Skip to content
Merged
Changes from 1 commit
Commits
Show all changes
36 commits
Select commit Hold shift + click to select a range
c2e0ae4
feat(dotnet): add Microsoft.Agents.AI.Tools.Shell with LocalShellTool
Apr 30, 2026
c397780
feat(shell): close Python parity gaps for LocalShellTool
Apr 30, 2026
8255bbc
feat(shell): add DockerShellTool sandboxed shell tier
Apr 30, 2026
7bc0932
test(shell): add DockerShellTool integration tests
Apr 30, 2026
ab3c0c9
style(shell): apply dotnet format pass
Apr 30, 2026
ad3fd51
docs(shell): add DockerShellTool walkthrough with sequence diagrams
Apr 30, 2026
d7e5707
PR 5604 review fixes (group a): libc DllImport, namespace cleanup, po…
May 1, 2026
3ac13b2
PR 5604 review fix (group b): add ShellKind.Sh for /bin/sh fallback
May 1, 2026
b74ea0b
PR 5604 review fix (group d): honor timeout=null, add DefaultTimeout
May 1, 2026
1579a83
PR 5604 review fix (group e): smart requireApproval default for Docke…
May 1, 2026
0022bb9
PR 5604 review fix (group c): wrap POSIX shell in setsid for correct …
May 1, 2026
b40ed60
.Net: DockerShellTool design + caller-cancel container leak fixes (PR…
May 1, 2026
6b30f56
.Net: Fill PR #5604 test coverage gaps for Shell tools
May 1, 2026
cb1ac20
feat(dotnet/shell): add ShellEnvironmentProvider for OS-aware shell i…
May 1, 2026
83b1270
fix(dotnet/shell): address PR review feedback round 3
May 1, 2026
4ce489f
Address PR #5604 round 4 review feedback
May 4, 2026
fe041a4
Add Async suffix to async test methods to satisfy IDE1006
May 4, 2026
3a0a513
Fix CPU busy-spin in WaitForSentinelAsync
May 4, 2026
3bfd332
Remove unused onCommand audit hook from shell tools
May 4, 2026
4972f2c
Align Shell csproj with Foundry.Hosting preview-package conventions
May 4, 2026
d5d2e25
Document why ShellEnvironmentProvider uses Instructions, not Messages
May 4, 2026
b55e86a
Clarify which probe failures ShellEnvironmentProvider swallows
May 4, 2026
5339a80
Strip cross-language and bug-history narrative from shell tool comments
May 4, 2026
f329604
Address PR #5604 round 5 review feedback
May 4, 2026
2a0af41
Address PR #5604 round 6 review feedback
May 4, 2026
ede4d33
Address PR #5604 round 7 review feedback
May 5, 2026
dfa1fa1
Address PR #5604 round 8 review feedback
May 5, 2026
1d09869
Address PR #5604 round 9 review feedback
May 5, 2026
7618e54
Address PR #5604 round 10 review feedback
May 6, 2026
94acaca
fix(dotnet): address PR #5604 round-3 review feedback
May 7, 2026
1351c55
fix(dotnet): address PR #5604 round-3 follow-up nits
May 7, 2026
106c2b1
Refactor shell tool: abstract ShellExecutor, options classes, Contain…
May 7, 2026
4f702df
Remove IsHardenedConfiguration; AsAIFunction defaults to approval-gated
May 7, 2026
2022da7
Replace ShellExecutionException/ShellTimeoutException with standard e…
May 8, 2026
ff7d8e4
Remove ShellPolicy.DefaultDenyList; default policy is empty
May 11, 2026
ba0b137
Document single-session ownership for persistent shell mode
May 11, 2026
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
Prev Previous commit
Next Next commit
test(shell): add DockerShellTool integration tests
Adds 9 end-to-end tests that exercise DockerShellTool against a live
Docker (or Podman) daemon. Tests are tagged [Trait("Category",
"Integration")] and auto-skip via Assert.Skip when no daemon is
available, so they are CI-safe.

Coverage:
- IsAvailableAsync probe
- Persistent mode basic command + state preservation across calls
- --network none blocks outbound DNS
- --read-only root prevents writes outside /tmp; /tmp tmpfs is writable
- --user 65534:65534 (nobody) is in effect
- Stateless mode: env vars do not leak across calls
- HostWorkdir bind-mount + read-only enforcement
- Environment variables passed via -e

Tests use debian:stable-slim (alpine ships only busybox sh, which
ShellSession persistent bash REPL cannot drive).

Run locally:
  dotnet test --filter "Category=Integration"
or filter by class on the test exe directly.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
  • Loading branch information
alliscode and Copilot committed Apr 30, 2026
commit 7bc09329d7e2150bcac5dd8798307c16ed6896ab
Original file line number Diff line number Diff line change
@@ -0,0 +1,195 @@
// Copyright (c) Microsoft. All rights reserved.

using System;
using System.Collections.Generic;
using System.IO;
using System.Threading.Tasks;
using Microsoft.Agents.AI.Tools.Shell;

namespace Microsoft.Agents.AI.Tools.Shell.UnitTests;

/// <summary>
/// End-to-end tests that exercise <see cref="DockerShellTool"/> against a live
/// Docker (or Podman) daemon. Tests auto-skip when no daemon is available, so
/// they're safe to run in CI.
/// </summary>
/// <remarks>
/// To run only these tests locally:
/// <code>
/// dotnet test --filter "Category=Integration&amp;FullyQualifiedName~DockerShellToolIntegrationTests"
/// </code>
/// or run the test exe directly with the trait filter.
/// </remarks>
[Trait("Category", "Integration")]
public sealed class DockerShellToolIntegrationTests
{
// Small, fast image that has bash. Pulled lazily on first run.
// Alpine ships only busybox sh, which the persistent shell session can't use.
private const string TestImage = "debian:stable-slim";

private static async Task<bool> EnsureDockerOrSkipAsync()
{
if (!await DockerShellTool.IsAvailableAsync().ConfigureAwait(false))
{
Assert.Skip("Docker (or Podman) daemon is not available on this machine.");
return false; // unreachable
}
return true;
}

[Fact]
public async Task IsAvailableAsync_ReturnsTrue_WhenDaemonRunning()
{
await EnsureDockerOrSkipAsync();
Assert.True(await DockerShellTool.IsAvailableAsync());
}

[Fact]
public async Task Persistent_RunsBasicCommand()
{
await EnsureDockerOrSkipAsync();

await using var tool = new DockerShellTool(image: TestImage, mode: ShellMode.Persistent);
await tool.StartAsync();

var result = await tool.RunAsync("echo hello-from-docker");

Assert.Equal(0, result.ExitCode);
Assert.Contains("hello-from-docker", result.Stdout);
}

[Fact]
public async Task Persistent_PreservesStateAcrossCalls()
{
await EnsureDockerOrSkipAsync();

await using var tool = new DockerShellTool(image: TestImage, mode: ShellMode.Persistent);
await tool.StartAsync();

var set = await tool.RunAsync("export DEMO=persisted-12345");
Assert.Equal(0, set.ExitCode);

var get = await tool.RunAsync("echo $DEMO");
Assert.Equal(0, get.ExitCode);
Assert.Contains("persisted-12345", get.Stdout);
}

[Fact]
public async Task NetworkNone_BlocksOutboundConnections()
{
await EnsureDockerOrSkipAsync();

await using var tool = new DockerShellTool(image: TestImage, mode: ShellMode.Persistent /* network defaults to "none" */);
await tool.StartAsync();

// Try to resolve a hostname; with --network none, even DNS should fail.
// Use getent (always present on debian) so we don't depend on optional tools.
var result = await tool.RunAsync("getent hosts example.com 2>&1; echo MARKER:$?");

Assert.Contains("MARKER:", result.Stdout);
// Non-zero status from wget proves network was blocked.
Assert.DoesNotContain("MARKER:0", result.Stdout);
}

[Fact]
public async Task ReadOnlyRoot_PreventsWritesOutsideTmp()
{
await EnsureDockerOrSkipAsync();

await using var tool = new DockerShellTool(image: TestImage, mode: ShellMode.Persistent);
await tool.StartAsync();

var rootWrite = await tool.RunAsync("touch /should-not-exist 2>&1; echo CODE:$?");
Assert.Contains("CODE:", rootWrite.Stdout);
Assert.DoesNotContain("CODE:0", rootWrite.Stdout);

var tmpWrite = await tool.RunAsync("touch /tmp/ok && echo TMP_OK");
Assert.Equal(0, tmpWrite.ExitCode);
Assert.Contains("TMP_OK", tmpWrite.Stdout);
}

[Fact]
public async Task NonRootUser_RunsAsNobody()
{
await EnsureDockerOrSkipAsync();

await using var tool = new DockerShellTool(image: TestImage, mode: ShellMode.Persistent);
await tool.StartAsync();

var result = await tool.RunAsync("id -u");

Assert.Equal(0, result.ExitCode);
// Default user is 65534:65534
Assert.Contains("65534", result.Stdout);
}

[Fact]
public async Task Stateless_RunsEachCommandInFreshContainer()
{
await EnsureDockerOrSkipAsync();

await using var tool = new DockerShellTool(image: TestImage, mode: ShellMode.Stateless);

var first = await tool.RunAsync("echo first; export STATE=set");
Assert.Equal(0, first.ExitCode);
Assert.Contains("first", first.Stdout);

// Stateless: env var must NOT survive
var second = await tool.RunAsync("echo \"second:[${STATE:-unset}]\"");
Assert.Equal(0, second.ExitCode);
Assert.Contains("second:[unset]", second.Stdout);
}

[Fact]
public async Task HostWorkdir_MountsAndIsReadOnlyByDefault()
{
await EnsureDockerOrSkipAsync();

var hostDir = Path.Combine(Path.GetTempPath(), "af-docker-shell-it-" + Guid.NewGuid().ToString("N")[..8]);
Directory.CreateDirectory(hostDir);
var sentinel = Path.Combine(hostDir, "from-host.txt");
await File.WriteAllTextAsync(sentinel, "host-content");

try
{
await using var tool = new DockerShellTool(
image: TestImage,
mode: ShellMode.Persistent,
hostWorkdir: hostDir,
mountReadonly: true);
await tool.StartAsync();

var read = await tool.RunAsync("cat /workspace/from-host.txt");
Assert.Equal(0, read.ExitCode);
Assert.Contains("host-content", read.Stdout);

// Read-only mount: write must fail
var write = await tool.RunAsync("echo bad > /workspace/should-fail 2>&1; echo CODE:$?");
Assert.DoesNotContain("CODE:0", write.Stdout);
}
finally
{
try { Directory.Delete(hostDir, recursive: true); } catch { /* best-effort cleanup */ }
}
}

[Fact]
public async Task EnvironmentVariables_ArePassedThrough()
{
await EnsureDockerOrSkipAsync();

await using var tool = new DockerShellTool(
image: TestImage,
mode: ShellMode.Persistent,
environment: new Dictionary<string, string>
{
["INJECTED_VAR"] = "injected-value-7777",
});
await tool.StartAsync();

var result = await tool.RunAsync("echo $INJECTED_VAR");

Assert.Equal(0, result.ExitCode);
Assert.Contains("injected-value-7777", result.Stdout);
}
}