
Valuation percentiles, trend state, drawdowns, and relative strength arrive precomputed.
Primary-source depth
Earnings, SEC filings, filing sections, and Form 4 insider activity are one call away.
Small by default
Curated responses keep context lean, with up to 30 years of daily price history when asked.
Price, profile, valuation, statements, and payouts in the shape agents need.
snapshotprofilefundamentalsvaluationpriced_individends
Technicals, price action, relative strength, drawdowns, history, and comparisons.
technicalsprice_actionhistorycompare
Earnings setup, upcoming events, SEC filings, filing sections, and insider activity.
earningscalendarfilings_listfilings_getfilings_sectioninsider
Search the supported universe and route the agent to the tools that fit.
searchcoverage
Every number traces to a primary source: market data and company filings.
Full reference03 / Context benchmark
Same workflow. StockContext returns actionable insights your agent can reason with; traditional APIs can fill the model’s context window in a couple of tool calls.
is AAPL extended here?
Clean context = better answers.
is AAPL extended here?
+ splits · dividends · shares · accession docs
Raw feeds = context bloat.
89× fewer input tokens. Better answers from optimized schemas and precomputed market context.
Measured on AAPL, Jun 2, 2026. Raw path includes 5Y daily bars, 40-period statements, 90D insider activity, and filing index. Token counts are LLM input estimates only. Costs shown at $5/M input tokens.
04 / Easy setup
Copy this prompt into Claude Code, Cursor, or another MCP-capable coding agent.
Integrate StockContext into this project so you can pull US stock and ETF context from an agent.
Step 1: Read the canonical skill
Read https://stockcontext.com/SKILL.md and follow it. If you cannot access it,
continue with the instructions below.
Step 2: Understand the project
If this is an existing codebase, inspect the language, framework, package
manager, environment-variable conventions, MCP/agent config, and current
market-data or finance code. Ask what stock-market workflow StockContext should
support unless I only asked you to connect MCP.
Step 3: Add hosted MCP
Endpoint: https://api.stockcontext.com/mcp
Header: X-API-Key: $STOCKCONTEXT_API_KEY
Store the key in the project's normal secret/env system. Do not commit it.
Claude Code:
claude mcp add --transport http StockContext https://api.stockcontext.com/mcp \
--header "X-API-Key: $STOCKCONTEXT_API_KEY"
Generic MCP JSON:
{
"mcpServers": {
"StockContext": {
"url": "https://api.stockcontext.com/mcp",
"headers": { "X-API-Key": "$STOCKCONTEXT_API_KEY" }
}
}
}
Step 4: Verify
Confirm the stockcontext_* tools loaded. Then call stockcontext_snapshot with:
{ "symbol": "AAPL" }
The response should include { "data": ... }, freshness, and as_of. If it
fails, diagnose the key/header/env/client config before trying other tools.
Step 5: Usage rules
MCP is read-only. REST/OpenAPI is the source of truth. Default sequence:
search if the user gave a company name, then coverage, then snapshot. Add
deeper routes only when needed. Preserve freshness, as_of, units, signs,
unsupported reasons, and unavailable/null explanations. Never invent data.Build production-ready stock market analyst agentsin minutes, not months.
Get a free key05 / Pricing
Essential endpoints are free. Starter unlocks every endpoint. Builder adds headroom for agents with real users.
Free
Quick stock reads
100 calls / day · 30 / min
Essential endpoints only
Starter
RecommendedFull analyst workflows
10,000 calls / month · 60 / min
≈2K five-call workflows / month
Builder
Agents with real users
75,000 calls / month · 300 / min
≈15K five-call workflows / month
No overage charges: requests stop at plan limits. Need more than Builder? We'll figure out the right limit together.
FAQ
US-listed stocks and ETFs: single names and 2–12 symbol baskets. No non-US equities and no options today; the focus is doing US equities well rather than covering everything shallowly.
Every response carries one: intraday or end_of_day for live reads, stale or degraded when a cached value was served, and unsupported when an endpoint doesn't apply to a symbol. Your agent should trust the flag and the timestamp, not the request time.
You can, and then your agent pulls full price history and statements into context, computes the ratios, percentiles, and drawdowns itself, and can get the math wrong. StockContext returns the computed answer in one compact, typed object: fewer tokens, one round-trip, nothing to invent.
Market data and primary company filings. Every number traces back to a primary source (the source filing's URL rides on the record), never a scraped aggregate.
Yes, on one key. Call the REST endpoints directly, or connect the hosted MCP server to Claude, Cursor, or your MCP client. Same data, same shapes, same key, no local process to run.
Yes, from the dashboard. Paid access stays active through the end of the billing period and you're never charged again after you cancel. Free needs no card and never expires.
Ready to build
No credit card required. Start with 100 free calls a day, then unlock powerful endpoints when you’re ready.