Decision Tracking
Track development decisions, record outcomes, and learn from patterns
Decision Tracking helps you log important development decisions, record their outcomes, and learn from patterns over time.
Overview#
Track decisions to:
- Remember context - Why was this decision made?
- Record outcomes - Did it work? What happened?
- Learn patterns - What decisions lead to success?
- Get recommendations - What worked before in similar situations?
How It Works#
Decision Logged Outcome Recorded Pattern Learned
───────────────────────────────────────────────────────────
"Use PostgreSQL" ──▶ "Success" ──▶ "PostgreSQL works
for user data" "Fast queries" well for user data"
(confidence: 0.85)
Logging Decisions#
Via CLI#
Loading code block...
Decision Types#
| Type | Description | Examples |
|---|---|---|
architecture | System design decisions | Tech stack, patterns |
database | Data model decisions | Schema, queries |
security | Security-related | Auth, encryption |
performance | Optimization choices | Caching, indexes |
ui | Interface decisions | Components, layout |
dependency | Library choices | npm packages |
process | Workflow decisions | Git flow, deployments |
Recording Outcomes#
When a Decision Works#
Loading code block...
When a Decision Fails#
Loading code block...
Outcome Statuses#
success- Decision worked as expectedpartial- Partially successful, some issuesfailure- Decision didn't work outpending- Still evaluating
Getting Recommendations#
Ask for past decisions that worked:
Loading code block...
Impact Scoring#
Decisions are automatically scored by impact:
| Level | Score | Description |
|---|---|---|
| Critical | 5 | Major architectural decisions |
| High | 4 | Important feature decisions |
| Medium | 3 | Standard development choices |
| Low | 2 | Minor implementation details |
| Minimal | 1 | Trivial decisions |
How Impact is Calculated#
Impact is calculated as: typeWeight x confidenceScore x outcomeMultiplier
- typeWeight - Based on decision type (architecture = 1.5, security = 1.4, etc.)
- confidenceScore - Your confidence when making the decision
- outcomeMultiplier - Success = 1.2, Failure = 0.8
Viewing Decisions#
Recent Decisions#
Loading code block...
Pending Outcomes#
Loading code block...
Decision Statistics#
Loading code block...
Pattern Learning#
The system automatically learns from your decisions over time. It identifies success patterns (e.g., "PostgreSQL decisions succeed 90% of the time") and anti-patterns to avoid (e.g., "hardcoded secrets have a 100% failure rate").
Best Practices#
- Log decisions as you make them - Context is freshest immediately
- Be specific about reasons - Future you will thank present you
- Record outcomes promptly - Don't wait too long to evaluate
- Review patterns regularly - Learn from your history
- Include metrics when possible - Quantifiable outcomes are more useful
Performance#
Decision tracking uses an append-only strategy for fast writes:
- O(1) write performance for logging decisions
- O(1) write performance for recording outcomes
- Periodic compaction keeps read performance optimal