Feature: Team Agent, Adv. Terminal, MCP Tools, GPU/CPU Instruments & CUDA Docker #384
+5,590
−142
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Edit: Initialize.sh to fix perms issue by adding -f force.
Edit: Prompts for Team Agent support Code Exe tool, input tool, and instructions to support individual terminal reset and better terminal management for complex coding tasks.
-both support for CPU based (debian slim) or gpu (CUDA based) images. -both utilize .sh script that creates and maintains a deticated instruments_venv for dependencies and include heartbeats that prevent terminal passback (api waste). -code_exe tool timeouts have been changes to a rolling window, if idle for 10 seconds it passes back, else there is no max exe time if a process is responsive.
…nhanced to detect common shell prompt patterns (like (venv) ...$, root@...:~#, etc.) using regex. When such a prompt is detected in the output, the function will immediately break and return the output, rather than waiting for the full timeout.
…UI errors during code execution tool)
…ng agent loop at the agent.py level. The call to self.call_extensions("message_loop_prompts", ...) in prepare_prompt is now wrapped in a try/except block.
…ases where the parsed string is just '+', '-', or empty, returning it as a string instead of raising a ValueError.
…n indentation-error prone methods.
…ggle, team leader planning phase, and task document handling, team leader integrate results task document handling *allows for persistent development between teams by leveraging team leader as an investigative task manager before and after team agent task execution cycles*
…tiple seperate f.write calls, complex string manipulation. encouraged reading, implement fix, and check syntax. context a little bulky but working for now.
…t if the last 128 characters repeat 5 times consecutively and will stop execution if this pattern occurs, indicating a likely infinite loop. Now, when a repeating output pattern is detected: An error message will be printed to the console, indicating the loop and that the session is being reset. The log will be updated. self.reset_terminal(session=session) will be called to reset the specific session where the loop occurred. A detailed message will be returned to the agent, including the captured output, the loop detection confirmation, a notification that the session was reset, and advice to review the problematic code/command.
…LM with less repeated context.
|
Added:
Edit:
Fix:
|
… Servers feat: (draft) support MCP Servers feat: install npx for local MCP Servers execution feat: add nest-asyncio as direct dependency feat: add pdf2image to requirements.txt feat: add local nginx for playwright file access feat: MCP Server Support (Part 1: local stdio servers)
…nfusion between 'mcp' package and 'mcp.py'. Edited dockerfile to 'latest' so it doesnt spin up old builds after image creation. Also edited preinstall to ensure the mcp related dependencies can load non-interactive.
…uto mcp install if present in settings config (on compose)
…nt to not get distracted, to continue process.
…talled when the system starts.
…ly cast to a string before any length checks or slicing operations are performed on it. I also renamed the variable from user_message_context to user_message_text for better clarity, as it primarily holds the textual part of the user's message. This should provide the agent with a more robust and helpful contextual block after each MCP tool execution.
|
Feature: MCP Tool Integration, Auto-Setup, and Enhancements (Post-#384) This PR builds upon the work in #384, primarily integrating Rafa's MCP tool implementation (from @Rafael-U #332, with commit history preserved via cherry-picking) and introducing several enhancements to make MCP server usage more robust and user-friendly. Key Updates in this PR:
Current Status: Initial reviews indicate that various MCPs are functioning as expected. There might still be some minor quirks to address. The This PR aims to significantly improve the MCP tooling experience within Agent Zero. |
deciduus
changed the title
|
Edits to clean up the PR based on Jan's feedback: |
|
Added new instrument for retrieving youtube video transcripts easily by providing link. Resolved merge conflicts. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary
This PR introduces several major enhancements and fixes across the codebase, focusing on improved agent tooling, Docker support, and advanced media generation capabilities.
🚀 Added
.shscript to create and manage aninstruments_venvfor dependencies.🛠️ Edited
-f(force) flag to resolve permissions issues.📝 Notes