What does dbs-decision do?
dontbesilent's Personal Decision System.
Turn any domain that requires long-term tracking (business, relationships, health, career, learning, investments…) into a local knowledge project. It features a four-layer structure, source tags, immutable snapshots, and a concept library that extracts patterns over time.
Trigger commands:
/dbs-decision/decision-system/case-filing/result-fillback/status-portrait
How It Works
1. Four Layers of Knowledge
- Raw Records – Capture every observation, event, or input as a timestamped entry.
- Tagged Sources – Tag each record with its source (e.g.,
#meeting,#report,#self-reflection) for traceability. - Immutable Snapshots – At key milestones, freeze a complete view of the current state. Snapshots are never modified; they serve as an unalterable history.
- Concept Library – Over time, patterns emerge from the snapshots and raw records. The system helps you extract, label, and link recurring concepts to build a growing knowledge graph of your domain.
2. Getting Started
To initialize a new decision project in a folder, run:
# Initialize a new domain (e.g., health tracking)
mkdir my-health-project
cd my-health-project
# This creates the four-layer directory structure
/path/to/dbs-decision init --domain health
The following directory tree will be created:
my-health-project/
├── raw/
│ └── README.md
├── sources/
│ └── README.md
├── snapshots/
│ └── README.md
└── concepts/
└── README.md
Each README.md explains how to use that layer.
3. Recording an Observation
Use the /case-filing trigger or directly create a markdown file in the raw folder:
---
date: 2025-03-25T09:00:00Z
tags: [#meeting, #client]
---
# Client Meeting Notes
- Attended by Alice and Bob.
- Decided to pivot to a subscription model.
- Risk: customer churn may increase in the short term.
4. Creating a Snapshot
When you've accumulated enough data (e.g., weekly, monthly), run:
# Create a snapshot with a descriptive name
/path/to/dbs-decision snapshot --name "2025-Q1 Review"
This copies all current raw entries and source tags into an immutable file under snapshots/2025-Q1-Review.md.
5. Building the Concept Library
After several snapshots, generate a concept map:
# Analyze snapshots and extract recurring patterns
/path/to/dbs-decision analyze --snapshots snapshots/*.md --output concepts/recurring-patterns.md
The concept library automatically detects repeated tags, timestamps, and keywords across snapshots, helping you identify trends and regularities.
Why This System?
- Discipline without rigidity. The four layers give structure, but you decide what and how much to record.
- Immutable history protects truth. Snapshots cannot be edited, so your past decisions remain trustworthy.
- Pattern learning is automatic. The concept library surfaces insights you might otherwise miss.
- Local-first. All data stays in your filesystem as markdown. No cloud provider required.
- AI-agent friendly. Every trigger command works with Claude Code, Cursor, Codex, or any AI agent that can read/write local files.
Example Use Cases
- Career tracking: Log meetings, project outcomes, feedback. Take monthly snapshots. Discover what conditions lead to promotion or burnout.
- Health monitoring: Record symptoms, workouts, meals. Snapshot weekly. Identify triggers for migraines or energy dips.
- Investment journal: Note decisions, market context, emotional state. Snapshot quarterly. Learn which patterns correlate with good calls.
- Relationship reflection: Capture significant conversations, conflicts, and joys. Snapshot annually. Recognize growth areas and recurring themes.
Requirements
- A local filesystem with write permissions.
- An AI agent (Claude Code, Cursor, Codex, etc.) that can execute file operations.
- Optional: Git for versioning snapshots (recommended).
Configuration
Set the default domain in a .dbs config file at the root of your project:
# .dbs/dbs-decision.yaml
domain: health
timezone: America/New_York
auto_tag: true
snapshot_frequency: weekly
License
MIT