FOX LAB
Artificial intelligence in practice. News, experiments, useful techniques and guides.
The BFDO Fox Score rating system
We rate materials on a five-fox scale in half-fox steps. A score of 4.5 means: “Four foxes and a tail!”, and 5 means:“Five foxes out of five!”. Materials without lab testing are clearly marked “Not rated yet”.
FOX LAB is preparing its first publications
The BlackFox Development editorial team is building an independent knowledge base of architecture guides and practical AI workflows. Materials will be published after real testing and editorial verification.
Architecture & Prompts
System instructions, zero-pill design and interaction contracts for LLMs.
Context optimization
Log compression, token isolation and faster work in long conversations.
Skills library
Open Skills for AI Studio, Claude Code and terminal agents.
Lab tests
Honest tests under the BFDO Fox Score protocol, with the conditions recorded.
Found: 0 articles; 3 Skills in preparation
Skills library (Public Skills)
Specifications and extension modules for AI agents. Available for a first look.
BFDO Architecture & Context Discipline
A planned template for reviewing project structure, separating layers, and checking dependencies and types before an agent makes changes.
BFDO Token & Attention Optimizer
Planned rules for managing an agent’s context window, compressing logs dynamically and removing irrelevant tokens in long sessions.
BFDO Terminal Sandbox Guard
A planned profile for controlling command execution in an isolated environment: network namespaces, protection against accidental deletion and monitoring file descriptors.