Same agent. Same code. Different behavior.
⸻
What this is
This is a real LangChain agent.py.
It has not been modified.
Instead, it has been wrapped with a second runtime layer that changes how it behaves during execution.
⸻
The idea
You normally improve software by changing its code.
This takes a different approach:
Attach behavior at runtime instead of modifying the source
⸻
Files in this repo
→ Original source (Layer 1, MIT licensed)
→ Ascended version (Layer 2 attached)
→ Artifact identity (CJPI score + fingerprint)
→ Pre-export verification (passed)
→ Debug + verify modes and capability breakdown
→ Original MIT license
→ Layer 2 license (CAAL-1.0)
→ Attribution and license separation
⸻
Before vs After
Original (agent.py)
• Stateless execution
• One-shot behavior
• No runtime mediation
⸻
Ascended (cmpsbl_agent.py)
• Memory across interactions
• Autonomous decision loop
• Runtime safety + cost controls
• Execution mediation layer
⸻
Verification
This artifact includes a full verification chain: • ✔ Layer 1 preserved byte-for-byte  • ✔ Layer 2 successfully attached • ✔ 6 runtime layers auto-wired • ✔ Entry point + 41 handlers detected
⸻
Try it
Run the original:
python agent.py
Run the ascended version:
python cmpsbl_agent.py
Ask the same questions and compare behavior.
⸻
Example
Try:
Explain recursion What did I ask before?
The responses will diverge.
⸻
What changed?
Not the code.
The execution.
⸻
Architecture
Layer 1:
• Original source (MIT licensed)
Layer 2:
• External runtime layer (CMPSBL)
Layer 2 attaches to execution without modifying Layer 1.
⸻
User Guide
See:
USER-GUIDE.html
Includes:
• capability breakdown
• debug mode
• verification steps
• pipeline details
⸻
License
• Layer 1: MIT (original source) 
• Layer 2: CAAL-1.0 (CMPSBL runtime) 
⸻
Why this exists
To test a simple question:
Can you change how software behaves without changing the software itself?
⸻
Discussion
Curious what people think.
Is this useful?
Or just a different way to structure the same ideas?