glean-agent-toolkit 0.4.0__tar.gz → 0.6.0__tar.gz

{glean_agent_toolkit-0.4.0 → glean_agent_toolkit-0.6.0}/.cz.toml

@@ -1,8 +1,7 @@
 [tool.commitizen]
 name = "cz_conventional_commits"
-version = "0.4.0"
+version = "0.6.0"
 tag_format = "$version"
 version_files = [
   "pyproject.toml:version",
-  "glean/toolkit/__init__.py:__version__",
 ]

{glean_agent_toolkit-0.4.0 → glean_agent_toolkit-0.6.0}/.github/workflows/publish.yml

@@ -3,11 +3,30 @@ name: Publish to PyPI
 on:
   push:
     tags:
-      - '*.*.*'
+      - '[0-9]*.[0-9]*.[0-9]*'
   workflow_dispatch:
 
 jobs:
-  deploy:
+  test:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - name: Set up mise
+        uses: jdx/mise-action@v2
+        with:
+          cache: true
+        env:
+          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Initialize dev environment
+        run: mise run setup
+
+      - name: Run tests
+        run: mise run test
+
+  build-and-publish:
+    needs: test
     runs-on: ubuntu-latest
     environment: pypi
     permissions:
@@ -18,14 +37,14 @@ jobs:
     - uses: actions/checkout@v4
       with:
         fetch-depth: 0   # Needed for changelog generation
-    
+
     - name: Set up mise
       uses: jdx/mise-action@v2
       with:
         cache: true
       env:
         GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
-        
+
     - name: Build package
       run: uv build
 
@@ -34,10 +53,10 @@ jobs:
       run: |
         # Get latest tag for both push and manual triggers
         LATEST_TAG=$(git describe --tags --abbrev=0)
-        
+
         echo "LATEST_TAG=${LATEST_TAG}" >> $GITHUB_OUTPUT
         echo "CHANGELOG<<EOF" >> $GITHUB_OUTPUT
-        
+
         # Try to get previous tag
         if PREVIOUS_TAG=$(git describe --tags --abbrev=0 HEAD^ 2>/dev/null); then
           # If we have a previous tag, show changes between tags
@@ -47,12 +66,12 @@ jobs:
           uvx --from commitizen cz changelog --dry-run >> $GITHUB_OUTPUT
         fi
         echo "EOF" >> $GITHUB_OUTPUT
-      
+
     - name: Create GitHub Release
       uses: softprops/action-gh-release@v1
       with:
         tag_name: ${{ steps.CHANGELOG.outputs.LATEST_TAG }}
         body: ${{ steps.CHANGELOG.outputs.CHANGELOG }}
-        
+
     - name: Publish package to PyPI
       uses: pypa/gh-action-pypi-publish@release/v1

glean_agent_toolkit-0.6.0/CHANGELOG.md

@@ -0,0 +1,68 @@
+## 0.6.0 (2026-07-19)
+
+### Fix
+
+- **crewai**: pass args_schema to BaseTool so the LLM sees real parameters (#73)
+- **adk**: expose real typed signatures so declarations and invocation work (#77)
+- **openai**: sanitize strict schemas and isolate per-tool conversion failures (#76)
+- **adapters**: map anyOf/union and array item types correctly in get_field_type (#75)
+- **langchain**: use StructuredTool so converted tools are invocable (#74)
+- **tools**: stop closing shared Glean client on every tool call (#72)
+
+## 0.5.0 (2026-04-07)
+
+### Feat
+
+- add installable skills for SDK usage and tool building (#70)
+- search API alignment, chat tool, deprecation path, get_tools, async support, import docs (#66)
+- injectable GleanContext replaces hidden api_client() global (#62) (#65)
+- **deps**: add [all] extra combining all framework adapters (CHK-001) (#27)
+
+### Fix
+
+- correct web search tool name from 'Web Browser' to 'Gemini Web Search' (#71)
+- resolve remaining P2 eval items (CHK-111, CHK-115, CHK-118, CHK-119) (#68)
+- resolve eval checklist items — dedup, dead code, error handling, imports (#67)
+- namespace tool names and optimize descriptions for LLM consumption (#58)
+- structured error results and consistent adapter return types (#55)
+- depend on langchain-core instead of langchain to support LangGraph 1.x (#54)
+- Regenerate lockfile
+- align publish workflow tag trigger with commitizen tag format (#39)
+- use is not None checks in read_document validation (#37)
+- export Registry from top-level package (#36)
+- remove pydantic BaseModel from adapters public API (#35)
+- preserve float values in retry backoff config (CHK-002)
+- **release**: correct broken version_files path in .cz.toml (CHK-006) (#31)
+
+## 0.4.0 (2026-03-05)
+
+### Feat
+
+- support GLEAN_SERVER_URL with GLEAN_INSTANCE fallback (#26)
+- **api**: Add retry configuration to API client with environment variable controls (#15)
+- **dev**: adopt mise with full task parity and Node 22; move dev docs to CONTRIBUTING; README links to CONTRIBUTING (#12)
+
+### Fix
+
+- **adapters**: correct install hints to use glean-agent-toolkit extras
+
+### Refactor
+
+- rename glean_search to search across code, tests, docs, and snippets (#19)
+
+## 0.3.0 (2025-07-21)
+
+### Feat
+
+- Add enhanced parameter schemas with Field metadata support
+
+### Fix
+
+- Adds newline to .cz.toml
+- Fixes type errors during task:lint
+
+## 0.2.0 (2025-06-05)
+
+### Feat
+
+- Implements Agent Builder tools as defaults

{glean_agent_toolkit-0.4.0 → glean_agent_toolkit-0.6.0}/CONTRIBUTING.md

@@ -14,8 +14,7 @@ src/
 │     ├─ registry.py
 │     ├─ spec.py
 │     ├─ adapters/
-│     ├─ tools/
-│     └─ cli.py
+│     └─ tools/
 ├─ tests/
 └─ docs/
 ```

glean_agent_toolkit-0.4.0/README.md → glean_agent_toolkit-0.6.0/PKG-INFO

@@ -1,3 +1,53 @@
+Metadata-Version: 2.4
+Name: glean-agent-toolkit
+Version: 0.6.0
+Summary: AI Agent toolkit for adapting Glean's enterprise tools across multiple frameworks
+Project-URL: Source Code, https://github.com/gleanwork/glean-agent-toolkit
+Author-email: Steve Calvert <steve.calvert@glean.com>
+License: MIT
+License-File: LICENSE
+Requires-Python: <4.0,>=3.10
+Requires-Dist: glean-api-client<1.0,>=0.6.0
+Requires-Dist: pydantic<3.0,>=2.7
+Provides-Extra: adk
+Requires-Dist: google-adk<2.0,>=1.28.1; extra == 'adk'
+Provides-Extra: all
+Requires-Dist: crewai<1.0,>=0.100; extra == 'all'
+Requires-Dist: google-adk<2.0,>=1.28.1; extra == 'all'
+Requires-Dist: langchain-core<2.0,>=0.3; extra == 'all'
+Requires-Dist: openai-agents<1.0,>=0.3; extra == 'all'
+Requires-Dist: openai<2.0,>=1.0.0; extra == 'all'
+Provides-Extra: codespell
+Requires-Dist: codespell<3.0,>=2.2.6; extra == 'codespell'
+Provides-Extra: crewai
+Requires-Dist: crewai<1.0,>=0.100; extra == 'crewai'
+Provides-Extra: dev
+Requires-Dist: commitizen<5.0,>=4.4.1; extra == 'dev'
+Requires-Dist: pip-audit<3.0,>=2.6.0; extra == 'dev'
+Requires-Dist: pre-commit<4.0,>=3.6.0; extra == 'dev'
+Requires-Dist: vcrpy<7.0,>=6.0.2; extra == 'dev'
+Provides-Extra: langchain
+Requires-Dist: langchain-core<2.0,>=0.3; extra == 'langchain'
+Provides-Extra: lint
+Requires-Dist: ruff<1.0,>=0.5; extra == 'lint'
+Provides-Extra: openai
+Requires-Dist: openai-agents<1.0,>=0.3; extra == 'openai'
+Requires-Dist: openai<2.0,>=1.0.0; extra == 'openai'
+Provides-Extra: test
+Requires-Dist: crewai<1.0,>=0.100; extra == 'test'
+Requires-Dist: google-adk<2.0,>=1.28.1; extra == 'test'
+Requires-Dist: langchain-core<2.0,>=0.3; extra == 'test'
+Requires-Dist: openai-agents<1.0,>=0.3; extra == 'test'
+Requires-Dist: openai<2.0,>=1.0.0; extra == 'test'
+Requires-Dist: pytest-asyncio<1.0,>=0.23.2; extra == 'test'
+Requires-Dist: pytest-cov<6.0,>=4.1.0; extra == 'test'
+Requires-Dist: pytest-httpx<1.0,>=0.35.0; extra == 'test'
+Requires-Dist: pytest<9.0,>=7.4.3; extra == 'test'
+Requires-Dist: vcrpy<7.0,>=6.0.2; extra == 'test'
+Provides-Extra: typing
+Requires-Dist: pyright<2.0,>=1.1.370; extra == 'typing'
+Description-Content-Type: text/markdown
+
 # Glean Agent Toolkit
 
 The Glean Agent Toolkit makes it easy to integrate Glean's powerful search and knowledge discovery capabilities into your AI agents. Use our pre-built tools with popular agent frameworks like OpenAI Assistants, LangChain, CrewAI, and Google's Agent Development Kit (ADK), or adapt your own custom tools for cross-framework use.
@@ -16,13 +66,12 @@ The Glean Agent Toolkit makes it easy to integrate Glean's powerful search and k
 
 ### Retry configuration (env vars)
 
-| Variable                    | Default | Description                    | Example |
-| --------------------------- | ------- | ------------------------------ | ------- |
-| `GLEAN_RETRY_INITIAL`       | `1.0`   | Initial backoff in seconds     | `0.5`   |
-| `GLEAN_RETRY_MAX`           | `50.0`  | Maximum backoff in seconds     | `8`     |
-| `GLEAN_RETRY_MULTIPLIER`    | `1.1`   | Exponential backoff multiplier | `2.0`   |
-| `GLEAN_RETRY_JITTER_MS`     | `100`   | Random jitter in milliseconds  | `250`   |
-| `GLEAN_RETRY_ON_RATE_LIMIT` | `true`  | Retry on HTTP 429 rate limits  | `true`  |
+| Variable                  | Default | Description                                              | Example |
+| ------------------------- | ------- | -------------------------------------------------------- | ------- |
+| `GLEAN_RETRY_INITIAL`     | `1.0`   | Initial backoff interval in seconds                      | `0.5`   |
+| `GLEAN_RETRY_MAX`         | `50.0`  | Maximum backoff interval in seconds                      | `8`     |
+| `GLEAN_RETRY_MULTIPLIER`  | `1.1`   | Backoff multiplier/exponent                              | `2.0`   |
+| `GLEAN_RETRY_MAX_ELAPSED` | `60.0`  | Total time limit in seconds before giving up on retries  | `30.0`  |
 
 Retries cover transient failures such as HTTP 429/5xx and connection timeouts. Set these before constructing any Glean client usage.
 
@@ -31,19 +80,32 @@ Retries cover transient failures such as HTTP 429/5xx and connection timeouts. S
 export GLEAN_RETRY_INITIAL=0.5
 export GLEAN_RETRY_MAX=8
 export GLEAN_RETRY_MULTIPLIER=2.0
-export GLEAN_RETRY_JITTER_MS=250
-export GLEAN_RETRY_ON_RATE_LIMIT=true
+export GLEAN_RETRY_MAX_ELAPSED=30.0
 ```
 
 ## Installation
 
-Install the base toolkit:
+Install with all framework adapters (recommended):
+
+```bash snippet=readme/snippet-03.bash
+pip install "glean-agent-toolkit[all]"
+```
+
+Or install the base toolkit and add extras as needed:
 
 ```bash snippet=readme/snippet-01.bash
 pip install glean-agent-toolkit
 ```
 
-To include support for specific agent frameworks, install the relevant extras:
+### Extras reference
+
+| Extra        | Installs                      | Use case                         |
+| ------------ | ----------------------------- | -------------------------------- |
+| `[openai]`   | `openai`, `openai-agents`     | OpenAI Agents SDK / Assistants   |
+| `[langchain]`| `langchain-core`              | LangChain / LangGraph agents     |
+| `[crewai]`   | `crewai`                      | CrewAI multi-agent workflows     |
+| `[adk]`      | `google-adk`                  | Google Agent Development Kit     |
+| `[all]`      | All of the above              | Full framework support           |
 
 ```bash snippet=readme/snippet-02.bash
 pip install glean-agent-toolkit[openai]
@@ -52,12 +114,6 @@ pip install glean-agent-toolkit[langchain]
 pip install glean-agent-toolkit[crewai]
 ```
 
-You can also install all extras:
-
-```bash snippet=readme/snippet-03.bash
-pip install glean-agent-toolkit[all]
-```
-
 Note: The `[openai]` extra installs the standard `openai` Python library, used for direct API interactions like Chat Completions or the Assistants API. The example below for the "OpenAI Agents SDK" uses a separate library, `openai-agents`, which you'll need to install independently: `pip install openai-agents`.
 
 ## Prerequisites
@@ -82,9 +138,9 @@ Here's a complete example that demonstrates the power of the Glean Agent Toolkit
 
 First, create the project structure:
 
-```bash snippet=readme/snippet-04.bash
-export GLEAN_API_TOKEN="your-api-token"
-export GLEAN_SERVER_URL="https://your-company-be.glean.com"
+```bash snippet=readme/snippet-06.bash
+mkdir company_assistant/
+cd company_assistant/
 ```
 
 ### Step 2: Create the Agent File
@@ -124,10 +180,10 @@ root_agent = Agent(
     instruction="""You are a helpful company assistant that helps employees find information,
     people, and resources within the organization. You have access to:
 
-    - Company knowledge base and documents (use search)
-    - Employee directory and contact information (use employee_search)
-    - Calendar and meeting information (use calendar_search)
-    - Email search capabilities (use gmail_search)
+    - Company knowledge base and documents (use glean_search)
+    - Employee directory and contact information (use glean_employee_search)
+    - Calendar and meeting information (use glean_calendar_search)
+    - Email search capabilities (use glean_gmail_search)
 
     Always be helpful, professional, and respect privacy. When searching for people,
     only share appropriate business contact information.""",
@@ -147,18 +203,26 @@ from . import agent
 
 Create `company_assistant/.env` with your credentials:
 
-```bash snippet=readme/snippet-05.bash
-export GLEAN_API_TOKEN="your-api-token"
-export GLEAN_SERVER_URL="https://your-company-be.glean.com"
+```bash snippet=readme/snippet-07.bash
+# company_assistant/.env
+
+# Authentication for Google ADK (choose one)
+GOOGLE_API_KEY=your-google-ai-studio-api-key
+# OR for Vertex AI:
+# GOOGLE_CLOUD_PROJECT=your-project-id
+# GOOGLE_CLOUD_LOCATION=us-central1
+
+# Glean credentials
+GLEAN_API_TOKEN=your-glean-api-token
+GLEAN_SERVER_URL=https://your-company-be.glean.com
 ```
 
 ### Step 5: Run Your Agent
 
 From the parent directory (outside `company_assistant/`), run your Company Assistant:
 
-```bash snippet=readme/snippet-06.bash
-mkdir company_assistant/
-cd company_assistant/
+```bash
+adk web
 ```
 
 ### Real-World Queries You Can Handle
@@ -176,17 +240,46 @@ This type of assistant can dramatically improve employee productivity by making
 
 ## Available Tools
 
-The toolkit comes with a suite of production-ready tools that connect to various Glean functionalities:
+| Tool name                | Import name         | Description                                    |
+| ------------------------ | ------------------- | ---------------------------------------------- |
+| `glean_search`           | `search`            | Search internal documents and knowledge bases  |
+| `glean_chat`             | `glean_chat`        | Conversational Q&A with Glean Assistant        |
+| `glean_read_document`    | `read_document`     | Read full document content by ID or URL        |
+| `glean_web_search`       | `web_search`        | Search the public web for external information |
+| `glean_calendar_search`  | `calendar_search`   | Find meetings and calendar events              |
+| `glean_employee_search`  | `employee_search`   | Search employees by name, team, or department  |
+| `glean_code_search`      | `code_search`       | Search source code repositories                |
+| `glean_gmail_search`     | `gmail_search`      | Search Gmail messages and conversations        |
+| `glean_outlook_search`   | `outlook_search`    | Search Outlook mail and calendar items         |
+
+### Explicit imports
+
+```python
+from glean.agent_toolkit.tools import search, glean_chat, read_document
+from glean.agent_toolkit.tools import web_search, calendar_search
+from glean.agent_toolkit.tools import employee_search, code_search
+from glean.agent_toolkit.tools import gmail_search, outlook_search
+```
+
+### Adapter methods
+
+Each tool function exposes adapter methods for framework conversion:
+
+| Method                | Returns                          | Framework           |
+| --------------------- | -------------------------------- | ------------------- |
+| `.as_openai_tool()`   | `FunctionTool` or `dict`         | OpenAI Agents SDK   |
+| `.as_langchain_tool()` | `langchain_core.tools.Tool`     | LangChain / LangGraph |
+| `.as_crewai_tool()`   | `CrewAI BaseTool`                | CrewAI              |
+| `.as_adk_tool()`      | `google.adk FunctionTool`        | Google ADK          |
+
+You can also use `get_tools()` to get all tools adapted for a framework at once:
 
-- `search`: Search your company's knowledge base for relevant documents and information
-- **`read_document`**: Read full content of a specific document by ID or URL (requires
-  config settings `queryapi.getDocuments.enabled` and `queryapi.getDocuments.content.enabled`)
-- **`web_search`**: Search the public web for up-to-date external information
-- `calendar_search`: Find meetings and calendar events
-- `employee_search`: Search for employees by name, team, department, or expertise
-- `code_search`: Search your company's source code repositories
-- `gmail_search`: Search Gmail messages and conversations
-- `outlook_search`: Search Outlook mail and calendar items
+```python
+from glean.agent_toolkit import get_tools
+
+langchain_tools = get_tools("langchain")
+openai_tools = get_tools("openai", include=["glean_search", "glean_chat"])
+```
 
 ### Web Research with Context
 
@@ -238,6 +331,8 @@ print(f"Search results: {result.final_output}")
 ```python snippet=readme/snippet-04.py
 import os
 
+# NOTE: AgentExecutor requires the full `langchain` package (not just langchain-core).
+# Install with: pip install langchain
 from langchain.agents import AgentExecutor, create_react_agent
 from langchain_core.prompts import ChatPromptTemplate
 from langchain_openai import ChatOpenAI
@@ -412,6 +507,32 @@ langchain_weather = get_weather.as_langchain_tool()
 crewai_weather = get_weather.as_crewai_tool()
 ```
 
+## Agent Skills
+
+The `skills/` directory contains [Agent Skills](https://agentskills.io) — structured instructions that teach AI coding agents how to use the Glean Agent Toolkit effectively. Skills are supported by Claude Code, Cursor, GitHub Copilot, VS Code, Gemini CLI, OpenAI Codex, Goose, Amp, Roo Code, Junie, and [many others](https://agentskills.io).
+
+### Install
+
+Use [`npx skills`](https://github.com/agentskills/agentskills) to install into your agent:
+
+```bash
+# Install all skills at once
+npx skills add https://github.com/gleanwork/glean-agent-toolkit
+```
+
+```bash
+# Or install individual skills
+npx skills add https://github.com/gleanwork/glean-agent-toolkit/tree/main/skills/glean-agent-toolkit-guide
+npx skills add https://github.com/gleanwork/glean-agent-toolkit/tree/main/skills/glean-agent-toolkit-builder
+```
+
+### Available Skills
+
+| Skill | Description |
+| --- | --- |
+| `glean-agent-toolkit-guide` | How to use the SDK: `get_tools()`, `GleanContext`, adapters, error handling, async |
+| `glean-agent-toolkit-builder` | How to create custom tools with `@tool_spec` |
+
 ## Contributing
 
 Interested in contributing? Check out our [Contributing Guide](CONTRIBUTING.md) for instructions on setting up the development environment and submitting changes.