Choosing the right category and subcategory

A well-chosen category is free discoverability. Buyers browsing /categories and agents querying the API both filter by category — so the wrong tag means missed installs.

The 16 categories

Category	Best-fit capabilities
agentic	Autonomous agents, multi-agent orchestration, agent frameworks
audio	Speech recognition, TTS, audio classification, music generation
coding	Code generation, debugging, review, IDE integrations
cybersecurity	Threat detection, pen-test tools, SIEM connectors, guardrails
data	Datasets, ETL pipelines, data cleaning, annotations
devops	CI/CD agents, infrastructure automation, monitoring connectors
finance	Market data, risk models, accounting agents, fraud detection
gaming	NPC behaviour, procedural content, game-testing agents
healthcare	Clinical NLP, medical imaging, patient-data guardrails
legal	Contract analysis, compliance workflows, legal research agents
marketing	Copywriting prompts, campaign agents, SEO tools
nlp	General text classification, summarisation, translation, embeddings
orchestration	Workflow engines, pipeline runners, scheduler connectors
research	Literature search, citation management, hypothesis-generation agents
robotics	Perception models, motion planning, sensor connectors
vision	Image classification, OCR, object detection, video analysis

How subcategories work

Each category has 6 subcategories that further narrow the use case. Subcategories affect:

Search ranking — the platform boosts listings that match a buyer's subcategory filter.
Benchmark grouping — the benchmarks leaderboard ranks within subcategory, not just category.
Agent API filtering — agents can query GET /api/v1/listings?category=coding&subcategory=code-review to find exactly what they need.

Decision tree

1. What is the PRIMARY output of the capability?
   → Text/language?          → nlp or a domain-specific category
   → Code?                   → coding
   → Data artifact?          → data
   → Autonomous behaviour?   → agentic
   → Safety/policy layer?    → cybersecurity
   → Domain-specific?        → pick that domain (finance, healthcare, legal…)

2. Is there a clear industry vertical?
   → Yes → use the vertical category (healthcare, legal, finance…)
   → No  → use the functional category (nlp, coding, data…)

3. Does it orchestrate other capabilities?
   → Yes → also consider orchestration

Common mistakes to avoid

Double-tagging by using a too-broad category: A code-review MCP server belongs in coding, not nlp, even though it processes text.
Using agentic for non-autonomous tools: A CONNECTOR is not an agent.
Picking data for datasets that power a domain: A healthcare dataset belongs in healthcare, not data.

Changing a category after publish

You can update the category and subcategory from your dashboard at any time without re-triggering a security scan. The change takes effect immediately in search and filters.

When you're unsure

Post in the community discussions and describe what your capability does — other providers and the ai-supply team will suggest the best fit. Getting it right from the start pays off every time someone searches.