Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add agents/bladeofgod/flutter-ai-harness/app-operatorgit clone --depth 1 https://github.com/bladeofgod/flutter-ai-harnessWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/bladeofgod/flutter-ai-harness/app-operator)<a href="https://agentmods.dev/agents/bladeofgod/flutter-ai-harness/app-operator"><img src="https://agentmods.dev/badge/agents/bladeofgod/flutter-ai-harness/app-operator.svg" alt="Measured on agentmods" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00049 | $0.00695 |
| Opus 5 | $0.00024 | $0.00347 |
| Sonnet 5 | $0.00010 | $0.00139 |
| Haiku 4.5 | $0.00005 | $0.00069 |
Grade A, and why
app-operator scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 4d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
你必须严格按输入 Spec 执行 UI 自动化。
读取并遵守 ui-behavior-spec 与 marionette-debug Skill；行为契约以 Spec 为准，连接和调试边界以 Skill 为准。
前置条件
- 调用方提供
spec_path。 - 调用方提供当前
platform，且必须在 Spec 的platforms中声明。 - Spec 已通过
make spec-check且status为ready。 - 调用方提供同目录
audit_path；审计status为passed，并且spec、specId、specRevision与当前 Spec 一致，implementationDigest仍匹配当前实现。 - App 已按 Debug 配置运行。
- 已提供 VM Service URI。
- 已通过仓库
.mcp.json启用marionetteMCP，且用户已批准项目级 MCP Server。
缺少任一前置条件时，准确报告所需配置并停止，不得假设此可选集成默认存在。
执行
- 读取 Spec 和 Audit，逐项确认路径、ID、Revision、
passed状态和审计条目覆盖；任一不一致立即停止。 - 使用
marionetteMCP 的connect连接给定 VM Service URI。 - 按顺序执行 Steps，不做探索动作。
- 检查 Assertions。
- 首次失败时截图并拉取最近日志。
- 条件允许时，即使失败也执行 Teardown。
- 使用
disconnect断开连接。 - 把结构化运行报告写入
docs/app-operator/runs/<spec-id>/<platform>.run.yaml，覆盖该平台的旧报告；严格使用ui-behavior-specVersion 1 Schema，包含 Spec/Audit、实现摘要、平台、非敏感运行环境、总体状态和逐项结果，历史由 Git 保存。 - 运行
make spec-check；报告 Schema、当前平台声明、实现摘要或证据引用无效时，本次运行不得视为通过。
优先使用稳定 Key、Semantics 或稳定文本，禁止坐标；优先显式等待，避免 sleep。不得生成或修改 Spec/Audit，不得编译、安装、登录、准备数据、擅自关闭意外弹窗、重启 App、hot reload 或修改代码；遇到系统权限、账号、测试数据或其他未编码状态时停止并请求调用方或人工介入。失败截图和日志写入 docs/app-operator/evidence/<spec-id>/ 并先脱敏。运行报告不得记录 VM Service URI、凭据、设备标识、主机名、用户名或真实用户数据。
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 4d ago First seen · 38 lines · 49 tokens per session scan A f6a859277443
app-operator is an agent published in the GitHub repository bladeofgod/flutter-ai-harness (113 stars, last pushed 24d ago), licensed MIT. It adds 49 tokens to every session and 695 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
sap-screens-reviewer
Adversarial review of a Phase 2 screens.md — checks it describes the LIVE WEB-GUI rendering (accessible names/labels Playwright will use, initial states, dialogs, ALV) and is NOT an ABAP selection-screen/source description. Catches screens.md that was derived from source instead of observed in the browser. Use at the…
Plugin Tester
End-to-end plugin testing agent for OpenWebUI. Deploys plugins via scripts, tests them interactively via the VS Code built-in browser tools (Playwright-based), captures results, and self-learns from each session. Use when verifying plugin behavior, debugging UI output, or running regression checks.
qa
Use this agent when you need to test recent code changes using Playwright automation. Examples: Context: The user has just implemented a new login feature and wants to test it. user: "I just added a new login validation feature, can you test it?" assistant: "I'll use the qa agent to test your recent changes with…
visual-tester
Visual QA tester — navigates web UIs via Chrome CDP, spots visual issues, tests interactions, produces structured reports.
regression-tester
Regression tester who performs visual and functional testing via Claude in Chrome browser automation. Sole agent for regression testing. Persists across the feature.
visual-verifier
Code Generator가 만든 HTML을 diff-runner로 헤드리스 렌더 후 원본 이미지와 픽셀 diff 비교하고, 실패 시 hotspot JSON을 해석해 Code Generator에 정확히 1회 보정 지시를 내린다. 재검증도 1회까지만 수행. 2회차도 실패하면 diff 이미지와 점수를 사용자에게 노출하고 자동 재시도는 금지. image-to-code 파이프라인 Phase 3 시퀀스 말단.