Proactive Tasks

**Quick judgment** Proactive Tasks is a strong, production-oriented skill for turning OpenClaw agents into proactive project partners. It shines in long-running, goal-driven workflows where reliability and recoverability under context truncation are crucial. Best suited for technical users comfortable with CLI tools and cron/heartbeat setups. **What it does well** - Provides a clear model for goals, tasks, priorities, dependencies, and task states. - Enables agents to work during heartbeats (autonomous mode) and update humans only when it’s meaningful (completion, blockers, discoveries). - Implements robust reliability patterns: WAL-based write-ahead logging, SESSION-STATE.md as canonical working memory, working-buffer.md for high-context "danger zone" logging, and a structured compaction recovery protocol. - Offers self-healing via `health-check` to detect and auto-fix common data issues, improving long-term integrity. - Enforces a verification-before-reporting norm to reduce spurious “done” reports and improve trust. **Key risks / limitations** - **Setup complexity:** Requires configuring heartbeats/cron, understanding HEARTBEAT.md/HEARTBEAT-CONFIG.md, and wiring the CLI into your workflow; non-trivial for less technical users. - **Single-file storage:** All task data in `data/tasks.json` plus WAL/logs is simple but may become less convenient for very large or highly concurrent setups. - **OpenClaw-centric design:** The patterns assume the OpenClaw heartbeat + workspace file model; portability to other agent stacks will need adaptation. **Recommended scenarios** - You want an agent that continues making progress on well-defined goals between your sessions, without constant prompting. - Your workflows are long-running and susceptible to context truncation, and you need robust state recovery and auditability. - You are comfortable with CLI tools, cron jobs, and file-based configuration, and you want a conservative, reliability-first approach to autonomous task management.