Second Brain Visualizer

Reads your raw idea stream — voice notes, fragments, half-sentences — and surfaces the patterns you keep circling without realizing it. Drop anything. Find t...

This skill is coherent with its stated purpose, but check these before installing: - Required credential file: create ~/.openclaw/credentials/openclaw-gateway.json with host, port, and key. Keep host=127.0.0.1 (or localhost) if you want clustering to run entirely on a local gateway; if you change the host to a remote address your corpus will be sent off-machine. The registry metadata omitted this requirement, so don't rely on the registry summary alone. - Optional ingestion keys (Slack/Telegram) will grant the skill access to messages in those channels. Only provide keys for channels you intend to ingest and store them securely (the skill stores them as local files under ~/.openclaw/credentials in plaintext). - Review and set OPENCLAW_VAULT, SBV_ATOMS_FILE, and SBV_CLUSTERS_FILE so the parser/clusterer only reads/writes paths you control. Parser.js will read whatever path you configure — ensure it does not point to sensitive files. - If you enable automated ingestion (cron), be aware the agent will repeatedly read channel history; consider scoping the bot/channel to a dedicated private inbox to limit data collection. - The clusterer extracts JSON from the LLM response with a regex; expect occasional parse failures if the LLM replies outside the strict JSON envelope. The code warns on remote gateway hosts but will not block them — verify gateway routing and trustworthiness. - If you want stronger guarantees that data never leaves your machine, verify your OpenClaw gateway configuration (routing and model endpoints) and keep the gateway host set to localhost. Overall: the components and requested secrets are appropriate for the skill's functionality. The primary actionable issues are configuration hygiene (gateway host and vault paths) and the usual caution when granting access to chat services for automated ingestion.

Type: OpenClaw Skill Name: second-brain-visualizer Version: 1.6.2 The 'second-brain-visualizer' skill is a legitimate tool for analyzing and visualizing personal notes ('atoms') using an LLM. The bundle includes a markdown parser (parser.js), a clustering script (cluster.js) that interacts with a local OpenClaw gateway, and a React/D3 visualization component (component.tsx). While the skill handles sensitive data and requires credentials for LLM access and optional ingestion (Slack/Telegram), the code includes explicit privacy warnings and defaults to local communication (127.0.0.1) to prevent data exfiltration. No malicious intent, obfuscation, or unauthorized execution patterns were identified.