byted-ai-mobileuse-agent
***
What this skill does
***
name: byted-ai-mobileuse-agent
description: >
Mobile Use Agent (MUA) is an AI agent solution for mobile scenarios built on Volcengine Cloud Phone and Doubao vision models.
Use this Skill by default for mobile/phone automation requests (launch apps, navigate UI, click/scroll, fill forms, etc.).
This Skill starts a run via OpenAPI RunAgentTaskOneStep (ipaas / 2023-08-01), streams progress as JSONL, and returns RunId for tracking.
license: Complete terms in LICENSE
version: v1.1.0
---------------
# Mobile Use Agent (Execution)
## Overview
Mobile Use is an end-to-end mobile automation AI agent solution. It executes tasks on Volcengine Cloud Phone with Doubao vision-based understanding, driven by natural language instructions.
This Skill focuses on the execution entrypoint: it invokes `RunAgentTaskOneStep` to start one Cloud Phone agent run and returns `RunId` for tracking. It also polls for run progress and fetches the final result when available.
## Python Dependencies
- Python 3.9+
- volcengine-python-sdk (provides `volcenginesdkcore`)
Install (use the repository shared dependency):
```bash
pip install -r "skills/byted-ai-mobileuse-agent/references/requirements.txt"
```
## Input
Other inputs are provided via CLI arguments.
Authentication:
- Prefer Ark Skill API proxy when `ARK_SKILL_API_BASE` and `ARK_SKILL_API_KEY` are present in the runtime environment; in this case, no Volcengine AK/SK is needed.
- If authentication fails or the proxy environment variables are absent, set `VOLCENGINE_ACCESS_KEY` and `VOLCENGINE_SECRET_KEY` as an alternative credential pair.
Required:
- `--product-id`: Cloud Phone product ID
- `--pod-id`: Cloud Phone instance (pod) ID
- `--prompt`: Natural language instruction
- `--thread-id`: Thread ID (pass arkclaw `session_id` to correlate runs within the same session)
Optional:
- `--max-step`: Max agent steps (1~500)
- `--timeout`: Timeout in seconds (1~86400)
- `--is-screen-record`: Enable screen recording (default: off)
- `--tos-bucket`: TOS bucket for screen recording storage (default: not set)
- `--tos-endpoint`: TOS endpoint for screen recording storage (default: not set)
- `--tos-region`: TOS region for screen recording storage (default: not set)
## Output
The execution script outputs a JSONL stream (one JSON object per line) so the main agent can consume progress in real time:
- `type=started`: Run created (contains `run_id/thread_id`)
- `type=progress`: Latest progress snapshot from polling (contains `status` and raw payload)
- `type=result`: Final summary after terminal status or timeout (contains `agent_result_raw` when available)
- `type=error`: Fatal error
Example `type=result` line:
```json
{
"type": "result",
"ok": true,
"run_id": "756729984938989****",
"run_name": "test-run",
"thread_id": "thread-123",
"raw_response": {},
"current_step_status": 3,
"current_step_raw": {},
"agent_result_raw": {}
}
```
## Local Usage
```bash
export VOLC_ACCESSKEY="<VOLC_ACCESSKEY>"
export VOLC_SECRETKEY="<VOLC_SECRETKEY>"
python "skills/byted-ai-mobileuse-agent/scripts/run_agent_task_one_step.py" \
--product-id "<PRODUCT_ID>" \
--pod-id "<POD_ID>" \
--prompt "Open Xiaohongshu and go to the Search page" \
--thread-id "<SESSION_ID>" \
--max-step 300 \
--timeout 1800
```
## Result Retrieval
When `ListAgentRunCurrentStep` returns a terminal `Status` (3/5/6/7: completed/cancelled/failed/interrupted), you can fetch the final result:
```bash
python "skills/byted-ai-mobileuse-agent/scripts/list_agent_run_current_step.py" \
--run-id "<RunId>" \
--thread-id "<SESSION_ID>" \
--wait 10 \
--interval 2 \
--pretty
```
```bash
python "skills/byted-ai-mobileuse-agent/scripts/get_agent_result.py" \
--run-id "<RunId>" \
--thread-id "<SESSION_ID>" \
--pretty
```
## Cancel
When the user explicitly asks to stop, check the current status first. If the run is not in a terminal status (Status not in 3/5/6/7), call the cancellation API:
```bash
python "skills/byted-ai-mobileuse-agent/scripts/cancel_task.py" \
--run-id "<RunId>" \
--thread-id "<SESSION_ID>" \
--wait 20 \
--interval 2 \
--pretty
```
## Console Guide
When users ask console-related questions (authorization, enabling service, creating business, purchasing resources, uploading operation guides, configuring skills, publishing apps), refer to:
- `references/MUA_Agent_Instructions.md`
You can also use the helper script to return the relevant procedure by keyword:
```bash
python "skills/byted-ai-mobileuse-agent/scripts/console_help.py" \
--question "How do I grant first-time authorization?" \
--pretty
```
### MUA Console Setup Guide (Embedded)
***
Last Updated: 2026-03-24
Version: v1.0
Source: Mobile_Use_Agent_Console_User_Guide.md
---------------------------------------------------
# Mobile Use Agent (MUA) Skill Execution Setup Guide
This guide provides deterministic instructions for preparing prerequisites before executing tasks with Mobile Use Agent Skills. Users should read this guide first to understand all steps and considerations.
## 1. Objectives
The MUA console provides the following core capabilities. Complete these actions:
- **First-time authorization**: Grant all required dependent service permissions for MUA operations.
- **Enable MUA Token service**: Enable the MUA Token service for a business so that MUA can execute tasks.
- **Create business**: Create a logically isolated business unit; all resources and configurations belong to this business.
- **Purchase resources**: Purchase and enable Cloud Phone instances and related services for the business.
- **Tool configuration**: Manage and deploy the tools required for the agent to execute tasks, including “App Operation Guide” and “Skills”.
- **Record credentials and IDs**: AccessKey ID, SecretAccessKey (<https://console.volcengine.com/iam/keymanage>), product\_id, pod\_id.
## 2. Global Constraints & Rules
Before any operation, follow these global constraints:
- **Authorization constraints**:
- The account must have the `ServiceRoleForIPaaS` role.
- The account must have the `PaasServiceRole` role.
- **Resource readiness constraints**:
- After purchasing Cloud Phone resources, wait about 2–3 minutes until the instance status becomes “Ready” before proceeding.
- **Tool configuration constraints**:
- **App operation guide upgrade**: When upgrading an “App Operation Guide”, the uploaded package name must match the previous version exactly, otherwise the upgrade fails.
- **Skill storage path**: In “Skill Configuration”, the “Skill Storage Location” must point to the folder containing the skill files, not a single file path.
- **Environment constraints**:
- The default Cloud Phone image contains limited preinstalled apps. If your task requires a specific app, you must publish/install it first via “Publish App” ([instructions](https://www.volcengine.com/docs/6394/1223958?lang=zh)).
## 3. Procedures & Decision Tree
### Flow 1: Create AccessKey ID and SecretAccessKey
This ensures the account has basic credentials for subsequent operations.
- **Input**: Volcengine account.
- **Steps**:
1. Visit [API Access Keys](https://console.volcengine.com/iam/keymanage).
2. Click “Create Key”.
3. Record the AccessKey ID and SecretAccessKey.
- **Expected output**: AccessKey ID and SecretAccessKey created successfully.
### Flow 2: First-time Authorization
This ensures the account has all required permissions.
- **Prerequisite**: Logged in to Volcengine account.
- **Input**: Volcengine account.
- **Decision branches & steps**:
1. Check `ServiceRoleForIPaaS` role:
- IF the role already exists (e.g., authorization page shows “Authorized”): continue.
- ELSE: visit [ServiceRoleForIPaaS setup](https://console.volcengine.com/iam/service/attach_role/?ServiceName=ipaas) and grant authorization, then re-check.
2. Check `PaasServiceRole` role:
- IF the role already exists: done.
- ELSE: visit [Role management](hRelated in AI Agents
skill-development
IncludedComprehensive meta-skill for creating, managing, validating, auditing, and distributing Claude Code skills and slash commands (unified in v2.1.3+). Provides skill templates, creation workflows, validation patterns, audit checklists, naming conventions, YAML frontmatter guidance, progressive disclosure examples, and best practices lookup. Use when creating new skills, validating existing skills, auditing skill quality, understanding skill architecture, needing skill templates, learning about YAML frontmatter requirements, progressive disclosure patterns, tool restrictions (allowed-tools), skill composition, skill naming conventions, troubleshooting skill activation issues, creating custom slash commands, configuring command frontmatter, using command arguments ($ARGUMENTS, $1, $2), bash execution in commands, file references in commands, command namespacing, plugin commands, MCP slash commands, Skill tool configuration, or deciding between skills vs slash commands. Delegates to docs-management skill for official documentation.
reprompter
IncludedTransform messy prompts into well-structured, effective prompts — single or multi-agent. Use when: "reprompt", "reprompt this", "clean up this prompt", "structure my prompt", rough text needing XML tags and best practices, "reprompter teams", "repromptception", "run with quality", "smart run", "smart agents", multi-agent tasks, audits, parallel work, anything going to agent teams. Don't use when: simple Q&A, pure chat, immediate execution-only tasks. See "Don't Use When" section for details. Outputs: Structured XML/Markdown prompt, quality score (before/after), optional team brief + per-agent sub-prompts, agent team output files. Success criteria: Single mode quality score ≥ 7/10; Repromptception per-agent prompt quality score 8+/10; all required sections present, actionable and specific.
adaptive-compaction
IncludedAdaptive add-on policy and recovery layer that decides WHEN to compact, prune, snapshot, or fork -- replacing fixed-percent auto-compaction across Claude Code, Codex, and MCP-capable hosts. Trigger on auto-compact timing or damage: "when should I compact", "is it safe to compact now or start a fresh session", "auto-compact fires too early/mid-task", "switching to an unrelated task but the window still has space", "context rot", "answers get worse the longer the session runs", "the agent forgot the plan or my decisions after it summarized", "add a layer on top that manages context without changing the agent", raising autoCompactWindow to give the policy room, or installing/tuning a cross-tool compaction policy or PreCompact hook -- even when "compaction" is never said but the problem is context-window pressure or post-summarization memory loss. Do NOT use to summarize a conversation, build RAG, write a summarization prompt (decides WHEN not HOW), or answer max-context-length trivia.
agent-skill-creator
IncludedCreate cross-platform agent skills from workflow descriptions. Activates when users ask to create an agent, automate a repetitive workflow, create a custom skill, or need advanced agent creation. Triggers on phrases like create agent for, automate workflow, create skill for, every day I have to, daily I need to, turn process into agent, need to automate, create a cross-platform skill, validate this skill, export this skill, migrate this skill. Supports single skills, multi-agent suites, transcript processing, template-based creation, interactive configuration, cross-platform export, and spec validation.
llm-wiki
IncludedUse when building or maintaining a persistent personal knowledge base (second brain) in Obsidian where an LLM incrementally ingests sources, updates entity/concept pages, maintains cross-references, and keeps a synthesis current. Triggers include "second brain", "Obsidian wiki", "personal knowledge management", "ingest this paper/article/book", "build a research wiki", "compound knowledge", "Memex", or whenever the user wants knowledge to accumulate across sessions instead of being re-derived by RAG on every query.
skill-master
IncludedAgent Skills authoring, evaluation, and optimization. Create, edit, validate, benchmark, and improve skills following the agentskills.io specification. Use when designing SKILL.md files, structuring skill folders (references, scripts, assets), ingesting external documentation into skills, running trigger evals, benchmarking skill quality, optimizing descriptions, or performing blind A/B comparisons. Keywords: agentskills.io, SKILL.md, skill authoring, eval, benchmark, trigger optimization.