Skip to main content
Glama

Stocklake — AI Stock Intelligence

by dev.stocklake

Get Signals

get_signals
Read-onlyIdempotent

AI-screened stock signals recently surfaced by the Stocklake pipeline — sourced from news analysis, sector screening, and sentiment signals.

Shows what the pipeline noticed in the last 24 hours (falling back to the most recent signals regardless of age if nothing has fired in that window — see window in the response). This reflects what the AI pipeline found, not whether Stocklake's own internal trading engine still holds it live — a signal it later dropped or let expire is still shown here, since that's a fact about our own trading state, not about the signal's informational value.

Parameters:

  • direction: "POSITIVE" | "NEGATIVE" | "NEUTRAL" (default: all). NEUTRAL covers both a flat/undecided read AND a genuinely two-sided idea (real opposing bull/bear theses on the same symbol) — in the latter case signal_score is the STRONGER of the two sides (see signal_score below), so a high score alongside NEUTRAL means "real conviction here, just no directional consensus," not "nothing going on." The two-sided detail is in rationale.

  • min_signal_score: minimum composite signal score 0-100 (default 60) — a blend of conviction/confidence/flag_score, source track record, and real technical factors. This is the field to filter on. Always compared against a single number, including for NEUTRAL/two-sided ideas — a result is never returned below your threshold on both sides.

  • min_conviction, min_flag_score: DEPRECATED, ignored for filtering — kept in the signature only so existing callers don't hard-fail; ai.stocklake.dev's internal scoring retired the raw conviction/confidence/flag_score triad in favor of signal_score. Passing a non-default value here has no effect and is logged for a planned removal.

  • source: filter by signal source — "news" | "screener" | "sentiment" | "social" (default: all)

  • limit: max results to return (default 50, max 50). Each returned signal counts as one call toward your daily limit.

Returns:

  • count: number of signals returned

  • window: "24h" (fresh activity found) or a fallback note when the last 24h had nothing and the response instead shows the most recent signals overall

  • signals[]: each with symbol, direction, signal_score (0-100), signal_score_band (human-readable label — "Weak"/"Moderate"/"Strong"/"Very Strong"), source, rationale, expires, flagged_at

Pro tier only — AI pipeline cost attached. For informational purposes only. Not financial advice.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
limitNoMax results to return (default 50, max 50). Each returned signal counts as one call toward your daily limit.
sourceNoFilter by signal source: "news" | "screener" | "sentiment" | "social" (default: all).
directionNoFilter by direction: "POSITIVE" | "NEGATIVE" | "NEUTRAL" (default: all). NEUTRAL covers both a flat/undecided read and a genuinely two-sided idea (see signal_score below).
min_convictionNoDEPRECATED, ignored for filtering — kept in the signature only so existing callers don't hard-fail; internal scoring retired the raw conviction/confidence/flag_score triad in favor of signal_score. Passing a non-default value here has no effect and is logged for a planned removal.
min_flag_scoreNoDEPRECATED, ignored for filtering — kept in the signature only so existing callers don't hard-fail; internal scoring retired the raw conviction/confidence/flag_score triad in favor of signal_score. Passing a non-default value here has no effect and is logged for a planned removal.
min_signal_scoreNoMinimum composite signal score, 0-100 (default 60) — a blend of conviction/confidence/flag_score, source track record, and real technical factors. This is the field to filter on.

Output Schema

TableJSON Schema
NameRequiredDescriptionDefault

No arguments

TDQS

A4.8/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Beyond the readOnlyHint and idempotentHint annotations, the description adds critical behavioral details: the fallback to older signals when nothing fires in 24 hours, the fact that dropped or expired signals are still shown, and the 'Pro tier only — AI pipeline cost attached' warning. No contradiction with annotations.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is lengthy but every sentence carries useful information—parameter semantics, edge cases, return structure, and cost. It's well-organized with clear sections and front-loaded with the core purpose, so the length is fully justified.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the tool's complexity, the description is comprehensive: it covers all parameters, return fields (count, window, signals), fallback behavior, deprecation, and the Pro tier cost. It complements the existing output schema by explaining the nuances of signal_score_band and the window field, so nothing an agent needs to call it correctly is missing.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters5/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

The description goes far beyond the input schema: it explains that NEUTRAL covers both flat and two-sided ideas, that signal_score reflects the stronger side for two-sided ideas, that min_signal_score is always compared against a single number, and that min_conviction/min_flag_score are deprecated and ignored. This is essential for correct filtering.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states that the tool returns 'AI-screened stock signals' from the Stocklake pipeline, sourced from news analysis, sector screening, and sentiment signals. It distinguishes itself from siblings like get_news_feed or get_market_movers by focusing on the pipeline's processed signals rather than raw data, so an agent can tell exactly what it does.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

It provides useful context on the 24-hour window and fallback behavior, and clarifies that signals reflect AI findings rather than the trading engine's current state. However, it does not explicitly name alternative tools or state when to prefer this over get_screener or get_news_feed, so the agent must infer usage from the description.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All Stocklake — AI Stock Intelligence toolsInspect
Try in Browser
RayrunBuild, deploy, and operate MCP servers without becoming an infrastructure team.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

TDQS

A4.3/5.0
Disambiguation4/5

Most tools carve out clearly distinct jobs: per-stock quotes, batch quotes, earnings calendars, market pulse, news, signals, screener, watchlist. A few pairs—notably get_stock vs get_stocks and get_market_assessment vs get_market_pulse—are close enough that an agent would need to read descriptions carefully to pick the right one.

Naming Consistency5/5

All tool names follow a uniform get_<snake_case_noun> pattern with no mixed conventions. The singular/plural pair get_stock and get_stocks is a deliberate batch-version distinction rather than an inconsistency, and compounds like earnings_intelligence vs market_assessment remain predictable.

Tool Count4/5

With 17 tools, the server is slightly over the ideal 3-15 range, but the count is well-justified by the breadth of the stock-intelligence domain: per-stock, batch, historical, market-level, earnings, insider, news, signals, screeners, and watchlists all have a dedicated entry point. It is a larger surface area than ideal, but none of the tools feel redundant or purely decorative.

Completeness4/5

The tool surface gives agents full read-oriented workflows: single-stock data, batch lookups, historical prices, indicator history, news, research bundles, insider activity, market health, earnings, signals, movers, sector intelligence, and screening all exist. The main gap is discovery—there is no symbol/name search tool beyond the screener category filters, and historical earnings-surprise data is absent—but agents can work around both via existing tools.

Resources