Skip to main content
Glama

Sliq

by com.getsliq
OverviewDiscussionsAdmin

Exa Search News

exa_search_news

A find-companies News-mode discovery source, used alongside search_news_events and tavily_search — the News-mode skill owns the routing (which signals fan out here). Each query retrieves the top recent news articles matching it, and each article comes back as pre-extracted company rows (name, domain, one-line event), so unlike tavily_search there is no article-extraction step. Covers narrative signals PredictLeads' typed categories miss (AI initiatives, role-scoped exec changes, plant incidents) and fills its zero-result gaps on typed events.

Compose queries in positive form describing the event sought, e.g. "manufacturing companies announcing new plant openings in the United States" — one query per signal, all in one call (the tool fans them out in parallel; one tool call is one LLM round). Don't embed exclusions — vector retrieval has no concept of "not", so naming an excluded term biases results towards it.

Each query retrieves the 10 most relevant recent articles. Pass lookback_days to bound how far back to look (default 30); articles published before that window are excluded server-side. A news scan passes the window from its trigger prompt.

Costs 0.5 Sliq credits per query that returns without error, so a batch of 4 queries costs 2 credits; a failed query is free. Free on BYO Exa (a connected Exa key). Soft-fails on insufficient credits: the charge is capped at the remaining balance.

Each company is then checked against criteria, one company at a time, reading its name and its article's headline and event, and only the matches come back in companies; the rest come back in rejected as {name, reason}. It adds a few seconds per ten companies.

When this runs in an agent, every company with a domain is saved to the Output tab as a company row (deduped by domain) with its verdict, carrying its article in data.signals — except a company whose domain is the article's own host, which is often the outlet's domain misread as the company's. The rejected ones are hidden from the list's default view. Dict with companies, the matching companies; rejected, the companies the check turned down, as {name, reason}; count, every company found, rejected ones included; and created_identifiers, the domains of matching companies this call newly added to the list. Each company row is {company_name, domain, event, headquarters, source_url, title, published_at, query, verdict, verdict_reason} — verdict is 'qualified', 'unclear' when the row says too little to judge, or null when the check couldn't run; domain is lowercased-bare or None when the article didn't reveal one (resolve those downstream before recording), event is a one-line description of what happened, source_url/title/published_at identify the article, and query names the originating query. Rows are deduped by domain across queries (first article wins).

If every query failed with an upstream error and nothing landed, also exa_available: False plus a human-readable error — a source outage, NOT a genuinely empty feed. Partial failure (some queries errored, others returned) stays silent and reports only the companies.

On insufficient Sliq credits the call is refused before running — it raises InsufficientCreditsError instead.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
queriesYesOne positive-form news query per signal. Pass a list of one for a single-signal lookup.
agent_idNoOptional — a specific agent to save the companies into. Omit it to use the running agent, which is the usual case.
criteriaYesWhat makes a company a match for this search, in plain words: the event it must show and the kind of company (e.g. "A manufacturer that itself opened or is opening a new US plant"). Each company is checked against it, and the check reads nothing else of your instructions, so make it no narrower than what this call searches for. A role that is part of the signal belongs here; who you'll contact at a matching company doesn't, since the check reads that as a rule about which companies match.
list_nameNoShort kebab slug naming the Output-tab list bucket. Absent, companies land in the 'default' list.
on_behalf_ofNoAn agency client's email, from list_agency_clients, to run this call in that client's account: their agents, settings, connected accounts, and credits. Omit it to act on your own account.
lookback_daysNoDays of news history to scan (default 30).

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed1 schema field changed
    • addedInput schema / properties / on_behalf_of
      Added value: +{
+  "anyOf": [
+    {
+      "type": "string"
+    },
+    {
+      "type": "null"
+    }
+  ],
+  "default": null,
+  "description": "An agency client's email, from list_agency_clients, to run this call in that client's account: their agents, settings, connected accounts, and credits. Omit it to act on your own account."
+}
  2. Changed1 schema field changed
    • changedInput schema / properties / criteria / description
      Previous value: -"What makes a company a match for this search, in plain words:\nthe event it must show and the kind of company (e.g. \"A manufacturer\nthat itself announced a new US plant\"). Each company is checked\nagainst it, and the check reads nothing else of your instructions,\nso make it no narrower than what this call searches for. A role that\nis part of the signal belongs here; who you'll contact at a matching\ncompany doesn't, since the check reads that as a rule about which\ncompanies match."New value: +"What makes a company a match for this search, in plain words:\nthe event it must show and the kind of company (e.g. \"A manufacturer\nthat itself opened or is opening a new US plant\"). Each company is\nchecked against it, and the check reads nothing else of your\ninstructions, so make it no narrower than what this call searches\nfor. A role that is part of the signal belongs here; who you'll\ncontact at a matching company doesn't, since the check reads that as\na rule about which companies match."
  3. Changed2 schema fields changed
    • addedInput schema / properties / criteria
      Added value: +{
+  "description": "What makes a company a match for this search, in plain words:\nthe event it must show and the kind of company (e.g. \"A manufacturer\nthat itself announced a new US plant\"). Each company is checked\nagainst it, and the check reads nothing else of your instructions,\nso make it no narrower than what this call searches for. A role that\nis part of the signal belongs here; who you'll contact at a matching\ncompany doesn't, since the check reads that as a rule about which\ncompanies match.",
+  "minLength": 1,
+  "type": "string"
+}
    • changedInput schema / required
      Previous value: -[
-  "queries"
-]New value: +[
+  "queries",
+  "criteria"
+]
  4. Changed3 schema fields changed
    • addedInput schema / properties / agent_id
      Added value: +{
+  "anyOf": [
+    {
+      "type": "integer"
+    },
+    {
+      "type": "null"
+    }
+  ],
+  "default": null,
+  "description": "Optional — a specific agent to save the companies into.\nOmit it to use the running agent, which is the usual case."
+}
    • addedInput schema / properties / list_name
      Added value: +{
+  "anyOf": [
+    {
+      "type": "string"
+    },
+    {
+      "type": "null"
+    }
+  ],
+  "default": null,
+  "description": "Short kebab slug naming the Output-tab list bucket.\nAbsent, companies land in the 'default' list."
+}
    • addedInput schema / properties / lookback_days / description
      Added value: +"Days of news history to scan (default 30)."
  5. First observed

TDQS

A4.6/5.0
Behavior5/5

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

Goes well beyond the annotations: credit cost (0.5/query, failed queries free, BYO Exa free, soft-fail capped at remaining balance), the per-company criteria check and its latency, Output-tab persistence with domain dedup and the article-host exclusion, `exa_available: False` on total outage vs silent partial failure, and `InsufficientCreditsError`. This is unusually thorough behavioral disclosure.

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

Conciseness4/5

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

Long but organized into summary/returns sections and front-loaded with purpose, then query-composition rules, then cost, then the criteria check and persistence. Almost every sentence carries non-redundant information, though the returns block is dense enough that a little tightening would help.

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?

For a 6-parameter, side-effecting discovery tool with no output schema, the description covers the full contract: return keys, row shape and verdict values, dedup rules, failure/outage signaling, cost model, and the credit-refusal path. Nothing an agent needs to invoke or interpret the call is missing.

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

Parameters4/5

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

Schema coverage is 100%, so the baseline is 3, but the description adds real meaning: `queries` should be positive-form, one per signal, fanned out in a single call, and `criteria` must be no narrower than the search, with the explicit note that a contact role belongs in criteria only when it is part of the signal. `lookback_days` semantics (server-side window, default 30) are also restated with purpose.

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?

Opens with a specific verb+resource: 'Discover companies from recent news via Exa's semantic news search.' It explicitly positions itself against siblings `search_news_events` and `tavily_search`, and clarifies the key differentiator (pre-extracted company rows, no article-extraction step). An agent can tell what this does and how it differs without opening another schema.

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?

Names the alternatives and explains what this source adds ('covers narrative signals PredictLeads' typed categories miss', 'fills its zero-result gaps on typed events') plus concrete query-composition rules (positive form, one query per signal, no embedded exclusions). It stops short of an explicit 'use this when X, use the sibling when Y' decision rule because routing is delegated to the News-mode skill, so not a full 5.

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

All Sliq toolsInspect
Try in Browser
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources