merge: merged dev to switch-to-pg

.github/workflows/generate-changelog.yml

@@ -0,0 +1,83 @@
+name: Julep Changelog Generation
+run-name: ${{ github.actor }} is generating changelog for the last two weeks using Julep
+
+on:
+  workflow_dispatch:
+
+jobs:
+  changelog_generation:
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Checkout code
+        uses: actions/checkout@v4
+        with:
+          fetch-depth: 0
+          ref: dev
+
+      - name: Setup GitHub CLI
+        run: |
+          echo "${{ secrets.GITHUB_TOKEN }}" | gh auth login --with-token
+
+      - name: Collect merged PRs from the last two weeks
+        id: collect_prs
+        run: |
+          if [[ "$OSTYPE" == "darwin"* ]]; then
+            date_threshold=$(date -v-14d +"%Y-%m-%d")
+          else
+            date_threshold=$(date -d '-14 days' +"%Y-%m-%d")
+          fi
+
+          echo "Fetching merged PRs since $date_threshold..."
+
+          merged_prs=$(
+            gh pr list --state merged --json number,title,body,author \
+              --search "merged:>=$date_threshold" \
+              --jq 'map({number, title, body, author: .author.login})'
+          )
+
+          if [ -z "$merged_prs" ] || [ "$merged_prs" = "null" ]; then
+            echo "No merged PRs found in the last two weeks."
+            echo "pr_data=[]" >> $GITHUB_ENV
+            echo '{"pr_data": []}' > pr_data.json
+            exit 0
+          fi
+
+          echo "pr_data=$merged_prs" >> $GITHUB_ENV
+          echo "$merged_prs" > pr_data.json
+
+      - name: Setup Python v3.10.12
+        uses: actions/setup-python@v5
+        with:
+          python-version: "3.10.12"
+
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          python -m pip install PyYAML julep
+
+      - name: Send PR data to Python script
+        if: steps.collect_prs.outputs.pr_data != '[]'
+        id: generate_changelog
+        run: |
+          if ! python scripts/generate_changelog.py; then
+            echo "Error: Failed to generate changelog"
+            exit 1
+          fi
+        env:
+          JULEP_API_KEY: ${{ secrets.JULEP_API_KEY }}
+          TASK_UUID: ${{ secrets.TASK_UUID }}
+          AGENT_UUID: ${{ secrets.AGENT_UUID }}
+
+      - name: Create Pull Request
+        if: success() && steps.collect_prs.outputs.pr_data != '[]'
+        uses: peter-evans/create-pull-request@v7
+        with:
+          token: ${{ secrets.GITHUB_TOKEN }}
+          commit-message: "chore(changelog): update CHANGELOG.md"
+          title: "Update CHANGELOG.md"
+          body: "This PR updates the changelog with PRs from the last 2 weeks."
+          branch: "update-changelog"
+          delete-branch: true
+          add-paths: |
+            CHANGELOG.md

CHANGELOG.md

@@ -1,15 +1,108 @@
-# TODO: Update changelog
-# SCRUM-29
+<div align="center" id="top">
+  <img src="https://socialify.git.ci/julep-ai/julep/image?description=1&descriptionEditable=Rapidly%20build%20AI%20workflows%20and%20agents&font=Source%20Code%20Pro&logo=https%3A%2F%2Fraw.githubusercontent.com%2Fjulep-ai%2Fjulep%2Fdev%2F.github%2Fjulep-logo.svg&owner=1&forks=1&pattern=Solid&stargazers=1&theme=Auto" alt="julep" width="640" height="320" />
+ </div>
+
+ <p align="center">
+   <br />
+   <a href="https://docs.julep.ai" rel="dofollow">Explore Docs (wip)</a>
+   ·
+   <a href="https://discord.com/invite/JTSBGRZrzj" rel="dofollow">Discord</a>
+   ·
+   <a href="https://x.com/julep_ai" rel="dofollow">𝕏</a>
+   ·
+   <a href="https://www.linkedin.com/company/julep-ai" rel="dofollow">LinkedIn</a>
+ </p>
+
+ <p align="center">
+     <a href="https://www.npmjs.com/package/@julep/sdk"><img src="https://img.shields.io/npm/v/%40julep%2Fsdk?style=social&amp;logo=npm&amp;link=https%3A%2F%2Fwww.npmjs.com%2Fpackage%2F%40julep%2Fsdk" alt="NPM Version"></a>
+     <span>&nbsp;</span>
+     <a href="https://pypi.org/project/julep"><img src="https://img.shields.io/pypi/v/julep?style=social&amp;logo=python&amp;label=PyPI&amp;link=https%3A%2F%2Fpypi.org%2Fproject%2Fjulep" alt="PyPI - Version"></a>
+     <span>&nbsp;</span>
+     <a href="https://hub.docker.com/u/julepai"><img src="https://img.shields.io/docker/v/julepai/agents-api?sort=semver&amp;style=social&amp;logo=docker&amp;link=https%3A%2F%2Fhub.docker.com%2Fu%2Fjulepai" alt="Docker Image Version"></a>
+     <span>&nbsp;</span>
+     <a href="https://choosealicense.com/licenses/apache/"><img src="https://img.shields.io/github/license/julep-ai/julep" alt="GitHub License"></a>
+ </p>
 
-# Changelog
+# **Julep AI Changelog for 12 December 2024** ✨
 
-All notable changes to this project will be documented in this file.
+Welcome to the latest Julep AI changelog! We’ve been busy refining and enhancing our systems to ensure a seamless experience for you. Dive into the exciting updates below!
 
-The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/),
-and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+## **Features** ✨
 
-## [Unreleased]
+- **Prometheus Summary for Cozo Queries**:
+  - A **Summary** metric has been introduced for query latency in `query_metrics_update()` within `counters.py`.
+  - The `increase_counter()` function has been renamed to `query_metrics_update()` across multiple files for consistency.
+  - **Why**: To improve performance monitoring and streamline code.
+  - **Impact**: Enhanced visibility into query performance and simplified codebase.
 
-### Added
+- **MMR Search and Configurable Parameters**:
+  - Introduced MMR search with configurable parameters, including `mmr_strength`, in `gather_messages.py`.
+  - Updated documentation and OpenAPI spec with new search parameters.
+  - **Why**: To offer flexible and powerful search capabilities.
+  - **Impact**: Users can now tailor search functionalities to better meet their needs.
 
-- v0.2 agents platform
+- **In-Memory Rate Limiter**:
+  - Added an in-memory rate limiter to `transition_step` using `RateLimiter` from `utils.py`.
+  - **Why**: To prevent system overload and ensure stable performance.
+  - **Impact**: Improved resource management and system reliability.
+
+## **Fixes** 🔧
+
+- **GitHub Action Workflow for Changelog Generation**:
+  - Fixed issues in the GitHub Action workflow by renaming and reconfiguring the workflow.
+  - **Why**: To ensure accurate changelog generation and avoid JSON structure errors.
+  - **Impact**: Streamlined changelog updates and enhanced workflow clarity.
+
+- **Grafana Dashboard and Temporal Metrics Port**:
+  - Resolved provisioning issues in Grafana dashboards and set a static port for Temporal metrics.
+  - **Why**: To maintain consistent monitoring and avoid configuration errors.
+  - **Impact**: Reliable dashboard visualization and metrics tracking.
+
+## **Improvements** 📈
+
+- **Metrics Refactoring**:
+  - Centralized initialization of `Counter`, `Summary`, and `Histogram` metrics in `counters.py`.
+  - Updated `query_metrics_update` to handle both sync and async functions.
+  - **Why**: To simplify metric management and enhance functionality.
+  - **Impact**: Improved performance tracking and easier maintenance.
+
+- **Async Support in Spider API**:
+  - Added async support and new methods to the Spider API, along with code refactoring.
+  - **Why**: To enhance the efficiency and flexibility of web crawling capabilities.
+  - **Impact**: Faster and more reliable API operations.
+
+## **Performance Enhancements** 🚀
+
+- **Integration with New Cozo RocksDB**:
+  - Updated configurations to integrate the latest Cozo with RocksDB, enhancing the engine and settings.
+  - **Why**: To leverage new features and improved performance of the Cozo database.
+  - **Impact**: Enhanced data processing speed and system robustness.
+
+- **Default Worker Settings**:
+  - Set default Temporal worker settings for improved concurrency and activity limits.
+  - **Why**: To optimize task handling and resource allocation.
+  - **Impact**: More efficient workflow management and reduced latency.
+
+## **Breaking Changes** 💥
+
+- **Deprecated Index References**:
+  - Removed unused index references from queries across multiple files to handle undefined indices.
+  - **Why**: To clean up the codebase and avoid potential errors.
+  - **Impact**: May require updates to dependent queries or scripts.
+
+## **Notes** 📝
+
+- **Attention Developers**: Please ensure you update your configurations and dependencies to align with the latest changes.
+- **Known Issues**: Some users might experience temporary delays in search functionalities as new indices are created.
+
+Thank you for using Julep AI! We’re committed to delivering the best possible experience and appreciate your support and feedback. Stay tuned for more updates!
+
+## Contributors
+
+Thank you to all our contributors who helped make this release possible!
+
+- [Dmitry Paramonov](https://github.com/whiterabbit1983) 🐇
+- [Ahmad Haidar](https://github.com/Ahmad-mtos) 🚀
+- [Diwank Tomar](https://github.com/creatorrr) 🌟
+- [Vedant Sahai](https://github.com/Vedantsahai18) 🔥
+- [Hamada Salhab](https://github.com/HamadaSalhab) 💡

agents-api/agents_api/activities/execute_system.py

@@ -17,6 +17,8 @@
     HybridDocSearchRequest,
     SystemDef,
     TextOnlyDocSearchRequest,
+    UpdateSessionRequest,
+    UpdateUserRequest,
     VectorDocSearchRequest,
 )
 from ..common.protocol.tasks import ExecutionInput, StepContext
@@ -111,10 +113,11 @@ async def execute_system(
             await bg_runner()
             return res
 
+        # Handle create session
         if system.operation == "create" and system.resource == "session":
             developer_id = arguments.pop("developer_id")
             session_id = arguments.pop("session_id", None)
-            data = CreateSessionRequest(**arguments)
+            create_session_request = CreateSessionRequest(**arguments)
 
             # In case sessions.create becomes asynchronous in the future
             if asyncio.iscoroutinefunction(handler):
@@ -123,14 +126,66 @@ async def execute_system(
             # Run the synchronous function in another process
             loop = asyncio.get_running_loop()
             return await loop.run_in_executor(
-                process_pool_executor, partial(handler, developer_id, session_id, data)
+                process_pool_executor,
+                partial(
+                    handler,
+                    developer_id=developer_id,
+                    session_id=session_id,
+                    data=create_session_request,
+                ),
+            )
+
+        # Handle update session
+        if system.operation == "update" and system.resource == "session":
+            developer_id = arguments.pop("developer_id")
+            session_id = arguments.pop("session_id")
+            update_session_request = UpdateSessionRequest(**arguments)
+
+            # In case sessions.update becomes asynchronous in the future
+            if asyncio.iscoroutinefunction(handler):
+                return await handler()
+
+            # Run the synchronous function in another process
+            loop = asyncio.get_running_loop()
+            return await loop.run_in_executor(
+                process_pool_executor,
+                partial(
+                    handler,
+                    developer_id=developer_id,
+                    session_id=session_id,
+                    data=update_session_request,
+                ),
+            )
+
+        # Handle update user
+        if system.operation == "update" and system.resource == "user":
+            developer_id = arguments.pop("developer_id")
+            user_id = arguments.pop("user_id")
+            update_user_request = UpdateUserRequest(**arguments)
+
+            # In case users.update becomes asynchronous in the future
+            if asyncio.iscoroutinefunction(handler):
+                return await handler()
+
+            # Run the synchronous function in another process
+            loop = asyncio.get_running_loop()
+            return await loop.run_in_executor(
+                process_pool_executor,
+                partial(
+                    handler,
+                    developer_id=developer_id,
+                    user_id=user_id,
+                    data=update_user_request,
+                ),
             )
 
         # Handle regular operations
         if asyncio.iscoroutinefunction(handler):
             return await handler(**arguments)
 
         # Run the synchronous function in another process
+        # FIXME: When the handler throws an exception, the process dies and the error is not captured. Instead it throws:
+        # "concurrent.futures.process.BrokenProcessPool: A process in the process pool was terminated abruptly while the future was running or pending."
         loop = asyncio.get_running_loop()
         return await loop.run_in_executor(process_pool_executor, partial(handler, **arguments))
     except BaseException as e:

agents-api/agents_api/activities/task_steps/prompt_step.py

@@ -172,6 +172,9 @@ async def prompt_step(context: StepContext) -> StepOutcome:
             for message in prompt
         ]
 
+    # Remove None values from passed_settings (avoid overwriting agent's settings)
+    passed_settings = {k: v for k, v in passed_settings.items() if v is not None}
+
     # Use litellm for other models
     completion_data: dict = {
         "model": agent_model,

agents-api/agents_api/activities/utils.py

@@ -466,10 +466,11 @@ def get_handler(system: SystemDef) -> Callable:
             return create_session_query
         case ("session", None, "update"):
             return update_session_query
-        case ("session", None, "delete"):
-            return delete_session_query
+        # TODO: Add support for create_or_update_session
         case ("session", None, "chat"):
             return chat
+        case ("session", None, "history"):
+            return get_history_query
 
         # TASKS
         case ("task", None, "list"):

agents-api/agents_api/autogen/Docs.py

@@ -14,11 +14,17 @@ class BaseDocSearchRequest(BaseModel):
         populate_by_name=True,
     )
     limit: Annotated[int, Field(ge=1, le=50)] = 10
+    """
+    The limit of documents to return
+    """
     lang: Literal["en-US"] = "en-US"
     """
     The language to be used for text-only search. Support for other languages coming soon.
     """
     metadata_filter: dict[str, Any] = {}
+    """
+    Metadata filter to apply to the search
+    """
     mmr_strength: Annotated[float, Field(ge=0.0, lt=1.0)] = 0
     """
     MMR Strength (mmr_strength = 1 - mmr_lambda)

agents-api/agents_api/autogen/Sessions.py

@@ -112,10 +112,41 @@ class RecallOptions(BaseModel):
         populate_by_name=True,
     )
     mode: Literal["hybrid", "vector", "text"] = "vector"
+    """
+    The mode to use for the search.
+    """
     num_search_messages: int = 4
+    """
+    The number of search messages to use for the search.
+    """
     max_query_length: int = 1000
-    hybrid_alpha: float = 0.7
-    confidence: float = 0.6
+    """
+    The maximum query length to use for the search.
+    """
+    alpha: Annotated[float, Field(ge=0.0, le=1.0)] = 0.7
+    """
+    The weight to apply to BM25 vs Vector search results. 0 => pure BM25; 1 => pure vector;
+    """
+    confidence: Annotated[float, Field(ge=0.0, le=1.0)] = 0.6
+    """
+    The confidence cutoff level
+    """
+    limit: Annotated[int, Field(ge=1, le=50)] = 10
+    """
+    The limit of documents to return
+    """
+    lang: Literal["en-US"] = "en-US"
+    """
+    The language to be used for text-only search. Support for other languages coming soon.
+    """
+    metadata_filter: dict[str, Any] = {}
+    """
+    Metadata filter to apply to the search
+    """
+    mmr_strength: Annotated[float, Field(ge=0.0, lt=1.0)] = 0
+    """
+    MMR Strength (mmr_strength = 1 - mmr_lambda)
+    """
 
 
 class RecallOptionsUpdate(RecallOptions):

agents-api/agents_api/common/protocol/tasks.py

@@ -106,7 +106,15 @@
         "finish_branch",
         "init_branch",
     ],
-    "finish_branch": ["wait", "error", "cancelled", "step", "finish", "init_branch"],
+    "finish_branch": [
+        "wait",
+        "error",
+        "cancelled",
+        "step",
+        "finish",
+        "init_branch",
+        "finish_branch",
+    ],
 }  # type: ignore
 
 valid_previous_statuses: dict[ExecutionStatus, list[ExecutionStatus]] = {