skills.rest
IgorGanapolsky
IgorGanapolskyCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

aistudio-harness

Validates agent harness configurations against Google AI Studio Agents patterns.

1|Updated Aug 27, 2025
Search Tags:#validation#agents.md#sandbox#agent harness#ai studio#tool allowlist#network allowlist
One-click install
npx skills add https://github.com/IgorGanapolsky/Random-Timer --skill aistudio-harness-igorganapolsky
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: aistudio-harness
Source: https://github.com/IgorGanapolsky/Random-Timer/tree/main/.claude/skills/aistudio-harness
Command: npx skills add https://github.com/IgorGanapolsky/Random-Timer --skill aistudio-harness-igorganapolsky

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve? Teams adopting Google AI Studio Agents patterns risk unsafe setups: unmanaged tool access, open network egress, unbounded token loops, and unverified outputs. This Skill evaluates AGENTS.md and SKILL.md templates, tool allowlists, network restrictions, sandbox isolation, and termination conditions before any agent harness is considered ready. ## Core Features & Use Cases - Harness File Evaluation: Checks that AGENTS.md and SKILL.md persona and workflow templates are properly mounted. - Tool and Network Allowlisting: Validates explicit tool toggles and domain allowlists instead of all-tools-on or open egress. - Termination and Sandbox Checks: Enforces named stop conditions, worktree isolation, and verified execution with read-back before claiming done. - Use Case: Before deploying an agent workflow, run the harness script with your allowed tools, permitted domains, and stop condition; proceed only when every JSON check returns ok and the process exits 0. ## Quick Start Ask the AI to evaluate your agent harness by running scripts/studio_agent_harness.py with your AGENTS.md, SKILL.md, allowed tools, permitted domains, and stop condition, then confirm every check passes.

Frequently Asked Questions about aistudio-harness

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I validate an AI agent harness before deployment?▼

Run scripts/studio_agent_harness.py with your AGENTS.md and SKILL.md files, allowed tools, permitted domains, stop condition, and isolation mode. Proceed only when every JSON check returns ok true and the process exits with code 0.

What is the AI Studio Agents pattern for AGENTS.md and SKILL.md?▼

AI Studio Playground Agents mount AGENTS.md plus SKILL.md templates for persona and workflow, toggle tools explicitly, lock an ephemeral sandbox, and require a network allowlist. This Skill replicates that method locally without hosted agents.

Does this Skill use the Gemini API or hosted Antigravity agents?▼

No. The Skill explicitly avoids creating Gemini managed agents and does not run production work in AI Studio Playground, where a single turn can consume 100k to 3M tokens. It only borrows the harness methodology.

Why should agent tool access use an allowlist instead of all tools on?▼

Allowlisting limits the agent to explicitly approved tools such as local_pytest or gh, reducing the blast radius of mistakes or prompt-driven misuse. The evaluate_tool_allowlist check fails closed when unapproved tools are configured.

When should I not run agent workloads in AI Studio Playground?▼

Avoid Playground for production work because a single prompt can loop until you name a stop condition, consuming unbounded tokens. Use explicit termination conditions and verified execution checks instead.

Related Skills

Explore More Software Engineering Skills→
basedgod55hjlbasedgod55hjl

create-agent

Create a custom agent definition file (.agent.md) for a specialized job.

Community
Basic
harnessharness
80

enforce-slsa

Add an SLSA Verification step to Harness pipelines for provenance attestations.

Official
Advanced
Lingtai-AILingtai-AI
692

token-usage

Analyzes LLM token costs, cache rates, and tool-call trends across LingTai agent logs.

Official
Intermediate
BLTGVBLTGV
2

rye-installer

Automate setup, migration, and validation of Rye PostgreSQL deployments.

Official
Intermediate