case-study 2026-10-10 case-studyelectronevidenceteachingauthorized

rea-playbook: Evidence → file:line → recreate (no engine required)

Install docs get you a server. **rea-playbook** shows what to do next: shrink a 1.8 MB Evidence blob into a readable file:line report, probe MCP, and finish a recreate loop — on an in-repo Electron demo, without Hopper/Ghidra.

Source meta

0 GitHub stars (snapshot)

Language: JavaScript

License: MIT

Updated: 2026-10-10

GitHub source ↗

rea.run rating

5/5 — Best net-new teaching layer

Quality. Minimal Electron target, Evidence→file:line extractor, MCP probe, verified recreate loop; CI + 28 tests.

Evidence. Executable playbook; pins rea-agents; no Ghidra/Hopper required for the JS path.

Limits. Teaching demo only — not a substitute for upstream docs or native engines.

For. Builders who finished install and need “what do I do with Evidence?”

What it is. MIT playbook (JS) with CI and 28 tests: minimal Electron app covering IPC/routing/DOM/fetch, an Evidence extractor, an MCP handshake probe, and a recreate sample that asserts 10/10.

Relation to REA. Pins rea-agents and teaches the Evidence contract. Complements /docs/install/ and the Electron example — it is a teaching layer, not a fork of REA.

Workflow. Run the demo target → ask REA for Evidence → extract line-referenced findings → recreate the feature in your stack → keep unknowns marked.

Evaluation. Highest teaching ROI among net-new 2026-10-10 finds. Star count is still zero; engineering signal is high.

Compliance

In-repo Electron demo only; authorized local analysis. No commercial targets.

Takeaways

Evidence files need a human-readable pass.
JS/Electron path can teach without native engines.
Own demo targets beat random binaries.

Related on rea.run