Agent skill

Gaia Harness Engineering

by amd in amd/gaia

Help a developer report a GAIA TUI, WebUI or agent problem, explicitly share a snapshot with Claude Code or Codex through MCP, and validate isolated worktree fixes.

MITAuto-check passed

Install Gaia Harness Engineering

skills CLI
$ npx skills add amd/gaia --skill gaia-harness-engineering -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install amd/gaia gaia-harness-engineering --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/amd/gaia.git skills-src && mkdir -p .claude/skills && cp -r skills-src/hub/agents/gaia/python/gaia_agent/skills/gaia-harness-engineering .claude/skills/gaia-harness-engineering && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
gaia-harness-engineering
GitHub stars
1.6k
Token cost
~882 tokens
SKILL.md length
468 words
Files
1
Skills in repo
44
Repo updated
First seen
Licence
MIT

At a glance

Help a developer report a GAIA TUI, WebUI or agent problem, explicitly share a snapshot with Claude Code or Codex through MCP, and validate isolated worktree fixes.

  • Works in 9 steps: Clarify expected and observed behavior.… → Select claude or codex with the… → Give the developer the job ID and… → …
  • Tasks that involve Git worktrees
  • Instructions only: no scripts, shell commands, URLs or credentials in SKILL.md

What it does

Gaia Harness Engineering is an agent skill from amd/gaia. Help a developer report a GAIA TUI, WebUI or agent problem, explicitly share a snapshot with Claude Code or Codex through MCP, and validate isolated worktree fixes.

Its SKILL.md is about 880 tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It works with Model Context Protocol. The repository describes itself as: Build AI agents for your PC. The licence is MIT.

When your agent uses it

  • Tasks that involve Git worktrees

Example prompts

  • “/gaia-harness-engineering”

Workflow steps

9 steps, taken from the first numbered list in SKILL.md.

  1. Clarify expected and observed behavior. Use engineering_status without a
  2. Select claude or codex with the developer. Prepare a concise summary and
  3. Give the developer the job ID and connection instructions. Use
  4. Diagnose before editing: reproduce safely using synthetic or read-only
  5. After diagnosis, use approve_engineering_code to request worktree scope.
  6. Have the coding app launch a separate agent/TUI/WebUI preview with isolated
  7. Use engineering_status for reported progress and preview details. Clearly
  8. Review all outgoing public text, code and commit history. Generalize private
  9. revoke_engineering_context stops future bridge reads. Already delivered

What it can do on your machine

Read from SKILL.md and the folder at commit 05fb50b. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    No scripts in the folder and no shell commands in SKILL.md.

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Gaia Harness Engineering loads about 882 tokens when it runs. Until then it costs about 47 tokens; SKILL.md has 468 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~47
When it runs · the whole SKILL.md, loaded when a task matches
~882

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from amd/gaia at commit 05fb50b, republished under its MIT licence (© amd). 468 words, ~882 tokens.

Download SKILL.mdSave it as .claude/skills/gaia-harness-engineering/SKILL.md (or your agent's skills folder).
name
gaia-harness-engineering
description
Help a developer report a GAIA TUI, WebUI or agent problem, explicitly share a snapshot with Claude Code or Codex through MCP, and validate isolated worktree fixes.

Developer handoff

This skill is enabled by the host only in developer mode. Start only when the developer reports a problem or requests an improvement. Do not automatically assess friction or start coding. Keep GAIA available for daily work.

  1. Clarify expected and observed behavior. Use engineering_status without a job ID to inspect source-cache initialization and coding-app setup. Surface failures and connection instructions; do not claim connection or task launch merely because the app exists.
  2. Select claude or codex with the developer. Prepare a concise summary and selected evidence. Do not collect entire histories, unrelated files or credentials. share_engineering_context shows the exact summary and context and requests fresh approval before the selected coding app can read them. The app may use a cloud provider; its existing filesystem permissions remain independent of this bridge. Never use shell, file tools or another MCP tool to bypass this approval. Bypass-permissions mode cannot approve sharing.
  3. Give the developer the job ID and connection instructions. Use open_engineering_app only when asked to open the app. An opened app or prefilled composer is not a submitted or running task. The developer may interact directly in their existing coding app. For Codex specifically, opening only brings the app forward: it does not create a task, select a folder or prefill a composer. Say this explicitly, then show the returned directory and exact prompt to paste into a new task. Never say "posted to Codex", "sent", or "look for the prefilled composer". A generated MCP config is not an installed or verified client connection.
  4. Diagnose before editing: reproduce safely using synthetic or read-only inputs, inspect configuration/model limits and actual build identities, check releases and existing issues/PRs. Do not replay destructive tools or external sends. Recommend a model/configuration change when evidence supports it; ask before changing the user's model. Unknown root cause stays unknown.
  5. After diagnosis, use approve_engineering_code to request worktree scope. All edits must be in an isolated worktree from the managed GAIA cache. Private snapshots stay outside Git; they must not become fixtures or commits.
  6. Have the coding app launch a separate agent/TUI/WebUI preview with isolated writable state, ports and explicit build revision. A worktree is not a security sandbox. Let the developer test and iterate. Share additional feedback only through append_engineering_context, which asks again.
  7. Use engineering_status for reported progress and preview details. Clearly distinguish app-reported results from locally verified checks. Native app conversations and detailed coding remain in that app.
  8. Review all outgoing public text, code and commit history. Generalize private examples. Workflow approval does not make raw context safe to publish. Follow repository contribution/review requirements; maintainers own merge.
  9. revoke_engineering_context stops future bridge reads. Already delivered data cannot be recalled. Stable release updates are a separate feature; never replace the running GAIA installation from a preview worktree.

© amd, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in hub/agents/gaia/python/gaia_agent/skills/gaia-harness-engineering of amd/gaia.

Open the folder on GitHubat commit 05fb50b

Compare with similar skills

Gaia Harness Engineering next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Gaia Harness Engineering compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Gaia Harness Engineering this skillamd/gaia1.6k—~882Automated safety check: PassMIT
TuiosGaurav-Gosain/tuios5k—~3.7kAutomated safety check: PassMIT
E2EAgentsMesh/AgentsMesh2.4k—~629Automated safety check: NotesCustom licence
Agent Deckasheshgoplani/agent-deck1k—~1.7kAutomated safety check: PassMIT
Puppetmaster Agent Orchestrationprofessorpalmer/Puppetmaster467—~3.2kAutomated safety check: PassMIT
Agent Manager Fleet TUIYoanWai/agent-manager576—~1kAutomated safety check: PassApache-2.0

Similar skills

  • Tuios

    Gaurav-Gosain/tuios

    Drive tuios from inside one of its panes. An agent skill from Gaurav-Gosain/tuios.

    5k GitHub stars~3.7k tokensUpdated today
    MobileAuto-check passed
  • E2E

    AgentsMesh/AgentsMesh

    Selects and runs the appropriate AgentsMesh end-to-end suite for Web, Desktop, MCP, or iOS, including worktree-specific environment setup and browser-level verification.

    2.4k GitHub stars~629 tokensUpdated 15 days ago
    Testing & QAAuto-check: notes
  • Agent Deck

    asheshgoplani/agent-deck

    agent-deck, the terminal session manager for AI coding agents.

    1k GitHub stars~1.7k tokensUpdated 2 days ago
    Agent WorkflowsAuto-check passed
  • Puppetmaster Agent Orchestration

    professorpalmer/Puppetmaster

    Operates and supervises Puppetmaster, a multi-agent orchestrator, through its MCP tools or CLI, picking the right verb for edits, reviews, audits and long-running jobs.

    467 GitHub stars~3.2k tokensUpdated today
    Agent WorkflowsAuto-check passed
  • Agent Manager Fleet TUI

    YoanWai/agent-manager

    Runs several coding-agent CLIs as real tmux sessions in one terminal UI, color-coded by whether each is working, waiting, idle or blocked.

    576 GitHub stars~1k tokensUpdated today
    Agent WorkflowsAuto-check passed
  • Agtx Task Sweep

    fynnfluegge/agtx

    Breaks a conversation's results into feature-level tasks and pushes them to the agtx kanban board, where each task gets its own worktree and agent session.

    1.7k GitHub stars~1.7k tokensUpdated 6 days ago
    Agent WorkflowsAuto-check passed

More from amd/gaia

All 44 skills in this repo
  • Adds a release eval scorecard to a GAIA hub agent by writing a harness adapter, running a real eval, and wiring the result into the agent's README and release gate.

    1.6k GitHub stars~2.6k tokensUpdated today
    Auto-check passed
  • Walks through releasing a GAIA sidecar agent as a frozen binary plus npm client through the tag-triggered Agent Hub CI pipeline, with a human gate before publishing.

    1.6k GitHub stars~3.6k tokensUpdated today
    Auto-check passed
  • Mines local Claude Code session transcripts with a deterministic Python pipeline to show what the agent is actually used for, how often it fails and what it costs.

    1.6k GitHub stars~2.3k tokensUpdated today
    Auto-check passed
  • Benchmarks AMD's GAIA agent against Claude Code and across models on quality, honesty, steps, tokens, time and real cost, using gaia eval tasks.

    1.6k GitHub stars~1.8k tokensUpdated today
    Auto-check passed
  • Guides safe code changes by finding the right file with grep or semantic search, reading before editing, reproducing bugs first, and proving a fix with a real test run.

    1.6k GitHub stars~2.1k tokensUpdated today
    Auto-check passed
  • Walks through scaffolding, writing and testing a new GAIA agent as a Python class with the SDK, from the base Agent subclass to registered tool methods.

    1.6k GitHub stars~1.5k tokensUpdated today
    Auto-check passed

Questions about Gaia Harness Engineering

What does Gaia Harness Engineering do?

Help a developer report a GAIA TUI, WebUI or agent problem, explicitly share a snapshot with Claude Code or Codex through MCP, and validate isolated worktree fixes. Gaia Harness Engineering is an agent skill from amd/gaia. Help a developer report a GAIA TUI, WebUI or agent problem, explicitly share a snapshot with Claude Code or Codex through MCP, and validate isolated worktree fixes.

When should I use Gaia Harness Engineering?

Gaia Harness Engineering fits situations like: tasks that involve Git worktrees.

How do I install Gaia Harness Engineering in Claude Code?

Run `npx skills add amd/gaia --skill gaia-harness-engineering -a claude-code`. Or copy the skill folder (hub/agents/gaia/python/gaia_agent/skills/gaia-harness-engineering in amd/gaia) into .claude/skills/gaia-harness-engineering in your project. Claude Code loads it when a task matches its description.

How do I install Gaia Harness Engineering in Codex?

Run `npx skills add amd/gaia --skill gaia-harness-engineering -a codex`. Or copy the skill folder (hub/agents/gaia/python/gaia_agent/skills/gaia-harness-engineering in amd/gaia) into .agents/skills/gaia-harness-engineering in your project. Codex loads it when a task matches its description.

Can I use Gaia Harness Engineering in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add amd/gaia --skill gaia-harness-engineering -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/gaia-harness-engineering, .gemini/skills/gaia-harness-engineering, .github/skills/gaia-harness-engineering and .opencode/skills/gaia-harness-engineering in your project.

What does Gaia Harness Engineering need to run?

SKILL.md names no scripts, command-line tools or credentials: Gaia Harness Engineering is instructions for the agent only.

Does Gaia Harness Engineering access the network?

SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.

Is Gaia Harness Engineering safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Gaia Harness Engineering use?

Gaia Harness Engineering is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Gaia Harness Engineering use?

About 882 tokens (SKILL.md is roughly 3.5k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Gaia Harness Engineering?

Skills that share tags, products or a category with Gaia Harness Engineering: Tuios (Gaurav-Gosain/tuios, 5k stars), E2E (AgentsMesh/AgentsMesh, 2.4k stars), Agent Deck (asheshgoplani/agent-deck, 1k stars) and Puppetmaster Agent Orchestration (professorpalmer/Puppetmaster, 467 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Gaia Harness Engineering?

amd (a GitHub organization) maintains it in amd/gaia, which has 1,580 GitHub stars. The repository holds 44 skills in this directory. The repository was last updated on October 8, 2026.

Source: amd/gaia on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.