.NET: Python/.Net: Agent Harness blog post accompanying samples part 3#6741
.NET: Python/.Net: Agent Harness blog post accompanying samples part 3#6741westey-m wants to merge 5 commits into
Conversation
moonbox3
added
documentation
github-actions
Bot
changed the title
github-actions
Bot
changed the title
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
There was a problem hiding this comment.
Automated Code Review
Reviewers: 5 | Confidence: 84%
✓ Correctness
This PR adds .NET and Python samples for a blog post (Part 3) on scaling agent capabilities. The code is well-structured and correct for a demo. One low-severity edge case exists in the valuation script where a zero price input would cause a ZeroDivisionError, though in practice the mock data always provides positive prices. No high-severity bugs found.
✓ Security Reliability
This PR adds sample code for a blog post demonstrating agent harness capabilities. The security posture is generally good: process arguments use safe list-based APIs (avoiding injection), sensitive operations are gated behind approval prompts, the shell executor has a timeout + deny-list + confined working directory, and security boundaries are clearly documented. The one notable gap is that the .NET SubprocessScriptRunner lacks a built-in process timeout (unlike its Python counterpart which has timeout=30), relying entirely on the caller-supplied CancellationToken.
✓ Test Coverage
This PR adds sample code (both .NET and Python) accompanying a blog post (Part 3 of 'Build your own claw' series). The PR introduces no library/production code — only sample applications and their supporting scripts. Following the established repository convention, harness/claw samples do not have associated unit tests (the existing Step01 sample likewise has no tests). The Python utility scripts (risk_score.py, valuation_metrics.py, subprocess_script_runner.py) are self-contained CLI tools intended as demos; while they would benefit from basic tests, the repo's existing pattern does not require them for sample code. No test coverage concerns rise to blocking level.
✓ Failure Modes
This PR adds sample code for a blog post demonstrating agent harness scaling capabilities. The code is well-structured for a sample, but there is one concrete resource leak on a failure path in the .NET FoundrySkills helper: if McpClient.CreateAsync throws (e.g., network error connecting to the Foundry Toolbox), the HttpClient created just before it is never disposed because the tuple return never executes and the caller's finally block only disposes the null variable.
✓ Design Approach
I found one blocking design issue in the Python Part 3 sample: it documents and instructs portfolio-based workflows, but the sample assets checked in under
working/only include trade confirmations, not theportfolio.csvfile those workflows require. That leaves one of the sample’s core demo paths broken out of the box.
Automated review by westey-m's agents
There was a problem hiding this comment.
Pull request overview
Adds “Part 3 — Scaling capabilities” harness samples for both Python and .NET, expanding the “Build your own claw” series with skills, a confined shell, CodeAct, background agents, and optional Foundry Toolbox MCP skills integration.
Changes:
- Added Python harness Part 3 runnable sample with local file-based skills (valuation + risk scoring), optional Foundry skills via Toolbox MCP, a confined shell tool, Monty CodeAct, and a background web-search research agent.
- Added .NET harness Part 3 runnable sample with equivalent capabilities (file skills + optional Foundry skills, confined shell executor, Hyperlight CodeAct, background research agent).
- Added sample working data (portfolio + trade confirmations) and updated sample readmes/indices to reference Part 3.
Reviewed changes
Copilot reviewed 38 out of 38 changed files in this pull request and generated 5 comments.
Show a summary per file
| File | Description |
|---|---|
| python/samples/02-agents/harness/build_your_own_claw/working/confirmations/trade confirmation 1.txt | Sample trade confirmation fixture for shell tidying demo |
| python/samples/02-agents/harness/build_your_own_claw/working/confirmations/SPY sell.txt | Sample trade confirmation fixture for shell tidying demo |
| python/samples/02-agents/harness/build_your_own_claw/working/confirmations/googl-jan.txt | Sample trade confirmation fixture for shell tidying demo |
| python/samples/02-agents/harness/build_your_own_claw/working/confirmations/copy of trade (3).txt | Sample trade confirmation fixture for shell tidying demo |
| python/samples/02-agents/harness/build_your_own_claw/working/confirmations/conf_AAPL.txt | Sample trade confirmation fixture for shell tidying demo |
| python/samples/02-agents/harness/build_your_own_claw/working/confirmations/2025-06-21_nvda.txt | Sample trade confirmation fixture for shell tidying demo |
| python/samples/02-agents/harness/build_your_own_claw/subprocess_script_runner.py | Python subprocess runner for file-based skill scripts |
| python/samples/02-agents/harness/build_your_own_claw/skills/valuation/SKILL.md | Valuation skill definition/instructions (Python) |
| python/samples/02-agents/harness/build_your_own_claw/skills/valuation/scripts/valuation_metrics.py | Valuation computation script (Python) |
| python/samples/02-agents/harness/build_your_own_claw/skills/valuation/references/valuation-guide.md | Valuation reference guide (Python) |
| python/samples/02-agents/harness/build_your_own_claw/skills/risk-scoring/SKILL.md | Risk-scoring skill definition/instructions (Python) |
| python/samples/02-agents/harness/build_your_own_claw/skills/risk-scoring/scripts/risk_score.py | Concentration risk scoring script (Python) |
| python/samples/02-agents/harness/build_your_own_claw/skills/risk-scoring/references/risk-bands.md | Risk band reference guide (Python) |
| python/samples/02-agents/harness/build_your_own_claw/README.md | Documents Part 3 sample and how to run/what to expect |
| python/samples/02-agents/harness/build_your_own_claw/claw_step03_scaling_capabilities.py | Main Python Part 3 runnable sample (skills/shell/codeact/background agents) |
| dotnet/samples/02-agents/Harness/README.md | Adds .NET Part 3 sample entry to harness samples index |
| dotnet/samples/02-agents/Harness/BuildYourOwnClaw/Claw_Step03_ScalingCapabilities/working/portfolio.csv | Sample portfolio fixture for .NET Part 3 |
| dotnet/samples/02-agents/Harness/BuildYourOwnClaw/Claw_Step03_ScalingCapabilities/working/confirmations/trade confirmation 1.txt | Sample trade confirmation fixture for .NET Part 3 |
| dotnet/samples/02-agents/Harness/BuildYourOwnClaw/Claw_Step03_ScalingCapabilities/working/confirmations/SPY sell.txt | Sample trade confirmation fixture for .NET Part 3 |
| dotnet/samples/02-agents/Harness/BuildYourOwnClaw/Claw_Step03_ScalingCapabilities/working/confirmations/googl-jan.txt | Sample trade confirmation fixture for .NET Part 3 |
| dotnet/samples/02-agents/Harness/BuildYourOwnClaw/Claw_Step03_ScalingCapabilities/working/confirmations/copy of trade (3).txt | Sample trade confirmation fixture for .NET Part 3 |
| dotnet/samples/02-agents/Harness/BuildYourOwnClaw/Claw_Step03_ScalingCapabilities/working/confirmations/conf_AAPL.txt | Sample trade confirmation fixture for .NET Part 3 |
| dotnet/samples/02-agents/Harness/BuildYourOwnClaw/Claw_Step03_ScalingCapabilities/working/confirmations/2025-06-21_nvda.txt | Sample trade confirmation fixture for .NET Part 3 |
| dotnet/samples/02-agents/Harness/BuildYourOwnClaw/Claw_Step03_ScalingCapabilities/TradingTools.cs | Approval-required “place_trade” tool wrapper (.NET) |
| dotnet/samples/02-agents/Harness/BuildYourOwnClaw/Claw_Step03_ScalingCapabilities/SubprocessScriptRunner.cs | .NET subprocess runner for file-based skill scripts |
| dotnet/samples/02-agents/Harness/BuildYourOwnClaw/Claw_Step03_ScalingCapabilities/StockTools.cs | Mock stock quote tool providing price + trailing EPS (.NET) |
| dotnet/samples/02-agents/Harness/BuildYourOwnClaw/Claw_Step03_ScalingCapabilities/skills/valuation/SKILL.md | Valuation skill definition/instructions (.NET sample content) |
| dotnet/samples/02-agents/Harness/BuildYourOwnClaw/Claw_Step03_ScalingCapabilities/skills/valuation/scripts/valuation_metrics.py | Valuation computation script (packaged with .NET sample) |
| dotnet/samples/02-agents/Harness/BuildYourOwnClaw/Claw_Step03_ScalingCapabilities/skills/valuation/references/valuation-guide.md | Valuation reference guide (.NET sample content) |
| dotnet/samples/02-agents/Harness/BuildYourOwnClaw/Claw_Step03_ScalingCapabilities/skills/risk-scoring/SKILL.md | Risk-scoring skill definition/instructions (.NET sample content) |
| dotnet/samples/02-agents/Harness/BuildYourOwnClaw/Claw_Step03_ScalingCapabilities/skills/risk-scoring/scripts/risk_score.py | Risk scoring script (packaged with .NET sample) |
| dotnet/samples/02-agents/Harness/BuildYourOwnClaw/Claw_Step03_ScalingCapabilities/skills/risk-scoring/references/risk-bands.md | Risk band reference guide (.NET sample content) |
| dotnet/samples/02-agents/Harness/BuildYourOwnClaw/Claw_Step03_ScalingCapabilities/ResearchAgent.cs | Background web-search-only research sub-agent builder (.NET) |
| dotnet/samples/02-agents/Harness/BuildYourOwnClaw/Claw_Step03_ScalingCapabilities/README.md | .NET Part 3 sample documentation (run instructions + features) |
| dotnet/samples/02-agents/Harness/BuildYourOwnClaw/Claw_Step03_ScalingCapabilities/Program.cs | .NET Part 3 runnable sample wiring skills/shell/codeact/background agents |
| dotnet/samples/02-agents/Harness/BuildYourOwnClaw/Claw_Step03_ScalingCapabilities/FoundrySkills.cs | Helper to connect to Foundry Toolbox MCP for centrally-managed skills |
| dotnet/samples/02-agents/Harness/BuildYourOwnClaw/Claw_Step03_ScalingCapabilities/Claw_Step03_ScalingCapabilities.csproj | New .NET sample project + content copying of skills/working data |
| dotnet/agent-framework-dotnet.slnx | Adds the new .NET sample project to the solution |
Motivation & Context
#6361
Description & Review Guide
Related Issue
Fixes #
Contribution Checklist
breaking changelabel (or add "[BREAKING]" to the title prefix, before or after any language prefix) — a workflow keeps the label and title prefix in sync automatically.