ooda
Here are 18 public repositories matching this topic...
Local-first agentic NetOps framework that connects LLMs to real network devices with unified intents, adapters, and full observability.
-
Updated
Mar 24, 2026 - Shell
Multi-modal sharp thinker — FeynTech + Drama modes for Claude, Perplexity, ChatGPT, Manus
-
Updated
Jun 26, 2026 - Shell
The WITNESS of The Wire — a witnessed substrate for agents. OODA-Observe across every browser tab and host; records every act so agents stay continuous, auditable, and federated. Recommends, never acts. (Record/replay test automation is one application, not the definition.)
-
Updated
Nov 1, 2026 - Go
HYDRA-C2: Multi-Domain Operations C2 Framework — 25 actors, 18 API endpoints, MIL-STD-2525B COP dashboard, OODA/kill-web doctrine, ADS-B/AIS/SDR signal intelligence
-
Updated
Mar 10, 2026 - Python
TravelExpress Web Application built with DotNetCore2
-
Updated
Jan 17, 2019 - CSS
OODA loop router for Claude Code & Grok — Observe, Orient (schwerpunkt), Decide, Act. Not a 4-box speed circle.
-
Updated
Sep 7, 2026 - Python
Turn OODA into an execution protocol for AI agents: purpose-first, evidence-based, verified, iterative. Part of a 5-repo agent-skill ecosystem (sloth-loop / architect-skills / obsidian-skills / ai-skill-output-template).
-
Updated
Oct 29, 2026 - JavaScript
Agent skills for popular work processes: GTD, OKR, PDSA, DMAIC, Design Thinking, Lean Startup, OODA, and research paper writing.
-
Updated
Aug 13, 2026 - Python
An application to improve the efficiency of rescue work after a flood, by estimating the locations of people in a stricken area and automatically allocating resources to rescue them.
-
Updated
Jun 5, 2018 - Python
ESRA for Claude Code: bounded, value-aligned skills and a privacy-preserving runtime for auditable improvement cycles.
-
Updated
Sep 24, 2026 - Python
Self-Improvement Framework for Supply Chain Agents
-
Updated
Oct 25, 2026 - Python
Portable, auditable Combat Hunter observation and decision skill for AI agents
-
Updated
Sep 29, 2026 - Python
Safety-first, model-agnostic agent framework. Hard constraints are code, not prompts. Works with OpenAI, Anthropic, Ollama, HuggingFace, and LayerCake.
-
Updated
May 21, 2026 - Python
Discover hidden correlations using EVEZ OODA loop + poly_c scoring. — by Steven Crawford-Maggard (EVEZ)
-
Updated
Jun 22, 2026 - Python
Add this topic to your repo
To associate your repository with the ooda topic, visit your repo's landing page and select "manage topics."