Ecosystem layer management repository for the Airymax AI Agent Runtime Platform. One of five management repositories under the airymaxhub umbrella.
Language: English | 简体中文
The ecosystem management repository is the application-facing ecosystem layer of the Airymax platform. It aggregates 5 leaf repositories as git submodules and provides the upper-layer ecosystem support that Agent applications consume at runtime — configuration management, prompt templates, reference examples, an open laboratory, and official skill definitions.
Within the Airymax four-layer architecture (Applications → Ecosystem → Daemon Services → Atoms), the ecosystem layer sits between end-user applications and the runtime daemons. It is the layer where Agent developers spend most of their time: they write agents against the SDK, then register them with manager/, compose prompts from prompts/, learn from examples/, extend the platform via openlab/, and reuse official capabilities from skills/.
This management repo only carries documentation, submodule wiring, and licensing. All implementation lives in the leaf repositories.
ecosystem/ # Management repository (this repo)
├── manager/ # Ecosystem manager leaf repo (submodule)
├── prompts/ # Prompt library leaf repo (submodule)
├── examples/ # Example agents leaf repo (submodule)
├── openlab/ # Open lab leaf repo (submodule)
├── skills/ # Official skills leaf repo (submodule)
├── .gitmodules # Submodule definitions
├── LICENSE # AGPL-3.0 + Apache-2.0 dual license full text
├── NOTICE # Copyright, trademark and third-party notices
├── README.md # This file (English)
└── README_zh.md # Chinese translation
| Module | Directory | Repository URL | Description |
|---|---|---|---|
| manager | manager/ |
git@atomgit.com:openairymax/manager.git |
Unified configuration & lifecycle management center — 11 JSON Schemas (~272 rules), 10 registered skills, 12 registered agents, 3 environment overlays (dev/staging/prod), sanitizer suppressions, security policies, deployment templates |
| prompts | prompts/ |
git@atomgit.com:openairymax/prompts.git |
Official prompt template library — 14 templates across 4 categories (Cognition / Memory / Security / System), registry, tuner framework (scorer / evaluator / A-B testing) |
| examples | examples/ |
git@atomgit.com:openairymax/examples.git |
10 runnable example agents with a graded learning path (Beginner → Intermediate → Advanced) — hello-agent, weather-agent, code-review-agent, research-agent, multi-agent-debate, etc. |
| openlab | openlab/ |
git@atomgit.com:openairymax/openlab.git |
Open laboratory — multi-agent orchestration core, 4 reference applications (DocGen / E-Commerce / Research / VideoEdit), 7 role agents, community contributions, JSON-RPC 2.0 marketplace |
| skills | skills/ |
git@atomgit.com:openairymax/skills.git |
5 official skills (code_review / text_summarization / security_audit / data_analysis / web_search) built on the SkillPlugin base class from sdk-python |
The ecosystem layer is the second layer of the Airymax four-layer architecture. It consumes runtime capabilities exposed by the Daemon Services layer (via the SDK) and provides the building blocks that Agent applications assemble:
┌──────────────────────────────────────────────────────────────────┐
│ Applications │
│ End-user Agent apps built on ecosystem + SDK │
├──────────────────────────────────────────────────────────────────┤
│ ★ Ecosystem (this management repo) ★ │
│ manager · prompts · examples · openlab · skills │
├──────────────────────────────────────────────────────────────────┤
│ Daemon Services │
│ 12 runtime daemons (gateway_d / llm_d / tool_d / sched_d / ...) │
├──────────────────────────────────────────────────────────────────┤
│ Atoms │
│ Micro-core primitives (corekern / coreloopthree / syscall / ...)│
└──────────────────────────────────────────────────────────────────┘
Each leaf repository plays a distinct role within the ecosystem layer:
| Leaf Repo | Role | Key Artifact |
|---|---|---|
| manager | Configuration root — single source of truth | configs/agentrt.yaml (v0.1.1 unified runtime config) |
| prompts | Prompt engineering — templates + evaluation | registry.yaml + tuner framework |
| examples | Learning path — runnable reference agents | 10 example agents with graded difficulty |
| openlab | Innovation surface — experimental & community | 4 reference applications + marketplace |
| skills | Reusable capabilities — official skill pack | 5 SkillPlugin subclasses |
- SDK — connects to a running AgentRT instance through the language SDKs (
sdk/management repo); the SDK exposes the double-layer API (Cognition / Safety / Tool / Chat nested clients) - Runtime —
manager/configs/agentrt.yamlis consumed by the AgentRT runtime at startup;manager/sanitizer/suppression files are consumed by the build toolchain at build/test time - Cupolas —
manager/sanitizer/andmanager/security/are co-owned with the Cupolas security module under a dual-responsibility model
- Agent developers — use
prompts/andskills/as building blocks; learn fromexamples/; deploy viamanager/configs - Operators — use
manager/deployment templates and monitoring configs for production rollouts - Researchers — use
openlab/for experimental features and community contributions - CI / CD pipelines — run
manager/tools/drift_detector.pyandmanager/tools/config_diff.pyas configuration validation gates
Note: The official Hooks collection (formerly
ecosystem/hooks/) was migrated tosdk-python/agentrt/hooks/during SP09.3. Import path changed tofrom agentrt.hooks import ....
The ecosystem layer ships Python tooling and configuration rather than compiled artifacts. Typical usage:
# Clone with submodules
git clone --recurse-submodules git@atomgit.com:openairymax/ecosystem.git
cd ecosystem
git submodule update --remote --checkout
# Validate a configuration file against its schema (manager/)
python -c "
import json, yaml
from jsonschema import validate
schema = json.load(open('manager/schema/kernel-settings.schema.json'))
config = yaml.safe_load(open('manager/kernel/settings.yaml'))
validate(instance=config, schema=schema)
"
# Run the operations toolset (manager/)
python manager/tools/src/drift_detector.py --action both --output drift_report.json
# Run example agents (examples/)
cd examples/hello-agent && python main.py
# Run skill tests (skills/)
python -m pytest skills/tests/ -v- This management repo —
mainonly. No feature branches are created here. - Leaf repositories — active development happens on
feature/official-hubs-01. Themainbranch on each leaf repo tracks the last stable release.
When cloning this repo with submodules:
git clone --recurse-submodules git@atomgit.com:openairymax/ecosystem.git
cd ecosystem
git submodule update --remote --checkoutDual-licensed under AGPL v3 + Apache 2.0 (SPDX: AGPL-3.0-or-later OR Apache-2.0). You may choose either license at your option. See LICENSE for the full text of both licenses and NOTICE for copyright, trademark and third-party notices.
Copyright (c) 2025-2026 SPHARX Ltd. All Rights Reserved.