build-persona-vaultlisted

# Build a PersonaVault Create one living knowledge asset without merging platform collectors into a brittle monolith. Keep acquisition platform-specific; normalize, merge, render, and verify through the shared PersonaVault protocol. ## Resolve the runtime 1. Resolve the repository root two directories above this `SKILL.md`; require its `package.json` name to be `persona-vault`. 2. Use Node.js 22 or newer. Invoke the CLI as `node <repo-root>/bin/personavault.mjs` when `personavault` is not on `PATH`. 3. Use an absolute local vault directory. Never store credentials, sessions, browser state, or raw private archives inside the repository. 4. Read [references/connectors.md](references/connectors.md) before collecting. Read [references/protocol.md](references/protocol.md) before changing the canonical schema. Read [references/recovery.md](references/recovery.md) after any partial or failed run. ## Lock the person and sources Treat the person, platform account, and content item as different identities. Confirm each platform's stable account ID and display name. Ask only when the evidence cannot establish that multiple accounts belong to the same person. Initialize once: ```bash personavault init --name "Display Name" --output "/absolute/vault/path" ``` Reuse the same vault on later runs. Refuse silent person-ID or source-account substitution. ## Collect through platform connectors - For WeChat public accounts, use `$mpvault-archive` or its `mpvault` CLI. Preserve its `man