causal-reasoner

name	causal-reasoner
version	0.2.0
description	Causal DAG engine — infers root cause from observed effects using Bayesian-weighted evidence chains
enabled	true
category	intelligence
tags	["causal","reasoning","root-cause","bayesian","diagnostics"]

🔮 causal-reasoner

Causal inference engine. Takes observed effects (CPU spike, memory pressure, deploy events) and backtracks through a directed acyclic causal graph with weighted evidence scoring.

Architecture

[git_push] ──→ [deploy] ──→ [cpu_load]
                              ├──→ [memory_usage]
[config_change] ─────────────┘
[high_traffic] ──────────────┘

Usage

python causal-reasoner/run.py --infer cpu_load              # infer root cause
python causal-reasoner/run.py --infer memory_usage --json    # JSON output
python causal-reasoner/run.py --evidence-path daemon/state.json  # read daemon state
python causal-reasoner/run.py --list-graph                   # dump causal DAG
python causal-reasoner/run.py --version
python causal-reasoner/run.py --dry-run --infer cpu_load

Integration

• Reads daemon state from skills/.daemon/state.json
• Writes reasoning logs to skills/causal-reasoner/logs/reasoning.jsonl
• Scoring: Bayesian-weighted evidence chain with confidence decay (0.85^n)
• Used by subconscious-daemon to suppress expected alerts (e.g. deployment → CPU spike)