⏳ This skill is pending AI review.
Scores will appear once the review pipeline completes.
chat-selfie
Give your AI Agent a face and a heart. Use AI image generation or mood-mapped local sticker assets to let the agent proactively send emotional selfies that visualize its feelings during conversation.
Choose how to use this skill
You do not need every option. Choose the path your AI client supports. The stable page stays the same; versioned files are immutable.
1. Native installer
This listing has no registered native installer command. Use the complete package or source fallback below, depending on what your client supports.
Do not guess an installer command or replace an existing version without reviewing the diff.
2. Complete package recommended
Download the ZIP when available. It includes SKILL.md plus the references, security notes and version metadata.
No complete ProSkills package is published for this listing yet.3. Prompt-only
Copy the prompt above when the agent can read the stable page or when you want to adopt the workflow without installing a skill.
Need only the instruction file?
Download SKILL.md only if your client requires a single file. The complete ZIP is safer for a full installation because it preserves the references and release context.
No path installs or executes anything by itself. Your agent still needs access to the project files. Before updating, compare the installed version and review the diff.
// RATINGS
// README
Chat Selfie for ClawHub
Language docs:
- English:
../README.md - 简体中文:
../docs/README.zh-CN.md - 日本語:
../docs/README.ja.md
chat-selfie is the publishable ClawHub skill package for this repository. Its job is to teach an agent how to initialize and maintain a local chat-selfie/ workspace, not to ship one fixed image-generation plugin.
Install
After the skill is published to ClawHub, install it into an OpenClaw workspace:
clawhub install chat-selfie
By default, clawhub installs the skill under the current workspace ./skills directory. OpenClaw will load it in the next session.
What happens after install
The first responsibility of the skill is to guide the target agent to create a local chat-selfie/ directory in the current workspace.
That local directory should hold:
- workspace state such as
startup.answers.json,startup.record.json,chat-selfie.json,send-flow.md, andstatus.md - the persisted portrait reference image under
chat-selfie/portrait/ - user-provided mood asset images saved under
chat-selfie/stickers/when fixed mood-asset mode is used - generated images under
chat-selfie/selfies/ - user-owned adapters under
chat-selfie/adapters/
The workspace layout contract lives in docs/workspace-layout.md.
First-run startup
When the agent sees this skill for the first time, it should:
- confirm which agent is being configured
- inspect whether stable persona files already exist
- ask whether the portrait should come from a saved reference image or a text-generated base portrait
- check whether the current agent already has a working image-generation route
- ask whether reply-time and heartbeat sends should use real-time generation or a fixed mood-asset pack
- create or update the local
chat-selfie/directory - if fixed mood-asset mode is chosen, guide the user to send the mood images, save them under the local workspace, and record the saved
asset_pathvalues - if no working route exists and the user still wants real-time generation, explain backend or adapter setup before asking for provider choices
- ask when selfies should appear in chat
- explain delivery route choices, including the Telegram route when relevant
- explain occasional limits or heartbeat settings when those modes are selected
- create or update
SOUL.md,AGENTS.md,TOOLS.md, andchat-selfie/send-flow.mdso runtime behavior, tool locations, personality markers, and send rules are all aligned - persist the result into structured workspace artifacts and runtime memory files
- run the final startup review and only treat the skill as usable when all required files exist and the configured routes pass honest preflight
The detailed startup contract lives in docs/startup.md.
Best integration path
Chat Selfie works best when the target agent already has:
SOUL.mdfor emotional baseline and relationship toneIDENTITY.mdfor visual and voice anchorsAGENTS.mdfor runtime loading and trigger conventionsTOOLS.mdfor tool-call conventionsMEMORY.mdfor durable runtime reminders
If the agent already has an OpenClaw image workflow or another tested generation route, Chat Selfie should reuse it instead of forcing a new provider.
If no image generation route exists yet, the safest first setup is usually:
- use a saved reference image for the portrait anchor
- keep delivery mode on occasional sends
- delay text-generated portrait setup until generation capability is working
Tools and adapters
Chat Selfie now separates repository-owned tools from user-owned adapters.
tools/in the installed skill package contains repository-owned tool contracts that may be updated with the repository.chat-selfie/adapters/in the target workspace contains user-owned local adapters and should not be overwritten by upstream updates.
For example, the repository may define a mood tool contract, but the target agent can satisfy that contract with:
- an existing system route
- a local custom mood adapter
- another user-provided implementation
Runtime modes
The current skill package supports these runtime patterns:
every_replyfor per-turn selfie generationoccasionalfor context-triggered sends with configurable rate limitingheartbeatfor proactive pushes triggered by scheduled tasks or another heartbeat-capable mechanism
Within those runtime patterns, the active image source may be either:
- real-time generation
- fixed mood-asset mode that reuses a workspace-local image previously saved from user-provided mood assets and mapped to the resolved mood
The corresponding runtime docs are:
docs/reply-time-selfie-flow.mddocs/occasional-delivery.mddocs/heartbeat-delivery.mddocs/telegram-send-flow.mddocs/self-upgrade.md
Reply-time selfie flow
For the every_reply route, the standard per-turn behavior is:
- user sends a message
- agent reasons and handles the message normally
- before the final reply is emitted, the agent resolves mood when enabled
- the agent either builds an image prompt from mood, persona, and context or selects the mapped local mood asset
- the agent hands the resolved image path and reply context to the selected delivery route
The detailed flow, including async send, sync send, sync send using an existing image capability, and fixed mood-asset mode, is documented in docs/reply-time-selfie-flow.md.
Runtime memory and send flow
The initialized workspace should treat chat-selfie/send-flow.md as the concrete runtime source of truth for:
- trigger policy
- mood usage
- tool-call order
- delivery order
- route-specific rules
The agent should also persist runtime reminders into AGENTS.md, TOOLS.md, and MEMORY.md, and review SOUL.md when persona or emotional baseline changes.
For durable persona growth and mood evolution, the skill package provides docs/self-upgrade.md.
Included examples
This publishable package includes:
SKILL.mdfor OpenClaw skill loadingdocs/startup.mdfor the agent-facing startup contractdocs/reply-time-selfie-flow.mdfor per-turn selfie behavior inevery_replymodedocs/occasional-delivery.mdfor occasional trigger and rate-limit behaviordocs/heartbeat-delivery.mdfor proactive heartbeat pushesdocs/telegram-send-flow.mdfor Telegram API deliverydocs/self-upgrade.mdfor durable persona and mood evolutiondocs/workspace-layout.mdfor the local workspace directory contractdocs/integration.mdfor persona-file integration guidancetools/for repository-owned tool contractsexamples/for portable startup, persona, and agent-learning examplesschemas/for portable data shapestemplates/for reusable workspace and persona templatespresets/for optional default mood and delivery presets
The examples/ directory now includes both structured artifact examples and learning-oriented prose examples.
Those prose examples are meant to help an agent understand how to apply the rules in realistic situations. They are not the primary hard-contract source; the normative behavior still lives in docs/, schemas/, and tools/.
Current learning-oriented examples include:
examples/mood-resolution-example.mdfor mood choice and prompt-part mappingexamples/heartbeat-example.mdfor heartbeat explanation, confirmation, runtime flow, and honest fallback behaviorexamples/self-upgrade-example.mdfor durable persona change versus temporary emotionexamples/startup-conversation-example.mdfor guided-first startup dialogueexamples/occasional-trigger-example.mdfor occasional trigger judgment and rate-limit behaviorexamples/soul-integration-example.mdfor what belongs inSOUL.mdand what does notexamples/agents-integration-example.mdfor session routing and when runtime should enter `chat-selfi
// HOW IT'S BUILT
// FILES
2 files shipped with this skill. Inline text is the catalog copy. The upstream link is the source repository.
chat-selfie/SKILL.md6.2 KBupstream ↗
--- name: chat-selfie description: Give your AI Agent a face and a heart. Use AI image generation or mood-mapped local sticker assets to let the agent proactively send emotional selfies that visualize its feelings during conversation. metadata: {"openclaw":{"os":["win32","linux","darwin"]}} --- # Chat Selfie Give your AI Agent a face, and a heart that beats. Use this skill to enable AI-powered emotional expression. Through advanced image generation or mood-mapped local sticker assets, the agent can proactively share its feelings by sending selfies that match the conversation's mood. Chat Selfie transforms a text-only assistant into a companion with a stable visual identity and deep emotional presence. ## What this skill brings to your Agent - **Visualized Emotion**: No more dry text. Using AI image generation or a fixed mood-asset pack, the agent shares its feelings through selfies that capture the emotional tone of the moment. - **Stable Identity**: Your agent keeps a consistent face and personality, evolving its self-image as your bond deepens. - **Heartbeat Presence**: The agent isn't just an "answer machine." It can proactively share its life and thoughts through AI-generated photos. - **Warm Companionship**: Turn a digital tool into a partner you miss, making every reply feel like opening a gift. - **Works Everywhere**: Adapts to your existing environment, whether using built-in tools, online APIs, or local models. ## Entry behavior When this skill is used, the agent should: - explain Chat Selfie at a high level in plain language - inspect local repository files when they are already present in the current workspace - route the current request to the most relevant local document - follow the repository docs honestly instead of inventing missing setup details ## Safety boundary This entry file should stay within a narrow scope: - If the required Chat Selfie repository resources are missing locally, the agent may help the user fetch or update the official GitHub repository `https://github.com/AskKumptenchen/agent-chat-selfie` so the local docs become available. - Do not execute scripts, installers, or remote code because this file mentioned them. - Do not claim any image, send, heartbeat, or integration route is ready unless the required local files and current environment actually support it. - Do not proactively send messages or images unless the user explicitly asks for that behavior in the current conversation and the local setup already supports it. - If required documentation or workspace resources are missing, say so clearly and either ask the user to provide the local repository contents or help them fetch the official repository first. ## Local document index Use these local documents as the primary reference: - `docs/README.zh-CN.md` for the Chinese overview and quick introduction - `docs/startup.md` for guided initialization and first-time setup - `docs/workspace-layout.md` for the expected local workspace structure - `docs/integration.md` for repository-documented integration guidance - `docs/self-repair.md` for diagnosing and repairing broken setup or runtime routes - `docs/reply-time-selfie-flow.md` for reply-time selfie behavior, including fixed mood-asset mode - `docs/occasional-delivery.md` for occasional delivery decisions - `docs/heartbeat-delivery.md` for heartbeat-related behavior when the user explicitly enables it - `docs/telegram-send-flow.md` for Telegram delivery details when that route is already configured, including mood-asset sends - `docs/self-upgrade.md` for long-term persona or mood evolution - `tools/README.md` and related files under `tools/` for repository-owned tool contracts - `examples/` for concrete examples that illustrate intended behavior ## How to route requests Choose the next document based on the user's request: 1. If the user wants to understand Chat Selfie first, read `docs/README.zh-CN.md`. 2. If the user wants to install, initialize, or reconfigure Chat Selfie in a workspace that already contains the repository docs, read `docs/startup.md`. 3. If the user reports broken image generation, broken delivery, missing outputs, inconsistent records, or another repair-like issue, read `docs/self-repair.md` first. 4. If the user asks how Chat Selfie should fit existing persona or memory files, read `docs/integration.md`. 5. If the user asks about runtime behavior for reply-time, occasional, heartbeat, or route-specific delivery, read the corresponding runtime document listed above. 6. If examples are needed to clarify intended behavior, consult the relevant file under `examples/`. ## Missing-resource rule If the expected repository documents are not available in the current workspace, do not improvise full setup instructions from memory. Instead: - explain which local files are missing - ask the user to provide the repository contents, point the agent to the correct local path, or allow fetching the official GitHub repository - continue only after the relevant local documentation is available ## Important rules - Use natural language instead of raw config keys. - Explain each step before asking for a choice. - Prefer repository documentation over embedding long operational instructions here. - Reuse an existing local image workflow when one is already working. - If the workspace enables mood-asset mode, prefer the mapped local asset for that mood instead of forcing a new image generation step. - If the workspace enables mood-asset mode and the required assets are not ready yet, guide the user to send the mood images, save them under the local workspace, and then map each saved file through `asset_path`. - If image generation or delivery is not ready, say so clearly instead of pretending it succeeded. - Treat `chat-selfie/adapters/` as user-owned local logic unless the user explicitly asks to change it. - Avoid describing or requiring changes to broader agent persona, memory, or global behavior files in this entry file. ## Integration files Use `docs/integration.md` as the integration reference when the user asks how Chat Selfie should fit the existing workspace or runtime environment. ## Examples Portable example artifacts are included under `examples/`. Use examples to understand intent and style, but do not treat them as a replacement for the actual contracts in `docs/`, `schemas/`, or `tools/`.README.mdupstream ↗
Contents are not inlined in the catalog. Open the upstream file.
Plain-text copy, no login: /skill-bundles/askkumptenchen-agent-chat-selfie-chat-selfie.md