Prompts and resources - StockContext

The server ships four read-only resources and four prompts alongside the tools.

Resources

Resources are JSON reference documents an agent can read without spending a tool call or a plan quota.

Resource	Returns
`stockcontext://reference/error-codes`	Error codes with their `retryable` flag and meaning.
`stockcontext://reference/sectors`	GICS sector to SPDR sector-ETF reference.
`stockcontext://reference/openapi`	The REST OpenAPI schema.
`stockcontext://coverage/supported-tickers`	Every supported stock and ETF ticker.

From an MCP client, list what is available and read one by URI:

{ "method": "resources/list" }

{
  "method": "resources/read",
  "params": { "uri": "stockcontext://reference/error-codes" }
}

Resources are best-effort. The stockcontext://reference/openapi resource embeds an { "error": ... } object if the upstream schema fetch fails, rather than failing the read.

Prompts

Prompt	Purpose
`decision_brief(symbol)`	Single-symbol context brief.
`earnings_risk(symbol)`	Earnings setup and risk for one symbol.
`insider_scan(symbol)`	Form 4 activity review for one symbol.
`compare_basket(symbols)`	2–12 symbol comparison.

Prompts are client-side templates. They tell the agent which tools to call in what order; they do not run a server-side workflow. Invoke one the way your client invokes prompts:

{
  "method": "prompts/get",
  "params": {
    "name": "decision_brief",
    "arguments": { "symbol": "AAPL" }
  }
}

The client gets back the template text and the agent calls the tools it names. The templates also carry the read guardrails: valuation percentiles and labels (near_5y_low, etc.) rank a multiple against the company’s own history only — never peers, sector, or fair value — so a low rank is not a cheapness verdict, and earnings guidance excerpts are source text, not parsed consensus. For the worked-out version of each flow, see agent recipes.

​Resources

​Prompts

Resources

Prompts