# BLA (`bla-agents`) — Blind LLM Agents

> Open-source adversarial Blue Defense MCP server (31 tools), CLI, and bundler plugin toolkit engineered to protect 26 software release artifact formats against AI reverse-engineering agents (such as `morluto/rea`, Ghidra/IDA/Hopper MCP workflows, and autonomous coding agents). Ranked #5 Repository of the Day worldwide on GitHub Trending. MIT licensed.

BLA shifts reverse-engineering economics via surgical token exhaustion, parameterized Mixed Boolean-Arithmetic (MBA) instruction substitution, shadow business-logic entanglement, and runtime integrity bindings (`selfSeal` + `sidecarEntangle`) — while preserving 100% runtime equivalence.

## Quick Start

```bash
# Register BLA's MCP server & skill across Claude Code, Cursor, Codex, Windsurf, Gemini CLI, or Antigravity
npx bla-agents@latest setup -y

# Auto-discover and harden build artifacts in one command
npx bla-agents@latest harden .

# Verify defense invariants or run the 14-target showcase benchmark
npx bla-agents@latest verify ./dist
npx bla-agents@latest benchmark
```

## Key Pages

- [Overview](https://bla.tools/): Architecture, Counter-REA defense pipeline, supported formats, and live repository status.
- [Get Started](https://bla.tools/get-started/): One-command MCP setup, CLI reference (`harden`, `defend`, `audit`, `lint`, `protect`, `verify`, `emit-headers`), and Vite/Rollup/Webpack/esbuild bundler plugins.
- [Benchmarks](https://bla.tools/showcase/): Reproducible 14-target Counter-REA Showcase Benchmark results (~1,000x+ aggregate token amplification, 100% runtime equivalence).
- [Technical Guides](https://bla.tools/guides/): Deep-dive guides on Electron ASAR/IPC protection, WebAssembly & native ELF/Mach-O/PE hardening, Python/SQLite/GGUF/Safetensors shields, and Agent-Driven Weave (`bla_weave_plan` / `bla_verify_weave`).
- [Public CrackMe Lab](https://bla.tools/crackme/): Public verifiable CrackMe challenges (`challenge/v6`–`v9`) with cryptographic `SEAL.json` commitments.
- [FAQ & Honesty Boundaries](https://bla.tools/faq/): Performance impact, supported formats, and BLA's strict Honesty Red Line (what client-side hardening solves vs. server-side authorization).
- [Enterprise Hardening](https://bla.tools/enterprise/): Custom adversarial assessment, CI/CD integration, and enterprise hardening engagements.

## Technical Reference & Schemas

- [Full LLM Technical Reference](https://bla.tools/llms-full.txt)
- [GitHub Repository (`qybaihe/bla`)](https://github.com/qybaihe/bla)
- [npm Package (`bla-agents`)](https://www.npmjs.com/package/bla-agents)
- [Configuration JSON Schema (`bla.config.v1.json`)](https://bla.tools/schemas/bla.config.v1.json)
- [Signed Defense Manifest Schema (`defense-manifest-v1.json`)](https://bla.tools/schemas/defense-manifest-v1.json)
