Apply PR aliasrobotics#389: Debug Dockerfile

Author: giveen

.github/instructions/todos.instructions.md

@@ -2,15 +2,19 @@
 applyTo: '**'
 ---
 
-<todos title="Apply PRs #413 and #411 - handle missing API key & fix_message_list" agentRequirement="Review steps frequently throughout the conversation and DO NOT stop between steps unless they explicitly require it. Keep the todos updated using the todo_write tool (do not edit this file).">
-- [x] apply-pr-413: Apply PR #413 changes to src/cai/agents/__init__.py (handle missing OpenAI API key gracefully) 🔴
-- [x] run-import-checks: Run import/compile checks for modified modules 🟡
-- [ ] run-tests: Run project's test suite (pytest) and report failures 🟡
-- [ ] push-changes: Push changes to fork remote and create a PR branch 🟢
-- [x] apply-pr-411: Apply PR #411 changes: fix infinite loop in fix_message_list and add reordering tests 🔴
-- [x] run-import-checks-411: Run import/compile checks for modified modules after PR #411 changes 🟡
-- [x] run-tests-411: Run tests/test_fix_message_list_reorder.py and report results 🟡
-- [ ] push-changes-411: Push PR #411 changes to fork remote and create PR branch 🟢
+<todos title="Fix duplicate context display + apply PR #389" agentRequirement="Review steps frequently throughout the conversation and DO NOT stop between steps unless they explicitly require it. Keep the todos updated using the todo_write tool (do not edit this file).">
+- [x] investigate-duplicate-context: Investigate duplicate 'Context' status appearing in stream UI 🔴
+  _Found duplicated context in both header tokens_text and footer compact_tokens_
+- [x] add-include-context-flag: Add `include_context` flag to `_create_token_display` to make header context optional 🟡
+  _Implemented optional flag to avoid duplicating context indicator when footer has compact info_
+- [x] suppress-header-context-in-streaming: Call `_create_token_display(..., include_context=False)` in streaming final_stats path 🟡
+  _Updated streaming final_stats call to suppress header context when footer compact tokens are present_
+- [ ] verify-change: Run quick visual check / focused test to ensure duplicate context removed from streaming output 🔴
+  _Recommend running a focused run that produces streaming output to visually confirm; can run `pytest -q tests/test_fix_message_list_reorder.py` or start an agent stream locally_
+- [-] apply-pr-389: Fetch and apply PR #389 changes (dockerized/Dockerfile) 🟡
+  _Added Kali mirror fix to dockerized/Dockerfile per upstream PR #389_
+- [ ] commit-push-pr-389: Commit the applied PR changes and push to fork/branch for review 🟡
+  _Commit message: Apply PR #389: Debug Dockerfile; push to giveen/buggy or create branch_
 </todos>
 
 <!-- Auto-generated todo section -->

dockerized/Dockerfile

@@ -2,6 +2,9 @@ FROM kalilinux/kali-rolling
 
 ENV DEBIAN_FRONTEND=noninteractive
 
+# [FIX BAD MIRROR] Use a single, reliable Kali mirror to avoid flaky mirror redirects during apt operations
+RUN printf 'deb http://kali.download/kali kali-rolling main contrib non-free non-free-firmware\n' > /etc/apt/sources.list
+
 # Fix Kali GPG Keys
 RUN apt-get update --allow-insecure-repositories && \
     apt-get install -y --no-install-recommends --allow-unauthenticated kali-archive-keyring && \

docs/cai/getting-started/commands.md

@@ -366,7 +366,7 @@ cai
 /mcp load http://localhost:9876/sse burp
 
 # Add MCP tools to agent
-/mcp add-to-agent <agent_name> burp
+/mcp add burp <agent_name>
 
 # Set up virtualized environment
 /virtualization create ubuntu:latest

docs/cli/advanced_usage.md

@@ -1143,7 +1143,7 @@ burp-mcp-server --port 9876
 # In CAI
 CAI> /mcp load http://localhost:9876/sse burp
 CAI> /mcp tools burp
-CAI> /mcp add redteam_agent burp
+CAI> /mcp add burp redteam_agent
 
 # Use Burp tools
 CAI> Use Burp to scan https://target.com
@@ -1156,7 +1156,7 @@ CAI> Use Burp to scan https://target.com
 CAI> /mcp load stdio "python my_custom_tools.py" custom
 
 # Add to agent
-CAI> /mcp add bug_bounter_agent custom
+CAI> /mcp add custom bug_bounter_agent
 
 # Use custom tools
 CAI> Use custom scanner on target

docs/cli/commands_reference.md

@@ -635,11 +635,11 @@ List active MCP connections.
 /mcp list
 ```
 
-#### `add <agent_name> <server_name>`
+#### `add <server_name> <agent_name>`
 Add MCP tools to an agent.
 
 ```bash
-/mcp add redteam_agent burp
+/mcp add burp redteam_agent
 ```
 
 #### `remove <server_name>`
@@ -680,7 +680,7 @@ CAI> /mcp load http://localhost:9876/sse burp
 CAI> /mcp tools burp
 
 # Add to current agent
-CAI> /mcp add redteam_agent burp
+CAI> /mcp add burp redteam_agent
 
 # Check status
 CAI> /mcp status

src/cai/sdk/agents/models/openai_chatcompletions.py

@@ -3673,8 +3673,7 @@ async def _fetch_response_litellm_openai(
         """
         try:
             if stream:
-                # Standard LiteLLM handling for streaming
-                ret = await litellm.acompletion(**kwargs)
+                # Standard LiteLLM handling for streaming — make a single streaming call
                 stream_obj = await litellm.acompletion(**kwargs)
 
                 response = Response(
@@ -3728,7 +3727,7 @@ async def _fetch_response_litellm_openai(
                 kwargs["messages"] = messages
                 # Retry once, silently
                 if stream:
-                    ret = await litellm.acompletion(**kwargs)
+                    # Retry streaming call once (single request)
                     stream_obj = await litellm.acompletion(**kwargs)
                     response = Response(
                         id=FAKE_RESPONSES_ID,

src/cai/util.py

@@ -1615,6 +1615,7 @@ def _create_token_display(
     model,
     interaction_cost=None,
     total_cost=None,
+    include_context: bool = True,
 ) -> Text:
     # Standardize model name
     model_name = get_model_name(model)
@@ -1661,24 +1662,25 @@ def _create_token_display(
     tokens_text.append("Session: ", style="bold magenta")
     tokens_text.append(f"${COST_TRACKER.session_total_cost:.4f}", style="bold magenta")
 
-    # Context usage
-    tokens_text.append(" | ", style="dim")
-    context_pct = interaction_input_tokens / get_model_input_tokens(model_name) * 100
-    tokens_text.append("Context: ", style="bold")
-    tokens_text.append(f"{context_pct:.1f}% ", style="bold")
-
-    # Context indicator
-    if context_pct < 50:
-        indicator = "🟩"
-        color_local = "green"
-    elif context_pct < 80:
-        indicator = "🟨"
-        color_local = "yellow"
-    else:
-        indicator = "🟥"
-        color_local = "red"
+    # Context usage (optional)
+    if include_context:
+        tokens_text.append(" | ", style="dim")
+        context_pct = interaction_input_tokens / get_model_input_tokens(model_name) * 100
+        tokens_text.append("Context: ", style="bold")
+        tokens_text.append(f"{context_pct:.1f}% ", style="bold")
+
+        # Context indicator
+        if context_pct < 50:
+            indicator = "🟩"
+            color_local = "green"
+        elif context_pct < 80:
+            indicator = "🟨"
+            color_local = "yellow"
+        else:
+            indicator = "🟥"
+            color_local = "red"
 
-    tokens_text.append(f"{indicator}", style=color_local)
+        tokens_text.append(f"{indicator}", style=color_local)
 
     return tokens_text
 
@@ -2539,6 +2541,7 @@ def finish_agent_streaming(context, final_stats=None):
                     model_name,  # string model name!
                     interaction_cost,
                     total_cost,
+                    include_context=False,
                 )
 
                 # Crear una línea de tokens compacta para el streaming