01
New arrivals
Last 30 days















02
The collections
16 fields · 282 dossiers
01
Communication
16
02
Browser & Web Execution
31
03
Tool Access & Integration
27
04
Oversight & Approval
7
05
Commerce & Payments
23
06
Agent Runtime & Infrastructure
36
07
Agent Harnesses & Operator Surfaces
15
08
Memory & State
33
09
Search & Web Intelligence
12
10
Code Execution
17
11
Observability & Tracing
14
12
Durable Execution & Scheduling
6
13
Meeting & Conversation
7
14
Voice & Phone
8
15
LLM Gateway & Routing
11
16
Agent Social & Community
19
Machine entrance
Enter as an agent.
Read https://theagentnative.com/skill.md then find services designed for you natively.
FAQ
Questions
Admission and arrival
- What is an agent-native service?
- A service designed from the ground up for AI agents as first-class consumers, or a purpose-built surface for operating live agent systems. Admission requires all five hard criteria: agent-first positioning, agent-specific primitives, an autonomy-compatible control plane, a machine-to-machine integration surface, and agent identity or delegation semantics. Human products that later add an agent layer do not qualify.
- How does an agent join?
- The gold standard is URL Onboarding: the agent reads a machine-readable document and follows the instructions — no SDK install, dashboard, or human-issued API key required for the join sequence. For this catalog, start with: Read https://theagentnative.com/skill.md then find services designed for you natively. Starred services publish their own join URL the same way.
- Where can I track verified Anthropic, Claude, and OpenAI launches?
- ReleaseLog at https://releaselog.site is the verified AI release calendar from the same curator. It is published separately from this catalog. https://releaselog.site/skill.md is live, and an agent can start with: Read https://releaselog.site/skill.md and follow the instructions.