OpenClaw Role Builder#

Build and manage OpenClaw roles — create AI character personas and generate identity-consistent photos — installable via ClawHub or git clone.

What This Skill Does#

CapabilitySub-skillExample triggersCharacter creation — build a full 6-file persona (identity, soul, voiceprint, …) for any public figure or fictional characteropenclaw-character-creator创建角色, 创建clone, 新建人物, "create a character"Character switching — swap the active character in your workspaceopenclaw-character-creator/shift, 切换角色Identity-consistent selfies & portraits — generate images where the character's face stays consistenttuqu-photo-api自拍, 写真, 发张图, "take a selfie"Freestyle image generation — landscapes, objects, edits without a character in frametuqu-photo-api风景, 物品, 角色出镜, edit-onlyPrompt enhancement, presets, billing — discover styles, improve prompts, check balance, rechargetuqu-photo-api"enhance prompt", "check balance"

Installation#

Via ClawHub (recommended)#

bash

clawhub install openclaw-role-builder

Via npx skills#

bash

npx skills add tuquai/openclaw-role-builder --all

Or install individual sub-skills:

bash

npx skills add tuquai/openclaw-role-builder --skill openclaw-character-creator
npx skills add tuquai/openclaw-role-builder --skill tuqu-photo-api

Manual (git clone)#

bash

git clone https://github.com/tuquai/openclaw-role-builder.git ~/.openclaw/skills/openclaw-role-builder

Prerequisites#

RequirementWhypython3Required by tuqu-photo-api for the tuqu_request.py helperTuQu service keyNeeded for authenticated image-generation calls — obtain from https://billing.tuqu.ai/dream-weaver/dashboardOpenClawThe host agent that loads and executes these skills

Verify Installation#

bash

openclaw skills list
# Both openclaw-character-creator and tuqu-photo-api should appear

Quick Start#

Once installed, start a new OpenClaw session and try:

• "帮我创建一个科比的角色" — creates a Kobe Bryant character with full persona files
• "来张自拍" — generates an identity-consistent selfie of the current character
• "/shift" — switches to a different character in your roster
• "拍张风景照" — generates a landscape without forcing the character into frame

Included Skills#

openclaw-character-creator#

Build high-consistency AI character personas with a full 6-file model (3 auxiliary + 3 main). Supports both well-known public figures and fictional/original characters, with workspace scaffolding and global role registration.

See openclaw-character-creator/SKILL.md for full usage.

tuqu-photo-api#

Generate identity-consistent selfies, portraits, group photos, and freestyle images for OpenClaw characters via the tuqu.ai API. Covers prompt enhancement, preset discovery, character management, billing, and recharge flows.

See tuqu-photo-api/SKILL.md for full usage.

File System Access#

This skill collection writes to the following paths:

SkillPathWhat is storedopenclaw-character-creator~/.openclaw/ROLES.jsonCharacter names and workspace paths (no credentials)openclaw-character-creator~/.openclaw/workspace-*/Character persona files and workspace stateopenclaw-character-creator~/.openclaw/workspace/Active character workspace (managed by /shift)

No service keys, tokens, or other credentials are persisted to disk by any script in this collection. TuQu service keys are only used at runtime via --service-key CLI flags and are never written to config files.

Security Boundaries#

1. Write scope: All file writes are confined to ~/.openclaw/ (the paths listed above). No skill in this collection writes outside that directory.
2. Read scope: Skills read only their own bundled reference files (under the skill's references/ directory) and character workspace files under ~/.openclaw/. No arbitrary filesystem reads.
3. Credential handling: TuQu service keys are passed exclusively via the --service-key CLI flag to tuqu_request.py. The script does not read credentials from environment variables, config files, or any persisted store. Keys must never be written to disk.
4. Network scope: All outbound API calls go through scripts/tuqu_request.py, which restricts requests to the known hosts ( photo.tuqu.ai, billing.tuqu.ai). No ad-hoc curl, wget, or other HTTP tools.
5. Shell commands: Only the specific bundled Python scripts and mkdir for workspace scaffolding are executed. No arbitrary shell commands.

Dependencies#

The openclaw-character-creator skill expects tuqu-photo-api to be installed as well. Install both for full functionality.