Baltor Get started

Harness tools

Slash commands

Saved prompts your harness runs as a command: debugging, reviews, changelogs, audits, refactors and migrations. Each keeps its source repository and licence.

11 shown here, from 2,818 commands in the library release of October 9, 2026.

What to watch for

  • A command that writes files should say so before it runs.
  • Commands work best with one clear job and a defined output.
  • Keep the command's source so an update can be compared with it.

Components

  1. Systematic debugging

    debug.md

    Systematic debugging: root cause before any fix, one hypothesis at a time, and a three-strikes rule that questions the architecture instead of stacking patches.

    File
    debug.md
    From
    github.com/azrtydxb/procoder

    CommandApache-2.03 files, 14.4 KBreads files, runs commands

  2. Auto-update CHANGELOG.md based on git commits

    changelog.md

    Auto-update CHANGELOG.md based on git commits

    File
    changelog.md
    From
    github.com/Data-Wise/craft

    CommandMIT3 files, 8.1 KBreads files, runs commands

  3. Keep your open PR merge-ready

    babysit-pr.md

    Keep your open PR merge-ready — fix CI failures, address review comments, resolve merge conflicts

    File
    babysit-pr.md
    From
    github.com/stevenmays/dotfiles

    CommandMIT3 files, 5.3 KBruns commands

  4. Audit only the git diff vs main

    audit-changes.md

    Audit only the git diff vs main (or specified base).

    File
    audit-changes.md
    From
    github.com/iktok90-design/ai-smart-contract-auditor

    CommandMIT3 files, 3.7 KBreads files, runs commands

  5. Audit third-party dependencies

    audit-deps.md

    Audit third-party dependencies — resolve installed versions, cross-reference known-vulnerable releases, and flag vendored code that has diverged from upstream.

    File
    audit-deps.md
    From
    github.com/iktok90-design/ai-smart-contract-auditor

    CommandMIT3 files, 5.7 KBreads files, runs commands

  6. Execute controlled behavior-preserving frontend refactoring with small steps

    fec-refactor.md

    Execute controlled behavior-preserving frontend refactoring with small steps, verification, rollback, and proof reporting.

    File
    fec-refactor.md
    From
    github.com/bovinphang/frontend-craft

    CommandMIT3 files, 3.0 KBnone declared

  7. Create the next sequentially-numbered ADR in house format

    adr.md

    Create the next sequentially-numbered ADR in house format (Status/Context/Decision/Consequences/Date) for a settled decision, and link it from ARCHITECTURE.md.

    File
    adr.md
    From
    github.com/MrEyratnz/software-factory

    CommandMIT3 files, 2.7 KBwrites files

  8. Comprehensive CI/CD audit

    audit-ci.md

    Comprehensive CI/CD audit.

    File
    audit-ci.md
    From
    github.com/dykyi-roman/awesome-claude-code

    CommandMIT3 files, 7.3 KBreads files, runs commands, writes files

  9. Audit Docker configuration

    audit-docker.md

    Audit Docker configuration.

    File
    audit-docker.md
    From
    github.com/dykyi-roman/awesome-claude-code

    CommandMIT3 files, 8.0 KBreads files, runs commands, writes files

  10. Generate comprehensive test cases from requirements or user stories

    write-test-cases.md

    Generate comprehensive test cases from requirements or user stories

    File
    write-test-cases.md
    From
    github.com/latestaiagents/agent-skills

    CommandMIT3 files, 3.0 KBnone declared

  11. Run an accessibility-only audit

    a11y.md

    Run an accessibility-only audit — focus indicators, skip links, alt text, prefers-reduced-motion, WCAG contrast.

    File
    a11y.md
    From
    github.com/Aboudjem/ui-ux-suite

    CommandMIT3 files, 3.3 KBuses the network, reads files, runs commands

Give your agent these components

Your harness downloads them through Baltor with Pro, $49 a month. Cancel any time from your account page.

Get started

Compare the plansSet up your harness

More from Harness tools

  • MCP server settingsConnection entries for MCP servers, written for Claude Code, Codex, Cursor and OpenCode.
  • Command-line programsInstall recipes with the published checksum for each platform, and a typed wrapper that checks the arguments before the program runs.
  • HooksScripts your harness runs before or after a step: block risky commands, check for secrets, enforce commit messages, and format or lint after an edit.
  • Tested functionsOne upstream Python function at a time, copied with the code it needs and a test: validators, parsers and formatters your agent can import without adding a dependency.
  • 3D and game file checkersStandard-library checkers for glTF and Godot files that catch what an engine would drop or misread.
  • Deterministic skillsSkills that give the same answer every time.
  • Corner casesThe inputs that break code, ready to test.
  • Stop starting from nothingFor a developer whose coding agent starts every task from nothing: give each step the guidance and reusable code it needs.
  • Harness FilesSkills, instruction files, tools, hooks and settings your harness picks up: what they are, how a harness gets them, and the setup for each harness.
  • Get set upConnect Claude Code, Codex, Cursor, OpenCode, VS Code or Pi over MCP, call the REST API, or give an agent a feed link. Then download your first file.
  • Top MCP serversA task-focused MCP shortlist with dated GitHub popularity, setup sources and clear limits on quality and Baltor usage claims.
  • Claude Code quickstartWhat you need, the .mcp.json entry to paste, the first search, the first download and the check that it worked, for Claude Code.