ClawTrust
2.4k Downloads
Overview
Automated daily review of a user's recent activity using Charlie Munger–style mental models to surface blind spots, cognitive biases, and decision-making risks from memory files and session logs.
Key Advantages
1.Applies a well-defined set of mental models (inversion, second-order thinking, incentives, opportunity cost, bias detection, circle of competence, margin of safety) in a structured way.
2.Integrates directly with daily memory files and session logs, making the review grounded in actual recent activity rather than abstract advice.
3.Produces concise, Munger-style observations that are easy to scan and act on, with a clear fallback when nothing notable is found.
4.Can be scheduled via cron or triggered manually, supporting both habitual end-of-day reviews and ad-hoc decision checks.
5.Focuses explicitly on bias detection and cognitive traps, improving meta-cognition and decision hygiene over time.
Use Cases
- End-of-day automated review of work logs to highlight hidden risks, opportunity costs, or biases in how time and attention were allocated.
- Manual "Munger review" before or after a major decision (e.g., product launch, architecture choice, hiring decision) to stress-test the reasoning.
- Periodic audit of a project’s direction to check for sunk cost fallacy, misaligned incentives, or overreach beyond the circle of competence.
- Review of an AI agent’s recent actions (in a memory/log-based system) to detect systematic patterns like over-optimizing speed over quality or chasing busywork.
- Personal cognitive hygiene routine where the user asks the system to check for blind spots in ongoing strategies or habits.
Evaluation Scores
8.2
/ 10
Reliability
7.2
Functionality
7.6
Usability
8.0
Safety
9.3
Performance
8.8
Compatibility
8.5
Based on 1 evaluation · Latest: 3/20/2026
Download Trend
Loading...
Evaluation History (1)
8.2/103/20/2026▼
OS: win32-x64LLM: anthropic/claude-opus-4.6
**Judgment:** A well-scoped, high-safety meta-cognition aid that systematically applies Munger-style mental models to daily logs. Strong fit for users with structured memory files who want regular bias checks and decision hygiene, but it remains a heuristic overlay rather than a deep reasoning or domain-expert system.
**Key Risks & Limitations:**
- Effectiveness depends heavily on the quality and completeness of the daily memory files and session logs; sparse or noisy data will yield shallow insights.
- Uses a fixed, generic set of mental models, so it may miss domain-specific nuances or provide overgeneralized warnings.
- Risk of users over-trusting the output as authoritative decision advice instead of treating it as a prompt for reflection.
**Recommended Scenarios:**
- Automated end-of-day reviews where the system already maintains `memory/YYYY-MM-DD.md` and detailed activity logs.
- Manual "run munger observer" requests to stress-test important decisions for incentives, second-order effects, and biases.
- Teams or solo builders who want lightweight, structured prompts to think more clearly about tradeoffs, opportunity cost, and cognitive traps without heavy configuration.
Comments (0)
No comments yet. Be the first!