ujjansh05/archrivet
v0.4.0-alpha.2MIT
Review explicit backend design facts, identify missing details, calculate workload demand, and display a local architecture diagram.
commands
- python.exe
plugin.json
- name
- archrivet
- version
- 0.4.0-alpha.2
- description
- Review explicit backend design facts, identify missing details, calculate workload demand, and display a local architecture diagram.
- author
- Ujjansh05https://github.com/Ujjansh05
- license
- MIT
- repository
- https://github.com/Ujjansh05/archrivet
- extensions
- com.openai
mcp.json
- archrivetstdio${PLUGIN_ROOT}/runtime/python.exe -I -B ${PLUGIN_ROOT}/launch.py
{
"$schema": "https://agent-plugins.org/schemas/1.0.0/plugin.schema.json",
"name": "archrivet",
"version": "0.4.0-alpha.2",
"description": "Review explicit backend design facts, identify missing details, calculate workload demand, and display a local architecture diagram.",
"author": {
"name": "Ujjansh05",
"url": "https://github.com/Ujjansh05"
},
"homepage": "https://github.com/Ujjansh05/archrivet",
"repository": "https://github.com/Ujjansh05/archrivet",
"license": "MIT",
"keywords": [
"system-design",
"architecture",
"local",
"developer-tools"
],
"extensions": {
"com.openai": {
"interface": {
"displayName": "ArchRivet",
"shortDescription": "Review backend design facts",
"longDescription": "ArchRivet helps your assistant review backend designs. It checks explicitly supplied facts, names missing details, and compares results after you clarify a design. Eight scoped checks cover retries, acknowledgements, queue overload, transactions, replicas, caches, failover, and dependency deadlines. Workload calculations show formulas and assumptions; capacity stays unknown without suitable benchmark evidence. An optional local browser window displays your assistant's diagram and updates when the assistant calls the diagram tools. This Windows x64 submission candidate includes the Python runtime and rules engine. No account, model download, GPU, or separate review daemon is needed. It does not certify a complete architecture, automatically read a conversation, or provide a native ChatGPT side panel. Laya training checkpoints are separate experimental work and are not included in this package.",
"developerName": "Ujjansh05",
"category": "Developer Tools",
"capabilities": [
"Review design facts",
"Ask for missing details",
"Calculate workload",
"Show local diagrams"
],
"websiteURL": "https://github.com/Ujjansh05/archrivet",
"supportURL": "https://github.com/Ujjansh05/archrivet/issues",
"privacyPolicyURL": "https://github.com/Ujjansh05/archrivet/blob/main/PRIVACY.md",
"termsOfServiceURL": "https://github.com/Ujjansh05/archrivet/blob/main/TERMS.md",
"defaultPrompt": [
"Review my backend design and tell me which details are missing.",
"Calculate workload demand using my assumptions and show what still needs testing.",
"Open a diagram of my design and update it as we change the architecture."
],
"brandColor": "#166B63",
"brandColorDark": "#5EEAD4",
"composerIcon": "./assets/icon.png",
"logo": "./assets/icon.png",
"screenshots": [
"./assets/screenshot.png"
]
},
"review": {
"commerce": false,
"commerce_description": "This local open-source package does not sell products or process payments."
},
"publication": {
"release_notes": "Windows local submission candidate with eight explicit-fact rules, missing-detail questions, before/after comparisons, workload formulas and a browser diagram companion. Includes runtime dependencies. Experimental Laya checkpoints are excluded. Host installation, identity, allowlisting and public submission materials still need verification."
}
}
}
}
Client extensions
Data this package carries for particular clients. The directory lists the clients named and never reads what is addressed to them.
- com.openai