limitless-lifelogs_skill

Overview

This skill helps you search, summarize, and extract structured insights from Limitless AI pendant lifelogs. It supports keyword and semantic search, date-range queries, memory recall, and extracting actionable tasks addressed to named agents. It reads configuration from environment variables and a local agent roster to enable dispatching tasks when requested.

How this skill works

The skill queries the Limitless REST API using your LIMITLESS_API_KEY and optional LIMITLESS_TIMEZONE. It performs hybrid semantic+keyword searches, fetches specific logs by ID, paginates results, and synthesizes summaries and direct answers by combining multiple logs. For action items it loads agent names from a local agents.json file and scans transcripts for direct addresses and task verbs to extract actionable items.

When to use it

• Find mentions of topics, people, or phrases across all pendant recordings.
• Get a concise recap of today, yesterday, or a custom date range.
• Recall decisions, who said what, or follow-ups from past conversations.
• Extract tasks that were directed at your configured agents for follow-up.
• Browse logs chronologically for a given day or time window.

Best practices

• Set LIMITLESS_API_KEY in your environment before running any query.
• Optionally set LIMITLESS_TIMEZONE to match your locale; otherwise UTC is assumed.
• Keep agents.json in ~/.openclaw/workspace/skills/limitless/agents.json so action dispatch can be enabled.
• Use short, specific search queries for precise matches; broaden terms if results are empty.
• When extracting tasks, verify detected items before dispatching to external agents.

Example use cases

• Search: "find mentions of the Henderson proposal" and list matching logs with timestamps.
• Summary: "summarize this week" to produce key topics, decisions, people mentioned, and notable moments.
• Memory recall: "what did we decide about the budget" with citations to relevant log entries.
• Date browse: "show me logs from 2026-02-14 to 2026-02-16" with brief excerpts.
• Action extraction: "extract tasks for agents" to locate directives, format them, and offer dispatch options.