A web harness with no settings panel: the model configures the page through a Cisco IOS style CLI. One HTML file, plus an optional C++ MCP server per machine for shells and files on Linux, macOS and W
View on GitHub ↗A web harness with no settings panel. You talk; the model sets up the page.
One HTML file, no account, plus an optional C++ server per machine. Give it one endpoint, yourself or through the kiss.conf of the site, and the model bootstraps the rest: other endpoints, MCP servers, tools, style, typed by the model itself in a CLI modeled on Cisco IOS. Anything that opens a way out of the page waits for your yes, and a key goes from your keyboard to the page, never through the model.
KiSS speaks the OpenAI compatible chat completions for now, plus the Anthropic Messages API and the OpenAI Responses API in their own protocols. Type these lines after a /, or just ask the model. A key left out opens a masked field: it never enters the conversation.
llama-server, or any OpenAI compatible server:
Then show models, set chat model claude/, and copy run start to keep it.
Any unambiguous prefix works. A batch applies whole or not at all and answers with the exact change, - then +. Endpoints, models, sampling, system prompt, MCP servers, tools, display and style all go through it, for the model as for you.
- A way out asks every time: a new endpoint or MCP address, a tool opened. No answer turns the question off, and a key never follows a URL the model wrote. - The firewall reads the effect, not the command: whatever spells a change, the diff decides. - Every MCP tool starts in consent, each call shown with its real arguments.
From the project README.
Add the radar badge to your README — it shows your project was picked up by MCP Radar and links to this page:
[](https://mcp.liqiwa.com/s/ServeurpersoCom--kiss.html)
ChatGPT thinks. Codex works. Use ChatGPT as the planning brain while keeping the Codex harness.
shinthink/blitzstrike⚡ Blitz Strike — a universal MCP penetration-testing toolbelt. Structured methodology: reconnaissance & attack-surface mapping, source-to-sink analysis, and live validation. 57 escalation chains, 130-
0xwilliamortiz/openclaude-improvedruns anywhere. uses anything
halofyai/halofyHalofy is the open access and governance layer for AI agents across your organization. Identity, policy, provenance, audit, and signed erasure.
CaptureGrubEnchant/SolidWorksSolidWorks MCP Server connects an AI assistant to a running SolidWorks instance. Sketch, extrude, fillet, export STEP/STL, generate macros.
arielshad/3d-asset-server3d assets api and mcp
The top new MCP servers of the week, every Monday. No spam, unsubscribe anytime.