coredipper commited on
Commit
4dd9037
·
verified ·
1 Parent(s): 1915f95

docs: improve README with usage instructions

Browse files
Files changed (2) hide show
  1. README.md +11 -10
  2. requirements.txt +1 -1
README.md CHANGED
@@ -11,19 +11,20 @@ license: mit
11
  short_description: Classify threats and route to different healing mechanisms
12
  ---
13
 
14
- # Operon Immunity Healing Router
15
 
16
- An API gateway pattern that classifies threats using InnateImmunity and routes to different healing mechanisms based on severity.
17
 
18
- ## Features
19
 
20
- - **Threat classification**: InnateImmunity detects injection, abuse, and structural issues
21
- - **Severity routing**: CLEAN -> passthrough, LOW -> chaperone repair, MEDIUM -> autophagy, HIGH -> reject
22
- - **Healing pipeline**: ChaperoneLoop for structural repair, AutophagyDaemon for content cleanup
23
- - **Presets**: Clean input, mild issues, moderate pollution, injection attack
24
 
25
- ## Motifs Combined
26
 
27
- InnateImmunity + ChaperoneLoop + AutophagyDaemon + Cascade
28
 
29
- [GitHub](https://github.com/coredipper/operon) | [PyPI](https://pypi.org/project/operon-ai/)
 
 
 
11
  short_description: Classify threats and route to different healing mechanisms
12
  ---
13
 
14
+ # 🛡 Operon Immunity Healing Router
15
 
16
+ Classify threats via InnateImmunity and route to escalating healing mechanisms -- passthrough, chaperone repair, autophagy cleanup, or hard reject -- like an immune system triaging pathogens by severity.
17
 
18
+ ## What to Try
19
 
20
+ 1. Select a preset from the **Preset** dropdown (e.g. "Clean input", "Mild issues", or "Injection attack") and click **Run Router** to see how the immunity layer classifies the threat and routes it to the appropriate healing mechanism.
21
+ 2. Type your own text into the **Input** textbox -- try mixing legitimate questions with injection attempts to see how the router heals instead of rejecting.
22
+ 3. Compare the "Moderate pollution" preset (routed to autophagy) vs. "Injection attack" (hard reject) to see how severity determines the response.
 
23
 
24
+ ## How It Works
25
 
26
+ InnateImmunity scans input for injection patterns and abuse, assigning an inflammation level. Clean input passes through; low-severity issues get chaperone structural repair; medium-severity triggers autophagy cleanup; high-severity is rejected outright -- healing what can be saved while blocking genuine threats.
27
 
28
+ ## Learn More
29
+
30
+ [GitHub](https://github.com/coredipper/operon) | [PyPI](https://pypi.org/project/operon-ai/) | [Paper](https://github.com/coredipper/operon/tree/main/article)
requirements.txt CHANGED
@@ -1,2 +1,2 @@
1
  gradio>=4.0
2
- operon-ai>=0.14.0
 
1
  gradio>=4.0
2
+ operon-ai>=0.15.0