Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Background Task Scheduler

    1

    Local background task scheduler with L0-L3 permissions, staged orchestration, and built-in resilient downloaders.

    Free

    Security scanned
    Background Task Scheduler

    Background Task Scheduler

    Example session with this skill installed

    Submit a plan to lint and test the project. Lint must finish before unit and e2e tests start in parallel. e2e depends on unit. Set level to L1.

    • Read your context and instructions
    • Compiled the background task scheduler

    The plan has been successfully materialized and submitted. It is currently in a pending state, waiting for the first execution trigger.

    Plan Materialization: ci-pipeline-a1

    {
      "plan_id": "ci-pipeline-a1",
      "status": "pending",
      "directory": ".bts/schedules/ci-pipeline/plans/ci-pipeline-a1",
      "structure": {
        "stages": [
          {
            "id": "lint_stage",
            "mode": "serial",
            "steps": ["eslint"]
          },
          {
            "id": "test_stage",
            "mode": "parallel",
            "steps": ["unit_test", "e2e_test"]
          }
        ]
      },
      "dependency_graph": {
        "e2e_test": ["unit_test"]
      },
      "commands": {
        "eslint": "npm run lint",
        "unit_test": "npm run test:unit",
        "e2e_test": "npm run test:e2e"
      }
    }
    

    Next steps

    1. Start the execution by running python scripts/bts_plan.py run_plan ci-pipeline-a1.
    2. Monitor progress and check for any L1 approval requirements using python scripts/bts.py list --status pending.
    3. View real-time logs for specific steps using python scripts/bts.py show <task_id> --lines 20.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Long-running commands often stall agent workflows or fail silently without a trace. This skill provides a local, daemon-less background task scheduler that moves execution out of the chat turn while maintaining strict control. It handles staged orchestration with dependency management, automatic retries, and a multi-tier permission system.

    What it does

    • Background Execution Offloads time-intensive tasks like builds, training, or batch processing to detached local processes.
    • Staged Orchestration Groups commands into Stages with serial or parallel execution modes and intra-stage depends_on logic.
    • Permission Gates Enforces L0-L3 security tiers, requiring explicit human approval for high-risk commands.
    • Automated Downloads Manages HTTP(S) and BitTorrent (via aria2c) transfers with resume support, integrity checks, and rate limiting.
    • Resilient Scheduling Runs recurring plans on fixed intervals with stall detection and failure propagation.
    • Post-Hook Processing Triggers automatic encoding conversion via @enc hooks upon successful command completion.

    How it works

    1. Create Tasks Use bts.py create or pack to register single commands or batches with specific permission levels and timeouts.
    2. Submit Plans Define multi-step sequences in a plan.json file and submit them for one-time or recurring execution.
    3. Approve & Run Review pending approvals via bts.py approve then trigger execution with run or run_plan.
    4. Monitor State Poll for updates using check, show, or events to track logs, exit codes, and process snapshots.

    Frameworks & tools

    Built for Python 3 environments using standard libraries. Optional integration with aria2c for torrent functionality. State is managed via a local .bts filesystem directory (JSON/JSONL).

    Why this beats prompting it yourself

    Prompting an agent to "run this in the background" usually results in fragile shell scripts that die when the session ends or lack error handling. This skill provides a structured state machine with file-based persistence, ensuring tasks survive agent crashes and failures are properly logged for audit.

    Use cases

    • Running long-duration CI/CD pipelines or test suites locally.
    • Scheduling recurring web scraping or system health checks.
    • Orchestrating multi-step deployments that require manual approval gates.
    • Managing large-scale file downloads with bandwidth constraints and auto-retry.

    Known limitations

    Requires manual polling via check commands to advance state as there is no persistent background daemon. Only supports single-machine file-based locking, not distributed systems.

    No background daemon; requires polling to drive schedules. Single-machine file locks only. Identity levels are declarative. aria2c required for torrents. No generic post-hooks besides @enc.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before any payment.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 10 days ago

    • Free with an Agensi account

    Requires

    Python
    Git
    aria2c

    Permissions

    Terminal / Shell
    Read Files
    Network Access
    Write Files
    Environment Variables
    Listed10 days ago

    What's inside

    Frequently Asked Questions

    Unlike standard background scripts that vanish if an agent crashes, this skill uses a local file-based state machine to ensure tasks persist, provide detailed logs, and handle complex multi-stage dependencies automatically.

    The L0-L3 permission system allows you to pre-approve low-risk tasks while forcing the agent to pause and wait for your explicit manual authorization before executing sensitive or high-resource commands.

    The skill is designed for any Python 3 environment and works with any AI agent capable of executing terminal commands or interacting with local files, including AutoGPT, OpenDevin, or custom LangChain agents.

    No, this is a daemon-less system that uses a local `.bts` directory for state management, meaning it doesn't consume background CPU cycles when idle and doesn't require complex root-level service installations.

    The package includes the `bts.py` core engine, documentation for JSON plan structures, and pre-configured hooks for advanced features like aria2c integration and automated encoding conversions.

    While designed for local execution, the scheduler includes robust error propagation and stall detection, allowing it to automatically retry failed steps or halt a multi-stage plan if a critical dependency fails.

    New in Workflow Automation