Validation

Validation Skill

Set up named validation commands (lint, test, typecheck, etc.) that act as repeatable CI-style checks for the project, then run and monitor them.

When to Use

• After building new functionality or making significant changes
• When setting up a project's quality gates for the first time
• When the validation command for a check needs to change (e.g., new test framework)
• To run validation commands and check their results
• To monitor the status of ongoing or past validation runs

When NOT to Use

• For one-off test runs — use the shell directly
• For deploy workflows — use the deployment skill

Available Functions

setValidationCommand(name, command)

Create or update a named validation command. If a validation with the same name already exists, it is silently updated (upsert behavior).

Parameters:

Returns: { success: true, name: string, command: string }

Example:

await setValidationCommand({ name: "lint", command: "npm run lint" });
await setValidationCommand({ name: "test", command: "npm test" });

clearValidationCommand(name)

Remove a previously registered validation command.

Parameters:

Returns: { success: true, name: string }

Example:

await clearValidationCommand({ name: "lint" });

getValidationCommands()

List all registered validation commands.

Parameters: None

Returns: { workflows: [{ name: string, tasks: [...], ... }] }

Example:

const commands = await getValidationCommands();

startValidationRun(commandIds)

Start a validation run that executes one or more registered validation commands.

Parameters:

Returns:

{
  "runId": "run-abc123",
  "status": "FAILED",
  "commands": [
    {
      "commandId": "lint",
      "shell": "npm run lint",
      "status": "FAILED",
      "exitCode": 1,
      "errorMessage": "Exited with: 1",
      "executionId": "exec-1",
      "durationMs": 3200,
      "logFilePath": "/tmp/validation/lint.log"
    },
    {
      "commandId": "test",
      "shell": "npm test",
      "status": "PASSED",
      "exitCode": 0,
      "errorMessage": "",
      "executionId": "exec-2",
      "durationMs": 5100,
      "logFilePath": "/tmp/validation/test.log"
    }
  ],
  "durationMs": 8300,
  "errorMessage": "",
  "runSummary": "- `npm run lint` FAILED (exit code 1): ESLint reported ..."
}

• status is one of: "RUNNING", "PASSED", "FAILED", "STOPPED", "ERROR", "TIMED_OUT"
• runSummary is an LLM-generated breakdown of the results with log line citations. Use it to understand what failed without reading full logs.
• Each command includes logFilePath pointing to the full output log.

Example:

await setValidationCommand({ name: "lint", command: "npm run lint" });
await setValidationCommand({ name: "test", command: "npm test" });
const run = await startValidationRun({ commandIds: ["lint", "test"] });

if (run.status !== "PASSED") {
  // run.runSummary explains what failed and cites log lines
  console.log(run.runSummary);

  // Inspect individual commands
  for (const cmd of run.commands) {
    if (cmd.status !== "PASSED") {
      console.log(`${cmd.shell} failed (exit ${cmd.exitCode}): ${cmd.errorMessage}`);
      console.log(`Full log: ${cmd.logFilePath}`);
    }
  }
}

stopValidationRun(runId, graceful?)

Stop a running validation run.

Parameters:

Returns: { success: true, runId: string }

Example:

await stopValidationRun({ runId: "run-abc123" });
await stopValidationRun({ runId: "run-abc123", graceful: false });

getValidationRun(runId)

Get the status and details of a specific validation run, including per-command results.

Parameters:

Returns: { validationRun: { id, status, startedAt, durationMs, commands: [...] } }

Example:

const run = await getValidationRun({ runId: "run-abc123" });
// run.validationRun.status is one of: "RUNNING", "PASSED", "FAILED", "STOPPED", "ERROR"

getValidationRuns()

Get all validation runs.

Parameters: None

Returns: { validationRuns: [{ id, status, startedAt, durationMs, commands: [...] }] }

Example:

const runs = await getValidationRuns();

Conventional Names and Commands

Use short, lowercase names. Here are common conventions by stack:

Example Workflow

After implementing a feature in a Node.js project:

// Set up validation commands for the project
await setValidationCommand({ name: "lint", command: "npm run lint" });
await setValidationCommand({ name: "test", command: "npm test" });
await setValidationCommand({ name: "typecheck", command: "npx tsc --noEmit" });

// Run all validations (blocks until complete)
const run = await startValidationRun({ commandIds: ["lint", "test", "typecheck"] });

if (run.status !== "PASSED") {
  // runSummary explains failures with log line citations
  console.log(run.runSummary);

  for (const cmd of run.commands) {
    if (cmd.status !== "PASSED") {
      console.log(`${cmd.shell} failed (exit ${cmd.exitCode}): ${cmd.errorMessage}`);
      console.log(`Full log: ${cmd.logFilePath}`);
    }
  }
}

// Or fetch it again later by ID
const result = await getValidationRun({ runId: run.runId });