mirror of
https://github.com/microsoft/agent-framework.git
synced 2026-06-16 21:04:09 +08:00
westey
96d242fa7f
* Move token params from HarnessAgent constructor to options Remove the required maxContextWindowTokens and maxOutputTokens constructor parameters from HarnessAgent and AsHarnessAgent, replacing them with optional MaxContextWindowTokens and MaxOutputTokens properties on HarnessAgentOptions. When both values are provided, compaction is enabled as before (in-loop CompactionProvider and chat reducer on the default InMemoryChatHistory Provider). When either is null, compaction is disabled entirely, making it opt-in. New constructor: HarnessAgent(IChatClient, HarnessAgentOptions?, ILoggerFactory?, IServiceProvider?) Closes #6333 Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * Improving comments. * feat: Add custom CompactionStrategy and DisableCompaction to HarnessAgentOptions Allow users to provide their own CompactionStrategy via options, with a clear priority system: 1. DisableCompaction=true: no compaction regardless of other settings 2. Custom CompactionStrategy provided: use it (token params ignored) 3. Both MaxContextWindowTokens and MaxOutputTokens set: default strategy 4. Otherwise: no compaction Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> * fix: Address PR review comments on compaction opt-in - Update chatClient param XML doc to reflect compaction is opt-in - Strengthen compaction tests to assert ChatReducer is null/not-null rather than just asserting construction succeeds Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com> --------- Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
96d242fa7f
·
2026-06-09 13:06:00 +00:00
History
Harness Step 04 — Code Execution (Hyperlight + Skills)
This sample demonstrates a HarnessAgent with all features enabled, plus:
- Hyperlight CodeAct — sandboxed Python code execution via
execute_code(requires KVM) - Skills — file-based skill discovery (a
regex-testerskill is included)
The agent can plan tasks, manage modes, store memories, read/write files, search the web, approve sensitive operations, discover and use skills, and execute arbitrary Python code — all pre-configured by the HarnessAgent.
Prerequisites
- .NET 10 SDK
- An Azure AI Foundry project endpoint
- KVM-capable host (the Hyperlight sandbox runs code in micro-VMs)
Environment Variables
| Variable | Description |
|---|---|
AZURE_AI_PROJECT_ENDPOINT |
Your Azure AI Foundry project endpoint |
AZURE_AI_MODEL_DEPLOYMENT_NAME |
Model deployment name (default: gpt-5.4) |
Running
dotnet run
What to Try
- Regex testing: "Help me write a regex that matches valid email addresses, then test it against some examples."
- Code execution: "Calculate the first 20 prime numbers using the Sieve of Eratosthenes."
- Skill + code combo: "I need a regex for ISO 8601 dates — test it thoroughly with edge cases."
Included Skill
The skills/regex-tester/ skill instructs the agent to validate regex patterns by executing Python test code in the Hyperlight sandbox. It includes a regex cheatsheet as reference material.
Features Enabled
| Feature | Description |
|---|---|
| TodoProvider | Task planning and tracking (/todos command) |
| AgentModeProvider | Mode switching (/mode command) |
| FileMemoryProvider | Persistent memory stored as files |
| FileAccessProvider | Read/write files in a working directory |
| ToolApproval | Don't-ask-again approval for sensitive tools |
| WebSearch | Built-in hosted web search |
| AgentSkillsProvider | Discovers and uses skills from the skills/ folder |
| HyperlightCodeActProvider | Sandboxed Python execution via execute_code |
| OpenTelemetry | Trace logging to a text file |