agentleFS
hjdspacehjdspace/SpaceCode/.agents/skills/dart-build-cli-app/SKILL.md

dart-build-cli-app

Skill4 starsChanged 4 months agoMITView on GitHub

Entrypoint structure, exit codes, cross-platform scripts. Use when building command line utilities, scripts, or applications.

4 other projects use a version of this.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

16 rules. About 0 are common, 0 are distinctive, and 16 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 16

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Run dart create -t cli <project_name> to scaffold a console application with basic argument parsing.

    line 23
  2. Very specific:

    Place executable entry points (files containing main()) exclusively in the bin/ directory.

    line 24
  3. Very specific:

    Place internal implementation logic in lib/src/ and expose public APIs via lib/<project_name>.dart.

    line 25
  4. Very specific:

    Import the args package to manage command-line arguments, flags, and subcommands.

    line 30
  5. Very specific:

    If building a simple script: Use ArgParser directly to define flags (addFlag) and options (addOption).

    line 32
  6. Very specific:

    If building a complex, multi-command CLI (like git): Implement CommandRunner and extend Command for each subcommand.

    line 33
  7. Very specific:

    Use the io package's ExitCode enum to return standard POSIX exit codes (e.g., ExitCode.success.code, ExitCode.usage.code).

    line 42
  8. Very specific:

    Use sharedStdIn from the io package if multiple asynchronous listeners need sequential access to standard input.

    line 43
  9. Very specific:

    Fail fast and with non-zero exit codes: Ensure operation failures result in descriptive error messages to stderr and appropriate non-zero …

    line 47
  10. Very specific:

    Use test_process and test_descriptor to write high-fidelity integration tests for your CLI.

    line 54
  11. Very specific:

    Create the mock filesystem before execution using await d.Descriptor.create().

    line 57
  12. Very specific:

    If testing locally during development: Use dart run bin/cli.dart.

    line 67
  13. Very specific:

    If bundling code assets and dynamic libraries: Use dart build cli. This runs build hooks and outputs to build/cli/_/bundle/.

    line 68
  14. Very specific:

    If distributing a standalone native executable: Use dart compile exe bin/cli.dart -o <output_path>.

    line 69
  15. Very specific:

    If distributing multiple apps with strict disk space limits: Use dart compile aot-snapshot bin/cli.dart.

    line 70
  16. Very specific:

    Use the --target-os and --target-arch flags with dart compile exe or dart compile aot-snapshot.

    line 76

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Building Dart CLI Applications

Contents

  • Project Setup & Architecture
  • Argument Parsing & Command Routing
  • Execution & Error Handling
  • Testing CLI Applications
  • Compilation & Distribution
  • Workflows
  • Examples

Project Setup & Architecture

Initialize new CLI projects using the official Dart template to ensure standard directory structures.

  • Run dart create -t cli <project_name> to scaffold a console application with basic argument parsing.
  • Place executable entry points (files containing main()) exclusively in the bin/ directory.
  • Place internal implementation logic in lib/src/ and expose public APIs via lib/<project_name>.dart.
  • Enforce formatting in CI environments by running dart format . --set-exit-if-changed. This returns exit code 1 if formatting violations exist.

Argument Parsing & Command Routing

Import the args package to manage command-line arguments, flags, and subcommands.

  • If building a simple script: Use ArgParser directly to define flags (addFlag) and options (addOption).
  • If building a complex, multi-command CLI (like git): Implement CommandRunner and extend Command for each subcommand.
  • Define global arguments on the CommandRunner.argParser and command-specific arguments on the individual Command.argParser.
  • Catch UsageException to gracefully handle invalid arguments and display the automatically generated help text.
  • Validate Help Text Accuracy: Ensure the help text provides all necessary information to run the tool. If the help text references a compiled executable name, and the user needs to add it to their PATH to run it that way, provide clear instructions on how to do so in the help text or description.

Execution & Error Handling

Leverage the io and stack_trace packages to build robust, production-ready CLI tools.

  • Use the io package's ExitCode enum to return standard POSIX exit codes (e.g., ExitCode.success.code, ExitCode.usage.code).
  • Use sharedStdIn from the io package if multiple asynchronous listeners need sequential access to standard input.
  • Wrap the application execution in Chain.capture() from the stack_trace package to track asynchronous stack chains.
  • Format output stack traces using Trace.terse or Chain.terse to strip noisy core library frames and present readable errors to the user.
  • Do not swallow exceptions in lower-level logic or storage classes unless recovery is possible. Let them bubble up or rethrow them so higher-level commands know operations failed.
  • Fail fast and with non-zero exit codes: Ensure operation failures result in descriptive error messages to stderr and appropriate non-zero exit codes (e.g., using exit(1) or triggering a 64 exit code after a caught UsageException).

Testing CLI Applications

[!IMPORTANT] All new commands and significant features must be covered by automated tests. Manual verification is not sufficient for testing logic. However, manual verification of help text and user experience (UX) is still required to ensure the interface is intuitive and correct.

Use test_process and test_descriptor to write high-fidelity integration tests for your CLI.

  • Define expected filesystem states using test_descriptor (d.dir, d.file).
  • Create the mock filesystem before execution using await d.Descriptor.create().
  • Spawn the CLI process using TestProcess.start('dart', ['run', 'bin/cli.dart', ...args]).
  • Validate standard output and error streams using StreamQueue matchers (e.g., emitsThrough, emits).
  • Assert the final exit code using await process.shouldExit(0).
  • Validate resulting filesystem mutations using await d.Descriptor.validate().

Compilation & Distribution

Select the appropriate compilation target based on your distribution requirements.

  • If testing locally during development: Use dart run bin/cli.dart. This uses the JIT compiler for rapid iteration.
  • If bundling code assets and dynamic libraries: Use dart build cli. This runs build hooks and outputs to build/cli/_/bundle/.
  • If distributing a standalone native executable: Use dart compile exe bin/cli.dart -o <output_path>. This bundles the Dart runtime and machine code into a single file.
  • If distributing multiple apps with strict disk space limits: Use dart compile aot-snapshot bin/cli.dart. Run the resulting .aot file using dartaotruntime.
<details> <summary>Cross-Compilation Targets (Linux Only)</summary>

Dart supports cross-compiling to Linux from macOS, Windows, or Linux hosts. Use the --target-os and --target-arch flags with dart compile exe or dart compile aot-snapshot.

  • --target-os=linux (Only Linux is currently supported as a cross-compilation target)
  • --target-arch=arm64 (64-bit ARM)
  • --target-arch=x64 (x86-64)
  • --target-arch=arm (32-bit ARM)
  • --target-arch=riscv64 (64-bit RISC-V)

Example: dart compile exe --target-os=linux --target-arch=arm64 bin/cli.dart

</details>

Workflows

Task Progress: Implement a New CLI Command

  • Create a new class extending Command in lib/src/commands/.
  • Define the name and description properties.
  • Register command-specific flags in the constructor using argParser.addFlag() or argParser.addOption().
  • Implement the run() method with the core logic.
  • Register the new command in the CommandRunner instance in bin/cli.dart using addCommand().
  • Create tests for the new command in the test/ directory using test_process or standard tests.
  • Run validator -> Execute dart run bin/cli.dart help <command_name> to verify help text generation.
  • Verify final UX: Compile the application using dart compile exe and run the resulting executable to verify the target user experience (e.g., ./bin/cli <command>).

Task Progress: Compile and Release Native Executable

  • Run validator -> Execute dart format . --set-exit-if-changed to ensure code formatting.
  • Run validator -> Execute dart analyze to ensure no static analysis errors.
  • Run validator -> Execute dart test to pass all integration tests.
  • Compile for host OS: dart compile exe bin/cli.dart -o build/cli-host
  • Compile for Linux (if host is macOS/Windows): dart compile exe --target-os=linux --target-arch=x64 bin/cli.dart -o build/cli-linux-x64

Examples

Example: CommandRunner Implementation

import 'dart:io';
import 'package:args/command_runner.dart';
import 'package:stack_trace/stack_trace.dart';

class CommitCommand extends Command {
  @override
  final String name = 'commit';
  @override
  final String description = 'Record changes to the repository.';

  CommitCommand() {
    argParser.addFlag('all', abbr: 'a', help: 'Commit all changed files.');
  }

  @override
  Future<void> run() async {
    final commitAll = argResults?['all'] as bool? ?? false;
    print('Committing... (All: $commitAll)');
  }
}

void main(List<String> args) {
  Chain.capture(() async {
    final runner = CommandRunner('dgit', 'Distributed version control.')
      ..addCommand(CommitCommand());

    await runner.run(args);
  }, onError: (error, chain) {
    if (error is UsageException) {
      stderr.writeln(error.message);
      stderr.writeln(error.usage);
      exit(64); // ExitCode.usage.code
    } else {
      stderr.writeln('Fatal error: $error');
      stderr.writeln(chain.terse);
      exit(1);
    }
  });
}

Example: Integration Testing with Subprocesses

import 'package:test/test.dart';
import 'package:test_process/test_process.dart';
import 'package:test_descriptor/test_descriptor.dart' as d;

void main() {
  test('CLI formats output correctly and modifies filesystem', () async {
    // 1. Setup mock filesystem
    await d.dir('project', [
      d.file('config.json', '{"key": "value"}')
    ]).create();

    // 2. Spawn the CLI process
    final process = await TestProcess.start(
      'dart',
      ['run', 'bin/cli.dart', 'process', '--path', '${d.sandbox}/project']
    );

    // 3. Validate stdout stream
    await expectLater(process.stdout, emitsThrough('Processing complete.'));

    // 4. Validate exit code
    await process.shouldExit(0);

    // 5. Validate filesystem mutations
    await d.dir('project', [
      d.file('config.json', '{"key": "value"}'),
      d.file('output.log', 'Success')
    ]).validate();
  });
}

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.