Laravel Tackle: Run an AI Coding Agent Inside Your Laravel Application

What Is Laravel Tackle?

Laravel Tackle, created by Jordan Dalton, is an AI coding agent that runs as a set of Artisan commands inside your Laravel application. Because it boots with the framework, it has direct access to the tools you already use: route listing, Telescope exceptions, database queries, and Laravel Pint for formatting after edits.

Safety controls — path restrictions, Artisan allowlists, and per-session spend limits — are enforced in PHP code, not in a prompt. The package is built on laravel/ai and defaults to Claude, but AI_CODE_PROVIDER lets you switch to OpenAI, Gemini, Groq, or a local model via Ollama.

Installation

Tackle requires PHP 8.3 and Laravel 12 or 13.

composer require jordandalton/laravel-tackle

php artisan vendor:publish --provider="Laravel\Ai\AiServiceProvider"
php artisan vendor:publish --tag="tackle-config"

Add your API key to .env:

ANTHROPIC_API_KEY=sk-ant-...

To use a local Ollama model instead:

AI_CODE_PROVIDER=ollama
AI_CODE_MODEL=deepseek-coder-v2
AI_CODE_PRICE_INPUT=0
AI_CODE_PRICE_OUTPUT=0

Tip: Commit or stash your work before the first session. The agent edits files in place unless worktree mode is enabled.

Available Agents

Tackle ships several focused agents:

• ai:code — interactive REPL with plan mode, slash commands, image attachments, and persistent history
• ai:run — non-interactive, single-task execution with --output=json for pipelines
• ai:fix — targeted fix from a pasted exception, a Sentry issue (--sentry=ID), or a GitHub issue (--issue=N)
• ai:review — reads a diff or pull request and posts inline review comments with severity levels
• ai:upgrade — migrates a Composer package across a major version using the package's own upgrade guide
• Self-healer — watches for failed queue jobs and scheduled tasks, patches code in an isolated worktree, and opens a pull request
• Tackle Remote — a companion package serving the same harness as a mobile browser UI

Safety Boundaries

All limits live in config/tackle.php and cannot be overridden by a prompt:

return [
    'budget_usd' => env('AI_CODE_BUDGET', 1.00),
    'shell' => [
        'local'      => env('AI_CODE_SHELL', 'approve'),
        'production' => env('AI_CODE_SHELL', 'off'),
    ],
    'artisan_allowlist' => [
        'local'      => ['make:*', 'migrate:*', 'db:seed', 'route:list', 'test'],
        'production' => ['route:list'],
    ],
    'protected_paths' => ['.env', '.env.*', 'storage/*', 'vendor/*', '.git/*'],
];

protected_paths blocks reads as well as writes, so the agent cannot read your .env. The budget is a hard stop — the session aborts when estimated spend crosses the limit, with a warning at 80%.

Self-Healing Queue Jobs

Enable self-healing with AI_CODE_HEALING_ENABLED=true, publish the migration, and start a worker on the healer queue. When a job fails, Tackle dispatches a healing agent that applies a minimal fix, runs your test suite, and either opens a pull request or merges directly in patch mode.

You can opt specific jobs out with an attribute:

use Tackle\Attributes\Healable;

#[Healable(false)]
class IssueRefund implements ShouldQueue
{
    public function handle(): void
    {
        // Skipped by the healer entirely.
    }
}

All healing attempts are logged to a tackle_healing_log table, queryable via php artisan tackle:healing-log.

MCP Server

Tackle can expose its Laravel-aware tools over MCP so external editors like Claude Code, Cursor, or Zed can call ListRoutes, QueryDatabase, ReadTelescopeEntry, and RunLarastan against your app:

claude mcp add tackle -- php artisan tackle:mcp

The exposed tool set defaults to read and analysis tools only — no file writes, no shell access.

Key Takeaways

• Safety rules are PHP code, not prompt instructions — they cannot be talked around
• Six specialised agents cover coding, fixing, reviewing, upgrading, and self-healing
• Worktree mode keeps edits isolated until you review the diff
• TACKLE.md lets you encode project conventions, boundaries, and gotchas for every session
• MCP support lets external editors use the same Laravel-aware tools
• Currently at v1.27.3, MIT-licensed, on GitHub at JordanDalton/laravel-tackle

Source: Laravel Tackle on Laravel News