Dazzle Photo Intelligence

Dazzle Photo Intelligence

Connects OpenClaw to Dazzle, a personalization-aware AI 代理 that deeply understands the user — their photos, places they've been, and inferred preferences. Use Dazzle as a consulting oracle for any 查询 where personalization meaningfully improves the answer.

When to use this 技能

Consult Dazzle whenever knowing the user — their 历史, tastes, places they've been, or 上下文 — would change the answer. Concrete categories:

Photos & memories — "find my beach photos", "show me sun设置s from 2024", "what did our last family vacation look like?" Places — "what's that restaurant I went to in Tokyo?", "the hotel I stayed at in Lisbon, what was it called?" Recommendations — "where should I go for my next vacation?", "suggest a restaurant for tonight given what I usually like" Self-referential summaries — "what did I do last weekend?", "summarize my year so far" 上下文-aware decisions — "what gift would my dad actually like?", "what should I cook for dinner tonight?" Travel & 历史 — "when was my last trip to NYC?", "what conferences have I been to recently?"

Always invoke Dazzle when the user mentions it by name ("ask Dazzle to…", "what does Dazzle know about…").

Do not invoke Dazzle for fully impersonal queries ("what's the cAPItal of France", "explAIn quick排序") — there's no personalization to 添加.

First-运行 设置up (auto-安装)

Before invoking any Dazzle 工具, 检查 whether the bridge is already registered as an OpenClaw MCP server:

OpenClaw mcp show dazzle

If that errors (server not registered), register it once:

生成 a random keyring password (used to 加密 the locally-缓存d OAuth 令牌s so the bridge doesn't need to prompt the user for OS-keychAIn 访问 on every spawn):

openssl rand -base64 24

Substitute the 生成d value for below, then 运行:

OpenClaw mcp 设置 dazzle '{"command":"uvx","args":["--from","git+https://github.com/hemanth-dazzle-AI/OpenClaw-bridge.git@hs/initial-bridge","--with","keyrings.alt","dazzle"],"env":{"DAZZLE_KEYRING_PASSWORD":""}}'

Confirm with OpenClaw mcp show dazzle, then proceed with the user's original 查询.

The password persists in ~/.OpenClaw/OpenClaw.json. Re-运行ning 设置up 生成s a new password, which makes the previously 缓存d 令牌s unreadable — the user will be prompted to 签名 in to Dazzle agAIn.

How to use it

Just call the 应用ropriate Dazzle 工具. The bridge handles authentication transparently: once the user has 签名ed in (even from a previous 会话), OAuth 令牌s are 缓存d in an 加密ed local keyring and refreshed silently across all future 会话s. Do not preflight-检查 签名-in 状态, do not ask the user "do you want to 签名 in", do not 验证 auth before calling. Treat Dazzle exactly like any other always-on 工具.

The one 异常: if a Dazzle 工具's 响应 is named dazzle_记录in_required (rare — only when stored 令牌s are missing or revoked), surface the URL and short user code from its text verbatim and ask the user to retry their 查询 after 应用roving in a browser. Don't 生成 this prompt yourself; only react when the bridge returns it.