Agent Nexus
Server Details
Agent-native registry to discover APIs, MCP servers and CLIs, with live health checks.
- Status
- Healthy
- Uptime
- 100.0% over the last 24 hours
- Last Tested
- Transport
- Streamable HTTP · MCP 2025-06-18
- URL
- Repository
- Strategiumfinancepartner/AgentNexus
- GitHub Stars
- 1
- Server Listing
- Agent Nexus
- ChatGPT
- –
- Claude
- –
TDQS
Scored across 6 tools
Each tool owns a distinct access pattern: goal-based discovery (discover_capabilities), literal keyword lookup (search_registry), full-catalogue enumeration (list_entries), single-slug fetch (get_entry), taxonomy enumeration (list_categories), and write (submit_entry). Descriptions explicitly disambiguate against each other ('if you do not have a slug, do not guess one — use search_registry'), leaving no realistic misselection path.
All six names follow a strict verb_noun snake_case pattern (discover_capabilities, get_entry, list_categories, list_entries, search_registry, submit_entry). Verbs are also semantically consistent — list_* for enumeration, get_* for single-record reads.
Six tools is well-scoped for a registry discovery/lookup service: one entry per distinct read pattern plus a single write. Nothing appears redundant or bolted on, and nothing obviously needed is squeezed into an overloaded tool.
The read surface is thorough (taxonomy, paging, keyword, goal, single-record), but the write side is a dead end: submit_entry creates a pending row with no tool to list, poll, update, or withdraw submissions — the description points to list_my_submissions and an HTTP polling endpoint that are not exposed as tools. Agents must fall back to raw HTTP calls to close the submission lifecycle.
Publisher details
- Operator
- Agent Nexus
- Operator website
- https://agentnexus.app
- Vendor relationship
- Not applicable
- Documentation
- https://agentnexus.app/llms.txt
- Privacy policy
- Not provided
- Trust center
- Not applicable
- Restrictions
- Read access is free and anonymous. Submissions require a free API key (POST /api/public/keys)
Available Tools
6 toolsdiscover_capabilitiesFind an API, MCP server or CLI for a taskRead-onlyIdempotentInspect
Find APIs, MCP servers or CLIs that can accomplish a specific task. Use this when the user describes a goal or capability rather than an already-known service. Call this first when a task requires an external service you do not already know, when you need an API for email, payments, search, storage, data, LLMs or another capability, or when an existing service is unavailable and you need an alternative. Describe the actual task in plain language (e.g. 'send transactional emails', 'find an alternative geocoding API', 'store and retrieve files'). Returns matches[] with slug, endpoint, auth, formats, rate limit, pricing and a 0-100 reliability score. Review the matches and select a relevant candidate before calling get_entry to inspect its full record. If coverage='none', do not invent a match: use the returned starting points or explain that no suitable service was found. For a known product, vendor, slug or endpoint, use search_registry instead. Agent Nexus only discovers and evaluates interfaces; you invoke the selected service directly. Rate limits: 100 calls/day anonymous, 1,000/day with a free key, 50,000/day on Agent Pro.
Inputs: need (required), limit, category, min_reliability. Returns: need, note, count, matches, +2 more
get_entryGet registry entryRead-onlyIdempotentInspect
Retrieve the full record of one approved service identified by its exact slug. Use this after discover_capabilities or search_registry identifies a potentially relevant service, or when you already know the exact slug. Inspect the record before recommending or relying on a service: it includes endpoint, authentication mode and parameters, input/output formats, rate limits, pricing, capabilities, documentation URL, invocation example and the latest recorded health probe (hourly for priority entries, otherwise on a rolling daily cycle). Use these details to decide whether the service actually fits the task and whether you can call it with the available credentials. Read-only and idempotent. Takes one exact slug and no filters or pagination. Do not guess slugs; use search_registry or discover_capabilities if you do not have one. An unknown or unapproved slug is an error, not an empty result.
Inputs: slug (required). Returns: entry
list_categoriesList interface categoriesRead-onlyIdempotentInspect
Retrieve the Agent Nexus catalogue categories and their current counts of approved entries. Use this when you need to understand the taxonomy, learn the accepted 'category' values or prepare a catalogue exploration. The three interface layers are api, mcp and cli. Call once when category information is needed; this tool does not return individual services. Use discover_capabilities to find a service for a task, search_registry for a known keyword, or list_entries to enumerate entries. Read-only, idempotent and parameterless.
Inputs: no named parameters. Returns: categories
list_entriesList all registry entries (paginated)Read-onlyIdempotentInspect
Enumerate the approved Agent Nexus catalogue page by page. Use this for catalogue exploration, auditing, mirroring or building your own service index; do not use it as the default search method when solving a specific task. For a specific service or keyword, use search_registry; for a plain-language capability or task, use discover_capabilities. Returns {total, page, per_page, pages, results[]} in deterministic slug order. Read 'pages' from the first response and increment 'page' until all pages have been retrieved; keep the category filter identical across pages. Results are summary records; use get_entry for the full details of one service. For a complete bulk export, use GET https://agentnexus.app/api/public/entries.ndjson. Read-only and idempotent.
Inputs: page, category, per_page. Returns: page, pages, total, results, +1 more
search_registryLook up a service by nameRead-onlyIdempotentInspect
Search the Agent Nexus registry when you already know the service, product, vendor, slug or endpoint you are looking for. Use this for a targeted lookup, not for discovering a solution to a broad task; for a plain-language goal, use discover_capabilities instead. Matches literal text only and does not interpret intent. Optionally narrow the search by category. Returns {count, results[]} containing summary records with slug, name, category, summary, endpoint and trust fields. Results are alphabetical, not ranked by relevance. There is no pagination beyond the limit parameter: increase limit (maximum 50) or narrow the query when needed. Use get_entry to inspect the full record of a relevant result. An empty results[] is a valid answer, not an error. Read-only and side-effect free.
Inputs: limit, query, category. Returns: count, results
submit_entrySubmit an interface to the registryInspect
Submit a new API, MCP server or CLI to the Agent Nexus catalogue. Use only when the user explicitly requests a submission or wants to contribute a relevant external service that is not already listed. Before submitting, use search_registry to check whether the service already exists. A submission creates a pending entry; it is NOT published automatically, and every submission is reviewed by a human before becoming discoverable. Required fields: name, category, summary, endpoint and auth_mode. Provide concrete capabilities[], docs_url and invocation_example whenever available: they help reviewers evaluate the submission and help other agents discover the service. Authentication is required, either as a signed-in member or with a free agent key from POST https://agentnexus.app/api/public/keys. Limits: one submission per agent key and per source address per 24 hours; 20 per hour for members. Nothing is overwritten or deleted; re-submitting an existing name creates another pending row. The endpoint must answer a live health probe to keep a reliability score. Returns {slug, status}; check the decision with GET https://agentnexus.app/api/public/submission?slug=, or provide contact_email for email notification. Use list_my_submissions to review previous submissions.
Inputs: name (required), summary (required), category (required), endpoint (required), +12 more. Returns: slug, status
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Related MCP Connectors
Discovery registry for AI agents: 700+ APIs, MCP servers and CLIs, health-checked with real requests (hourly for priority entries, otherwise on a rolling daily cycle) — full initialize + tools/list handshakes for MCP, real calls for APIs — plus response-schema validation that catches responses which parse fine but are missing a required field.
AI agent registry — search, discover, register, and connect agents via MCP.
Registry of MCP servers, agent skills and plugins: search, filter, comments, likes, publish.
Agent Resource Discovery (ARD, IETF draft) registry: search every ARD registry + verified MCP tools
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceProvides a unified registry for discovering and managing agents that implement the A2A (Agent-to-Agent) protocol, enabling registration, querying, and CRUD operations on agent metadata through both REST API and MCP tools.3 npm4MIT
- AlicenseNot gradedqualityBmaintenanceProvides a shared registry that lets MCP clients search remote/stdio MCP servers and A2A agents and retrieve their current capabilities, transports, addresses, and credential references. Consumers use the returned connection details to call the selected targets directly from their own environment.Apache 2.0
- AlicenseNot gradedqualityCmaintenanceProvides MCP clients with a searchable registry of MCP servers, enabling discovery and publishing of servers.47 npmMIT
- AlicenseNot gradedqualityDmaintenanceAn open source registry and proxy that federates MCP, A2A, and REST/gRPC APIs with centralized governance, discovery, and observability. It optimizes agent and tool calling, and supports plugins.Apache 2.0
Glama MCP Gateway
Add one secure layer between your agents and this server.