SCUM Resource Skill

System Consumption Utilization Monitor - Libkind-style compositional resource management.

SCUM Score Formula

SCUM(p) = α·MEM(p) + β·CPU(p) + γ·TIME(p) + δ·STALE(p)

Where:

• MEM(p): Memory usage normalized to system total
• CPU(p): CPU% averaged over sampling window
• TIME(p): Total CPU time consumed (cumulative sin)
• STALE(p): Time since last meaningful I/O (zombie indicator)

Default weights: α=0.4, β=0.2, γ=0.2, δ=0.2

GF(3) Classification

Quick Commands

# Calculate SCUM scores for top processes
scum-score

# Kill processes above threshold
scum-kill 80

# Show resource allocation as ACSet
scum-acset

# Libkind-style resource rebalancing
scum-balance

Babashka Implementation

#!/usr/bin/env bb
(require '[babashka.process :refer [shell]])

(defn parse-top []
  (->> (shell {:out :string} "top" "-l" "1" "-stats" "pid,command,cpu,mem,time" "-o" "mem" "-n" "30")
       :out
       str/split-lines
       (drop 12)
       (map #(str/split % #"\s+"))
       (filter #(> (count %) 4))))

(defn calc-scum [{:keys [mem cpu time]}]
  (let [mem-score (* 0.4 (/ mem 100))
        cpu-score (* 0.2 (/ cpu 100))
        time-score (* 0.2 (min 1.0 (/ time 3600)))
        stale-score 0.0]  ; TODO: track I/O
    (int (* 100 (+ mem-score cpu-score time-score stale-score)))))

(defn scum-report []
  (println "PID\tSCUM\tMEM\tCPU\tCOMMAND")
  (println "---\t----\t---\t---\t-------")
  (doseq [[pid cmd cpu mem time] (parse-top)]
    (when-let [scum (calc-scum {:mem (parse-double mem)
                                 :cpu (parse-double cpu)
                                 :time 0})]
      (printf "%s\t%d\t%s\t%s\t%s%n" pid scum mem cpu cmd))))

Libkind Resource Algebra

Sophie Libkind's compositional approach: resources form a resource theory (symmetric monoidal category where morphisms are resource transformations).

ACSet Schema for Processes

@present SchProcess(FreeSchema) begin
  Proc::Ob          # Processes
  Resource::Ob      # Resources (MEM, CPU, FD, NET)
  
  uses::Hom(Proc, Resource)      # Process uses resource
  amount::Attr(uses, Float64)    # How much
  
  parent::Hom(Proc, Proc)        # Process tree
  scum::Attr(Proc, Int)          # SCUM score
end

Resource Rebalancing via Colimits

# Identify processes that can share resources
# Pushout along common resource usage
function rebalance(procs::ACSet)
  # Find processes using same resource type
  shared = @acset_colim procs begin
    p1::Proc; p2::Proc; r::Resource
    uses(p1) == r
    uses(p2) == r
  end
  # Compute fair allocation as coequalizer
  coequalizer(shared)
end

Kill Interface

# Interactive: shows SCUM scores, asks before kill
scum-kill --interactive

# Automatic: kills all above threshold
scum-kill 85 --auto

# Dry run: shows what would be killed
scum-kill 70 --dry-run

# Kill by name pattern
scum-kill --pattern "java|python" --threshold 60

Current Top SCUM Offenders

Based on live system data:

Integration with Gay.jl

using Gay

# Color processes by SCUM score
function color_process(scum_score, seed=1069)
  # Map SCUM to hue: 0→120° (green), 100→0° (red)
  hue = 120 * (1 - scum_score/100)
  Gay.color_at_hue(seed, hue)
end

Voice Narration

When killing SCUM, announce with say-narration skill:

say -v Samantha "Terminating java process. SCUM score 87. Memory reclaimed: 5.2 gigabytes."

Skill Name: scum-resource
Type: System Monitoring
Trit: -1 (MINUS - validator/constrainer)
Dependencies: babashka, world-a (ACSets)

Scientific Skill Interleaving

This skill connects to the K-Dense-AI/claude-scientific-skills ecosystem:

Graph Theory

• networkx [○] via bicomodule
  • Universal graph hub

Bibliography References

• general: 734 citations in bib.duckdb

Cat# Integration

This skill maps to Cat# = Comod(P) as a bicomodule in the equipment structure:

Trit: 0 (ERGODIC)
Home: Prof
Poly Op: ⊗
Kan Role: Adj
Color: #26D826

GF(3) Naturality

The skill participates in triads satisfying:

(-1) + (0) + (+1) ≡ 0 (mod 3)

This ensures compositional coherence in the Cat# equipment structure.