docs: use relative paths for MP4 videos in all READMEs to ensure correct rendering

Author: knowrohit

README.md

@@ -3,7 +3,7 @@
 **What it really means to be a 10x engineer—and the tool built for that reality.**
 
 <p align="center">
-  <video src="https://github.com/SRSWTI/axe/raw/main/assets/mp4/axe_axe_axe_ending.mp4" width="100%" autoplay loop muted playsinline>
+  <video src="assets/mp4/axe_axe_axe_ending.mp4" width="100%" autoplay loop muted playsinline>
     axe in action
   </video>
 </p>
@@ -110,7 +110,7 @@ To demonstrate the precision advantage, we built a minimal CLI agent implementat
 ### Example 1: Basic Edit Operations
 
 <p align="center">
-  <video src="https://github.com/SRSWTI/axe/raw/main/assets/mp4/axe_comparison.mp4" width="100%" autoplay loop muted playsinline>
+  <video src="assets/mp4/axe_comparison.mp4" width="100%" autoplay loop muted playsinline>
     comparison
   </video>
 </p>
@@ -125,7 +125,7 @@ The difference is clear. The basic agent searches blindly, while axe-dig underst
 When asked to explain how call flow tracking works, both agents found the context—but the results were dramatically different.
 
 <p align="center">
-  <video src="https://github.com/SRSWTI/axe/raw/main/assets/mp4/axe_explain_dig.mp4" width="100%" autoplay loop muted playsinline>
+  <video src="assets/mp4/axe_explain_dig.mp4" width="100%" autoplay loop muted playsinline>
     call flow part 1
   </video>
 </p>
@@ -143,7 +143,7 @@ axe-dig didn't just use fewer tokens—it provided **better understanding** of h
 The difference compounds with follow-up questions. When we asked about caller information:
 
 <p align="center">
-  <video src="https://github.com/SRSWTI/axe/raw/main/assets/mp4/axe_call_flow_compounding.mp4" width="100%" autoplay loop muted playsinline>
+  <video src="assets/mp4/axe_call_flow_compounding.mp4" width="100%" autoplay loop muted playsinline>
     call flow part 2
   </video>
 </p>
@@ -158,7 +158,7 @@ The difference compounds with follow-up questions. When we asked about caller in
 In the mlx-lm codebase, when asked how to compute DWQ targets:
 
 <p align="center">
-  <video src="https://github.com/SRSWTI/axe/raw/main/assets/mp4/axe_better_inference.mp4" width="100%" autoplay loop muted playsinline>
+  <video src="assets/mp4/axe_better_inference.mp4" width="100%" autoplay loop muted playsinline>
     better inference
   </video>
 </p>
@@ -212,7 +212,7 @@ The result: **axe works brilliantly with both local and cloud models**, because
 Here's axe running with **srswti/blackbird-she-doesnt-refuse-21b**—a 21B parameter model from our Bodega collection, running entirely locally:
 
 <p align="center">
-  <video src="https://github.com/SRSWTI/axe/raw/main/assets/mp4/axe_subagents.mp4" width="100%" autoplay loop muted playsinline>
+  <video src="assets/mp4/axe_subagents.mp4" width="100%" autoplay loop muted playsinline>
     local model demonstration
   </video>
 </p>
@@ -391,7 +391,7 @@ chop dfg src/db.py update_user
 **Subagents in action:**
 
 <p align="center">
-  <video src="https://github.com/SRSWTI/axe/raw/main/assets/mp4/axe_subagents.mp4" width="100%" autoplay loop muted playsinline>
+  <video src="assets/mp4/axe_subagents.mp4" width="100%" autoplay loop muted playsinline>
     subagents
   </video>
 </p>
@@ -438,7 +438,7 @@ pytest tests/
 Hit **Ctrl+X** to toggle between axe and your normal shell. No context switching. No juggling terminals.
 
 <p align="center">
-  <video src="https://github.com/SRSWTI/axe/raw/main/assets/mp4/axe_axe_sample_toggle_shell.mp4" width="100%" autoplay loop muted playsinline>
+  <video src="assets/mp4/axe_axe_sample_toggle_shell.mp4" width="100%" autoplay loop muted playsinline>
     shell toggle
   </video>
 </p>
@@ -547,7 +547,7 @@ Understanding code isn't just about reading—it's about **seeing** the structur
 The dashboard provides real-time visualization for:
 
 <p align="center">
-  <video src="https://github.com/SRSWTI/axe/raw/main/assets/mp4/axe_axe_future.mp4" width="100%" autoplay loop muted playsinline>
+  <video src="assets/mp4/axe_axe_future.mp4" width="100%" autoplay loop muted playsinline>
     dashboard visualization
   </video>
 </p>

docs/AXE-DIG.md

@@ -291,7 +291,11 @@ To demonstrate the precision advantage, we built minimal CLI agent implementatio
 
 When asked to explain how call flow tracking works:
 
-![call flow explanation](../assets/axe_gif_explain_dig.gif)
+<p align="center">
+  <video src="../assets/mp4/axe_explain_dig.mp4" width="100%" autoplay loop muted playsinline>
+    call flow explanation
+  </video>
+</p>
 
 **Left:** Basic agent had to read the entire file after grepping for literal strings. **44,000 tokens**.  
 **Right:** axe-dig used **17,000 tokens** while discovering:
@@ -305,7 +309,11 @@ axe-dig provided better understanding with fewer tokens.
 
 When asked about caller information:
 
-![compounding effect](../assets/axe_gif_call_flow_compounding.gif)
+<p align="center">
+  <video src="../assets/mp4/axe_call_flow_compounding.mp4" width="100%" autoplay loop muted playsinline>
+    compounding effect
+  </video>
+</p>
 
 **Left:** Started wrong, inferred wrong, continued wrong.  
 **Right:** Had more context from the start, leading to precise answers.
@@ -314,7 +322,11 @@ When asked about caller information:
 
 In the mlx-lm codebase, when asked how to compute DWQ targets:
 
-![better inference](../assets/axe_gif_better_inference.gif)
+<p align="center">
+  <video src="../assets/mp4/axe_better_inference.mp4" width="100%" autoplay loop muted playsinline>
+    better inference
+  </video>
+</p>
 
 **Left:** Explained the concept generically.  
 **Right:** axe actively searched the codebase and found the actual implementation.

examples/README.md

@@ -163,13 +163,21 @@ pytest tests/test_db.py
 
 Hit **Ctrl+X** to seamlessly switch between axe and your normal shell:
 
-![shell toggle](../assets/axe_gif_axe_sample_toggle_shell.gif)
+<p align="center">
+  <video src="../assets/mp4/axe_axe_sample_toggle_shell.mp4" width="100%" autoplay loop muted playsinline>
+    shell toggle
+  </video>
+</p>
 
 ### Precision vs. Basic Search
 
 See the difference between basic grep-based agents and axe-dig's intelligent code understanding:
 
-![comparison](../assets/axe_gif_comparison.gif)
+<p align="center">
+  <video src="../assets/mp4/axe_comparison.mp4" width="100%" autoplay loop muted playsinline>
+    comparison
+  </video>
+</p>
 
 **Left:** Basic CLI agent with grep  
 **Right:** axe with axe-dig tools

src/axe_cli/agents/README.md

@@ -8,7 +8,11 @@ axe isn't limited to one main agent. You can create subagents and tasks for *any
 
 Need a dedicated security researcher? A ruthlessly precise code reviewer? A creative copywriter? axe can create and deploy specialized subagents based on your exact requirements. These subagents help you complete tasks better, faster, and more efficiently—operating with lethal precision to divide and conquer complex workflows.
 
-![subagents in action](../../assets/axe_gif_subagents.gif)
+<p align="center">
+  <video src="../../../assets/mp4/axe_subagents.mp4" width="100%" autoplay loop muted playsinline>
+    subagents in action
+  </video>
+</p>
 
 **Subagents enable parallel task execution:** Spawn multiple specialized agents to work on different aspects of a problem simultaneously, each with their own context and tools.