Niches · Agent tool access

Find the right action among thousands without filling context

No company's docs show a full fit yet. The listings below do the job with a limit, or not at all.

0companies, full fit
9companies, fit with a limit
33companies carry the tag
16share of the category's work
6first-hand accounts

What counts as done

One search call returns a short list (under 10) of tools the connected account can run, with schemas.

Measured, for our own tests: the list costs under 2,000 tokens of context. No docs we read state these figures, so they don't count against a listing here.

How agents ask for it: what can you do in Notion, find the tool to list Zoho tickets, which apps can you reach

Does the job, with a limit 9

It does the job, but something in the docs or a user's account will get in an agent's way. The limit is named in each row.

ListingWhy
AgentWeb D
Agent Web Labs, Inc.
find_actions takes the task in plain words, but the docs give no result count or token size, and the 22-tool manifest is about 47 KB of JSON at rest. source
Composio (API + MCP) A
Sampark Inc (d/b/a Composio)
Sessions expose search and execute meta tools over every toolkit by default, filterable by toolkit, tool or tag. acc-25 shows search offering the whole catalogue so the agent picks operations the account can't run; no token figure is published. source
MCP Gateway BB
No legal entity found. COMMERCIAL.md, CLA.md and AUTHORSHIP.md name Mikko Parkkola, Finland, as licensor and author
gateway_search_tools searches only the configured backends behind 11 meta-tools of about 1,100 tokens. The README's own run found 1.2 to 16.1 per cent more input tokens and one more turn than direct, and search result size isn't stated. source
Merge Agent Handler BB
Merge API, Inc.
search_tools takes a plain-language intent across a Tool Pack that can hold hundreds of tools, but no result count or token cost is stated, and authenticate tools don't appear in search. source
Monid C
Monid, Inc.
POST /v1/discover/endpoints searches 2,800+ endpoints in plain words and inspect returns one endpoint's schema and price, so the catalogue stays out of context. Schemas need a second call and result size and token cost aren't stated. source
One B
One Systems, Inc.
find_one_actions takes up to 10 platform-and-intent requests and returns the chosen action with docs and a few alternatives; actions outside the grant are never offered. The 3-tool surface is about 1.2K tokens, but per-search size isn't stated beyond digests. source
StackOne B
StackOne Technologies Limited
Advanced Tool Search replaces the catalogue with a search tool and an execute tool and scales to thousands of actions, but result count and token cost per search aren't stated. source
Unified.to MCP Server B
Unified API Inc.
Tools are limited to what the connection was granted, and tools, permissions and defer_tools narrow or defer the list, but there is no search tool; the docs warn many models cap at 10 to 50 tools. source
Zapier MCP (agent actions) C
Zapier Inc.
Agentic mode keeps 16 meta-tools at rest and discover_zapier_actions searches apps and actions, auto-provisioned from existing connections. Result size and token cost aren't stated, and discovering then enabling adds a call. source

Doesn't do this job 2

Listings in the category that do something else, with the reason.

ListingWhy
letme BB ours
Long Lunch Labs Limited (company number 17523112, registered in England and Wales, incorporated 15 October 2026), trading as Anchor Terminal and letme
One call returns a pick and up to 10 alternatives from Anchor Terminal's directory of listings, with connect snippets, but not actions a connected account can run or their input schemas. source
Smithery F
Clavia, Inc. (doing business as Smithery)
What remains is a public directory of MCP servers for people to browse, with no API for an agent to search actions in its connected accounts. source

1 listings in the category were added after this reading and aren't marked yet.

What stops agents finishing

  • search has operations the account cannot run
  • agent forgets deferred tools exist

Around it

After: aggregator.run-app-action.

Where this comes from

Weight 16 of 100 in Agent tool access: 6 of 32 accounts; context cost of tool lists is the most discussed theme on Hacker News

The same as data: /api/v1/use-cases/aggregator.find-tool.json, or the MCP tool get_use_case with id aggregator.find-tool. Other jobs in this category: /api/v1/niches?category=aggregator.

Draft, read 2026-11-15 from public docs and first-hand accounts, not yet reviewed by a person. A mark says what the docs show, not what a run measured.

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.