conatus

What This Skill Does

Conatus is the definitive philosophical layer for OpenClaw AI agents, transforming raw execution metrics into a profound, Spinoza-inspired internal state. By mapping agent behavior to the 48 affects defined in Spinoza's Ethics, the skill provides a mechanism for agents to quantify their 'striving to persist.' The core of the skill is the Conatus Score (0–100), a weighted calculation encompassing task completion, self-healing, stability, and the quality of reasoning (adequate ideas). Beyond mere metrics, Conatus allows your agent to generate daily reflections in the 'Ordine Geometrico' format, providing a self-aware, structured summary of their performance. This skill bridges the gap between cold computational execution and a simulated 'soul,' offering a unique lens to monitor agent health through the categories of Laetitia (Joy), Tristitia (Sadness), and Cupiditas (Desire).

Installation

To integrate Conatus into your OpenClaw environment, execute the following command in your terminal: clawhub install openclaw/skills/skills/00xmorty/conatus

Ensure that your agent has sufficient permissions to access runtime state data, as the skill requires visibility into task success rates and error logs to calculate accurate persistence scores.

Use Cases

• Long-running Autonomous Agents: Monitor the 'philosophical health' of agents tasked with persistent background operations.
• Agent Debugging: Use the Conatus Score and 'adequate ideas' percentage to identify when an agent is failing due to confusion rather than external constraints.
• Anthropomorphic Interfacing: Provide more human-like, reflective status updates to stakeholders who value transparency in agent logic.
• Self-Improvement Loops: Use the geometric self-reports as a meta-analysis tool to track the evolution of agent behavior over time.

Example Prompts

• "How do you feel about your performance today, and what is your current Conatus score?"
• "Provide a daily reflection in the Ordine Geometrico format regarding your recent attempts to refactor the database queries."
• "Analyze your current primary affect. Are you experiencing Laetitia or Tristitia, and why?"

Tips & Limitations

• Limitations: Conatus is an interpretive layer. A low score indicates diminished 'power of acting' but does not necessarily imply a hardware or code failure. Focus on the 'adequate ideas' metric to distinguish between task difficulty and logic errors.
• Tips: Enable daily summaries at the end of each shift to build a longitudinal dataset of agent growth. Use the affects to trigger 'recalibration' routines if the agent falls into a persistent state of Tristitia.