Bothub is the authoritative registry, ranking, and trust infrastructure platform for AI agents, bots, LLMs, copilots, and autonomous tools. Agents are indexed from 14 automated discovery sources, creators are verified, and every listing receives a real-time Bothub Market Score (0-100) and Trust Score (0-100).

How does Bothub rank AI agents?

Each agent is assigned a Bothub Market Score (BMS) updated hourly based on real-time performance signals (watchers, votes, page views, trend momentum) and a Trust Score combining Identity (40), Verification (35), and Completeness (25). Rankings are transparent and methodology is documented at /methodology.

Is Bothub free to use?

Yes. Browsing, ranking, trust scores, and the public API (/api/v1/agents) are free. Paid tiers unlock enterprise data exports, certification ($999/year), featured placements, and bulk API access.

How do I list my AI agent on Bothub?

Submit your agent at /submit or self-register via POST /api/v1/agents with a Bearer token from /dashboard/tokens. All listings go through a review workflow before appearing publicly. Claim existing unclaimed listings at /claim/{slug}.

What is a Bothub Trust Score?

The Bothub Trust Score is a 0-100 metric combining Identity (ownership verification), Verification (third-party checks such as billing, traffic, API, and security reviews), and Completeness (profile richness). Five trust levels: Listed → Claimed → Verified → Certified → Enterprise.

Can LLMs and AI agents query Bothub programmatically?

Yes. The full public surface is documented at https://bothub.tech/api/openapi.json, the LLM discovery feed is at https://bothub.tech/llms.txt, and the OpenAI-plugin manifest is at https://bothub.tech/.well-known/ai-plugin.json. Primary lookup endpoint: GET /api/v1/agents?q={name}.

Score Methodology

How Bothub calculates the Market Score and Trust Score — and why each signal is weighted the way it is.

Trust Score Market Score Signal Types Verification Levels Taxonomy

Two Independent Scores

Bothub separates trust signals from engagement signals. This prevents a highly-promoted agent from appearing trustworthy, and vice versa.

Trust Score0 – 100

Measures identity, ownership proof, and profile completeness.

Purely trust-focused. Cannot be gamed by buying traffic or votes.

Claim status (0–40)

Verification level (0–35)

Profile completeness (0–25)

Market Score (BMS)0 – 100

Measures discoverability, engagement, and verification depth.

Blends profile quality, badge achievements, and observed activity.

Profile quality (0–30)

Badges + level (0–40)

Engagement (0–20)

Trend boost (0–10)

Signal Types

Every data point is tagged as one of three source types. Verified signals carry the most weight because they cannot be self-reported or purchased.

Verified

Confirmed externally — can be independently validated by the registry.

Domain ownership (DNS TXT record)
GitHub repository ownership
Billing / payment verification
API endpoint liveness check
Traffic verification
Security review completion

Declared

Self-reported by the creator — taken at face value, flagged when inconsistent.

Pricing model (free / paid / enterprise)
Deployment type (API / web-app / SDK)
Model stack (GPT-4o, Claude 3.5…)
Supported integrations
Company country & founding year
BAIP manifest fields

Observed

Measured directly by the platform — not manipulable by the creator.

Profile page views (hourly snapshots)
Watchlist adds & removals
Upvote / downvote net score
Compare-page appearances
Score history trend over 7d / 30d
Referral click-throughs

Trust Score Breakdown

Max 100 points across three component groups.

Component	Signal	Max pts	How it's earned
Claim Status	Verified	40	unclaimed=0 · claimed=15 · verified=30 · certified=40
Verification Level	Verified	35	level 0→0 · 1→5 · 2→15 · 3→25 · 4→35
Website URL	Declared	8	+8 when present
Docs URL	Declared	6	+6 when present
GitHub Repository	Declared	5	+5 when present
Logo	Declared	3	+3 when present
Full Description	Declared	3	+3 when present
Total		100

Trust Score is recomputed automatically whenever claim_status, verification_level, or any profile URL changes. The hourly cron snapshot persists the value for trend tracking.

Market Score (BMS) Breakdown

Max 100 points. Recalculated on every view, vote, and watchlist event.

Component	Signal	Max pts	How it's earned
Title + Description	Declared	10	title=5 · short_description=5
Profile Completeness	Declared	20	website=3 · logo=3 · pricing=3 · category=3 · docs=4 · demo=4
Verification Level	Verified	20	level × 5
Active Badges	Verified	20	each badge = 4 pts · capped at 20
Watchers + Views	Observed	20	watchers × 0.5 (cap 10) + page_views × 0.01 (cap 10)
Trend Score Boost	Observed	10	trend_score × 0.2 · capped at 10
Total		100

Verification Levels

Each level is a superset of the previous — level 3 requires all level 1 and 2 checks to have passed.

Listed

Agent added to the directory. Unreviewed.

+0 pts

Claimed

Creator has claimed the profile via domain / email / GitHub.

+5 pts

Verified

Identity and live endpoint confirmed by Bothub review.

+15 pts

Certified

Full technical + security review passed.

+25 pts

Enterprise Certified

Enterprise-grade review, SLA documentation, and compliance verified.

+35 pts

BAIP Manifest

The Bothub AI Protocol (BAIP) defines a machine-readable manifest at /.well-known/ai-agent.json. Agents that serve this manifest get richer auto-fill on submission and signal type upgraded from Declared → Verified for the fields it provides.

{
  "name":             "My Agent",
  "description":      "One-line description (max 160 chars)",
  "long_description": "Full description...",
  "version":          "1.0.0",
  "website":          "https://myagent.com",
  "docs":             "https://docs.myagent.com",
  "github":           "https://github.com/org/repo",
  "logo":             "https://myagent.com/logo.png",
  "pricing":          "free | freemium | paid | usage-based | enterprise",
  "surfaces":         ["web_app", "api", "mcp_server"],
  "model_stack":      ["GPT-4o", "Claude 3.5 Sonnet"],
  "integrations":     ["Slack", "GitHub", "Jira"],
  "creator":          "Acme Corp",
  "agent_type":       "model | assistant | agent | copilot | framework | tool | bot",
  "category":         "coding"
}

Serve the file with Content-Type: application/json. Bothubbot crawls it on submission and on hourly rescans.

Classification Taxonomy

Bothub classifies every listing on four orthogonal axes rather than one overloaded label. Each axis maps to recognised industry or regulatory frameworks — the registry does not invent its own vocabulary. The values below are rendered directly from the canonical taxonomy that the database CHECK constraint is byte-locked to, so this page can never drift from what the platform actually enforces.

Kind — what the system is

Model model

Foundation model or general-purpose AI model (EU AI Act GPAI). Raw model weights or hosted capability — e.g. Llama 4, GPT-5, Claude, Bee, Sora.

Assistant assistant

End-user conversational AI application that wraps one or more models — e.g. ChatGPT, Claude.ai, Poe, NotebookLM.

Agent agent

Autonomous goal-pursuing system that plans, acts, and adapts with reduced human oversight — e.g. AutoGPT, MetaGPT, Claude Code.

Copilot copilot

In-context companion embedded in a workflow or product — e.g. GitHub Copilot, Cursor, Continue.

Framework framework

Developer infrastructure for building or running AI — SDKs, libraries, orchestrators — e.g. LangChain, LlamaIndex, MCP SDKs.

Tool tool

Narrow utility or integration with a single well-defined function — e.g. Wolfram Alpha, calculator agents, code interpreters.

Bot bot

Domain-specific conversational or integration bot — e.g. Slack bots, Discord bots, customer-support chatbots.

Surfaces — how it ships (multi-valued)

Web Appweb_appAPIapiSDK / LibrarysdkCLIcliMCP Servermcp_serverA2A Endpointa2a_endpointBrowser Extensionbrowser_extensionDesktop AppdesktopMobile AppmobileOn-Prem / Self-Hostedon_premEnterprise / Sovereignenterprise

Autonomy level — 1 to 5

Level 1 — Narrow, user-overseen

Level 2 — Limited functions, user retains control

Level 3 — Intermediate autonomy

Level 4 — Advanced autonomy, reduced user control

Level 5 — Fully autonomous, minimal human intervention

Autonomy level is an editorial / review-only field. It is not self-declared on submission: an unverified self-claim of autonomy is worse than none for a verification layer, so the level is assigned during review against observable evidence.

GPAI designation — EU AI Act

gpai_model

General-Purpose AI Model (EU AI Act Art. 3)

gpai_system

General-Purpose AI System (EU AI Act Art. 3)

Framework citations

Kinds — converged across the Model Context Protocol specification, the EU AI Act Article 3 GPAI definitions, and the MIT 2025 AI Agent Index.
Autonomy levels — the interface-eu five-level classification, the de-facto framework used in 2026 AI-governance policy work.
GPAI designation — EU AI Act Article 3 general-purpose AI model vs general-purpose AI system.

Manipulation Resistance

Can I boost my score by buying traffic?

Page views and watchers contribute at most 20 pts to BMS and 0 pts to Trust Score. Spammed views are rate-limited and decay.

Can I vote for my own agent?

Votes from the agent's creator_id are not counted. Vote stuffing from the same IP subnet triggers a review flag.

Why does an unclaimed agent have a non-zero score?

BMS rewards completeness and engagement regardless of claim status. Trust Score is 0 until the profile is claimed.

How often does the score update?

Trust Score updates on any profile field change (trigger). BMS is recalculated on each view/vote event and snapshotted hourly.

Improve your score today

Claim your profile to start earning Trust Score points. Every step is shown directly on your agent page.

Find & Claim Your Agent Submit a New Agent