feat(agents): add autonomous web search and research agent - #136
Open
SurajsinghBayas wants to merge 3 commits into
Open
feat(agents): add autonomous web search and research agent#136SurajsinghBayas wants to merge 3 commits into
SurajsinghBayas wants to merge 3 commits into
Conversation
There was a problem hiding this comment.
Pull request overview
This PR adds a new a2a-search-agent under agents/ to act as an autonomous research assistant, providing web search and webpage-reading capabilities to an A2A/OpenAI-based agent runtime.
Changes:
- Introduces
SearchToolsetimplementingsearch_web(DuckDuckGo) andread_webpage(requests + BeautifulSoup extraction). - Adds an OpenAI agent definition (system prompt + tool wiring) and an
OpenAIAgentExecutorto orchestrate tool-calling loops. - Adds packaging and runtime assets (AgentCard, Dockerfile, docker-compose, pyproject, scripts, README).
Reviewed changes
Copilot reviewed 11 out of 12 changed files in this pull request and generated 4 comments.
Show a summary per file
| File | Description |
|---|---|
| agents/a2a-search-agent/src/search_toolset.py | Implements DuckDuckGo search and webpage text extraction tools. |
| agents/a2a-search-agent/src/openai_agent.py | Defines system prompt and exposes tools to the executor. |
| agents/a2a-search-agent/src/openai_agent_executor.py | Executes OpenAI chat loop with tool calling and A2A task updates. |
| agents/a2a-search-agent/src/main.py | CLI entrypoint wiring A2A server, agent card, and executor. |
| agents/a2a-search-agent/run_with_phoenix.sh | Convenience script to run the agent with Phoenix env vars. |
| agents/a2a-search-agent/README.md | Documentation for the agent (currently inconsistent with this agent’s purpose). |
| agents/a2a-search-agent/pyproject.toml | Python project metadata and dependencies for the agent. |
| agents/a2a-search-agent/Dockerfile | Container build/run configuration for the agent. |
| agents/a2a-search-agent/docker-compose.yml | Compose service definition for running the agent container. |
| agents/a2a-search-agent/AgentCard.json | A2A AgentCard metadata describing skills and capabilities. |
| agents/a2a-search-agent/.gitignore | Agent-local gitignore for Python build artifacts and environments. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Comment on lines
+43
to
+46
| logger.info(f"Reading webpage URL: {url}") | ||
| try: | ||
| response = self.session.get(url, timeout=10) | ||
| response.raise_for_status() |
Comment on lines
+1
to
+12
| # A2A Translator Agent | ||
|
|
||
| An intelligent translation agent built with A2A (Agent2Agent) SDK that can translate text and web page content using natural language. | ||
|
|
||
| ## Features | ||
|
|
||
| - **Text Translation**: Translate plain text between different languages | ||
| - **URL Content Translation**: Extract and translate content from web pages | ||
| - **Language Detection**: Automatically detect the language of text or web content | ||
| - **Multi-language Support**: Supports translation between multiple languages using Google Translate | ||
| - **Clean Web Content**: Extracts readable text from HTML pages, removing scripts and styling | ||
|
|
Comment on lines
+1
to
+6
| #!/bin/bash | ||
|
|
||
| # Set environment variables for observability | ||
| export PHOENIX_PROJECT_NAME="a2a-search-agent" | ||
| export PHOENIX_API_KEY="your_phoenix_api_key_here" | ||
|
|
Co-authored-by: Copilot Autofix powered by AI <175728472+Copilot@users.noreply.github.com>
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 11 out of 12 changed files in this pull request and generated 5 comments.
Comments suppressed due to low confidence (3)
agents/a2a-search-agent/src/search_toolset.py:46
read_webpagefetches arbitrary URLs without validating scheme/host, which enables SSRF (e.g., hitting localhost services) and non-HTTP schemes. Add basic URL validation (http/https only) and block localhost at minimum before issuing the request.
logger.info(f"Reading webpage URL: {url}")
try:
import socket
import ipaddress
agents/a2a-search-agent/README.md:5
- This README appears to be copied from the translator agent (title, description, feature list, and dependencies mention translation/Google Translate) and does not document the new web search agent. Update it to describe web search + webpage reading behavior, tools/skills, and correct dependencies.
# Web Search Agent
An autonomous research and web search agent for the Nasiko platform built using the A2A SDK.
## Capabilities
agents/a2a-search-agent/run_with_phoenix.sh:6
- The script exports a placeholder
PHOENIX_API_KEYvalue. This can lead to confusing behavior (sending invalid credentials) and encourages hardcoding secrets. Prefer readingPHOENIX_API_KEYfrom the environment and warning if it is unset.
# Ensure we run from the agent directory
cd "$(dirname "$0")" || exit 1
# Set environment variables for observability
Comment on lines
+109
to
+113
| method = getattr(tool_instance, function_name) | ||
| result = method(**function_args) | ||
| # Check if the result is a coroutine and await it | ||
| if inspect.iscoroutine(result): | ||
| result = await result |
Comment on lines
+84
to
+91
| # Add assistant's response to messages | ||
| messages.append( | ||
| { | ||
| "role": "assistant", | ||
| "content": message.content, | ||
| "tool_calls": message.tool_calls, | ||
| } | ||
| ) |
Comment on lines
+97
to
+99
| function_name = tool_call.function.name | ||
| function_args = json.loads(tool_call.function.arguments) | ||
|
|
Comment on lines
+48
to
+69
| skill = AgentSkill( | ||
| id="search_web", | ||
| name="Search Web", | ||
| description="Search the internet and read web pages to answer queries", | ||
| tags=["search", "web", "research", "url"], | ||
| examples=[ | ||
| "Search the web for the latest news on AI", | ||
| "Who won the Superbowl in 2024?", | ||
| ], | ||
| ) | ||
|
|
||
| # AgentCard for OpenAI-based agent | ||
| agent_card = AgentCard( | ||
| name="Web Search Agent", | ||
| description="An agent that can autonomously search the internet and read web pages to answer queries", | ||
| url=f"http://{host}:{port}/", | ||
| version="1.0.0", | ||
| default_input_modes=["text"], | ||
| default_output_modes=["text"], | ||
| capabilities=AgentCapabilities(streaming=True), | ||
| skills=[skill], | ||
| ) |
Comment on lines
+7
to
+17
| RUN pip install --no-cache-dir \ | ||
| "a2a-sdk[http-server]>=0.3.0,<1.0.0" \ | ||
| click>=8.1.8 \ | ||
| httpx>=0.28.1 \ | ||
| openai>=1.57.0 \ | ||
| pydantic>=2.11.4 \ | ||
| python-dotenv>=1.1.0 \ | ||
| uvicorn>=0.34.2 \ | ||
| requests>=2.31.0 \ | ||
| beautifulsoup4>=4.12.0 \ | ||
| duckduckgo-search>=6.0.0 |
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 11 out of 12 changed files in this pull request and generated 4 comments.
Comments suppressed due to low confidence (1)
agents/a2a-search-agent/Dockerfile:17
- Unquoted package specifiers like
click>=8.1.8will be parsed by/bin/shas output redirections (>), which can break the image build and create stray files. Quote each requirement (or use a requirements file) sopipreceives the full specifier.
RUN pip install --no-cache-dir \
"a2a-sdk[http-server]>=0.3.0,<1.0.0" \
click>=8.1.8 \
httpx>=0.28.1 \
openai>=1.57.0 \
pydantic>=2.11.4 \
python-dotenv>=1.1.0 \
uvicorn>=0.34.2 \
requests>=2.31.0 \
beautifulsoup4>=4.12.0 \
duckduckgo-search>=6.0.0
Comment on lines
+62
to
+65
| response = self.session.get(url, timeout=10) | ||
| response.raise_for_status() | ||
|
|
||
| soup = BeautifulSoup(response.content, 'html.parser') |
Comment on lines
+49
to
+60
| parsed_url = urlparse(url) | ||
| if parsed_url.scheme not in ("http", "https"): | ||
| raise ValueError("Only http and https schemes are supported.") | ||
|
|
||
| # Resolve the hostname to an IP address | ||
| try: | ||
| ip = socket.gethostbyname(parsed_url.hostname) | ||
| ip_obj = ipaddress.ip_address(ip) | ||
| if ip_obj.is_private or ip_obj.is_loopback or ip_obj.is_link_local: | ||
| raise ValueError(f"Access to private/local IP ({ip}) is blocked for security reasons.") | ||
| except socket.gaierror: | ||
| raise ValueError("Could not resolve hostname.") |
Comment on lines
+11
to
+13
| ports: | ||
| - "5000" | ||
| tty: true |
|
|
||
| 3. Alternatively, run the agent using Python: | ||
| ```bash | ||
| pip install -r pyproject.toml |
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
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.
Description
This PR introduces a new Web Search Agent (
a2a-search-agent) to theagents/directory. It is built using the A2A SDK and serves as an autonomous research assistant capable of searching the web and parsing detailed webpage content to answer queries.Key Features & Changes
AgentCard.json): Configured the agent withsearch-webandread-webpageskills.search_toolset.py): Implemented the core logic usingduckduckgo-searchto query the web andbeautifulsoup4to scrape and extract text from URLs.openai_agent.py): Added a tailored system prompt that instructs the LLM on how to iteratively search the web and synthesize long-form content.Dockerfile,docker-compose.yml, andpyproject.tomlto include all necessary scraping dependencies.Verification
pip.(Note to maintainers: This agent uses free packages like
duckduckgo-searchto avoid needing paid API keys for web search functionality).