Module: PWN::AI::Agent::Metrics

Defined in:
lib/pwn/ai/agent/metrics.rb

Overview

PWN::AI::Agent::Metrics is the telemetry layer of the pwn-ai learning loop. Every tool dispatch performed by PWN::AI::Agent::Loop is recorded here (name, success, duration, last error) and persisted to ~/.pwn/metrics.json.

PromptBuilder re-injects a compact effectiveness summary into the system prompt on every turn, so the model gains awareness of which tools historically succeed vs. fail on THIS host and can adapt its tool selection accordingly. This is one half of the closed feedback loop that lets pwn-ai continuously make itself smarter (the other half is PWN::AI::Agent::Learning).

PER-ENGINE SEGMENTATION

A local Ollama model and a frontier model do NOT have the same per-tool success rate — blending them mis-advises the local model about itself. Every record now also increments an :engines sub-bucket; summary/to_context accept engine: to surface only that engine's telemetry so the TOOL EFFECTIVENESS block becomes a genuine per-engine learned policy.

Constant Summary collapse

METRICS_FILE = 
File.join(Dir.home, '.pwn', 'metrics.json')
HALF_LIFE_DAYS = 
14.0
CUSUM_K = 
0.15
CUSUM_H = 
0.6
WINDOW = 
30
PRM_MIN_N =

P18/P2 — rolling mean step_reward advantage for a tool. Sample-efficiency gate: < PRM_MIN_N samples → 0 (no rank noise). Shrinkage: adv *= min(1, n/PRM_FULL_N) so sparse signal cannot dominate UCB. Zero-variance windows (all +1 or all -1 from a single session) damp to 0.5×.

5
PRM_FULL_N = 
20

Class Method Summary collapse

Class Method Details

.advantage(opts = {}) ⇒ Object

Supported Method Parameters

a = PWN::AI::Agent::Metrics.advantage(name: 'shell')

C1 — tool.success_rate − global_rate over the rolling window.



310
311
312
313
314
315
316
317
318
319
320
321
322
323
 
# File 'lib/pwn/ai/agent/metrics.rb', line 310

public_class_method def self.advantage(opts = {})
  data = load[:tools] || {}
  t    = data[opts[:name].to_s.to_sym]
  return 0.0 unless t

  # P20 — local/global from effective_rate so bandit tracks judge
  # when the handler-ok proxy is hacked (proxy_distrust high).
  local = effective_rate(name: opts[:name])
  rates = data.keys.map { |k| effective_rate(name: k) }
  global = rates.empty? ? 0.5 : (rates.sum / rates.length)
  (local - global).round(3)
rescue StandardError
  0.0
end

.append_jsonl(opts = {}) ⇒ Object 



106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
 
# File 'lib/pwn/ai/agent/metrics.rb', line 106

public_class_method def self.append_jsonl(opts = {})
  path = File.join(Dir.home, '.pwn', 'logs', 'tool_metrics.jsonl')
  FileUtils.mkdir_p(File.dirname(path))
  row = {
    ts: Time.now.utc.iso8601,
    session: Thread.current[:pwn_session_id],
    tool: opts[:name],
    latency_ms: (opts[:duration].to_f * 1000).round,
    timeout_used: opts[:timeout],
    outcome: opts[:success] ? 'ok' : 'err',
    error_class: opts[:error].to_s.split(':').first,
    bytes_out: opts[:bytes_out].to_i
  }
  File.open(path, 'a') do |f|
    f.flock(File::LOCK_EX)
    f.puts(JSON.generate(row))
  end
  row
rescue StandardError
  nil
end

.authors ⇒ Object

Author(s)

0day Inc. [email protected]



716
717
718
 
# File 'lib/pwn/ai/agent/metrics.rb', line 716

public_class_method def self.authors
  "AUTHOR(S):\n  0day Inc. <[email protected]>\n"
end

.calibration(opts = {}) ⇒ Object

Supported Method Parameters

cal = PWN::AI::Agent::Metrics.calibration(engine: :ollama)



529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
 
# File 'lib/pwn/ai/agent/metrics.rb', line 529

public_class_method def self.calibration(opts = {})
  store = load[:calibration] || {}
  buckets = if opts.key?(:engine) && !opts[:engine].to_s.empty?
              [store[opts[:engine].to_s.to_sym]].compact
            else
              store.values
            end
  c = buckets.each_with_object({ n: 0, brier_sum: 0.0, p_sum: 0.0, a_sum: 0.0 }) do |b, acc|
    next unless b.is_a?(Hash)

    acc[:n] += b[:n].to_i
    acc[:brier_sum] += b[:brier_sum].to_f
    acc[:p_sum] += b[:p_sum].to_f
    acc[:a_sum] += b[:a_sum].to_f
  end
  return { n: 0, brier: nil } unless c[:n].positive?

  n = c[:n].to_f
  { n: c[:n], brier: (c[:brier_sum] / n).round(4), mean_predicted: (c[:p_sum] / n).round(3), mean_actual: (c[:a_sum] / n).round(3), overconfidence: ((c[:p_sum] - c[:a_sum]) / n).round(3) }
end

.calibration_green?(opts = {}) ⇒ Boolean

Supported Method Parameters

PWN::AI::Agent::Metrics.reset



553
554
555
556
557
558
559
560
561
 
# File 'lib/pwn/ai/agent/metrics.rb', line 553

public_class_method def self.calibration_green?(opts = {})
  cal = calibration(engine: opts[:engine])
  return false if cal[:n].to_i < 8
  return false if cal[:overconfidence].nil?

  cal[:overconfidence].to_f <= 0.08
rescue StandardError
  false
end

.changepoints(opts = {}) ⇒ Object

Supported Method Parameters

cps = PWN::AI::Agent::Metrics.changepoints

E1 — tools whose CUSUM tripped (success_rate regime change). The caller (Mistakes.record / Curriculum) triggers extro_snapshot + correlate on these so a Mistake caused by env drift is tagged cause: :env_drift and does NOT count toward [REPEATING].



493
494
495
496
497
498
499
500
501
502
503
504
 
# File 'lib/pwn/ai/agent/metrics.rb', line 493

public_class_method def self.changepoints(opts = {})
  within = (opts[:within_secs] || 3_600).to_i
  now = Time.now.utc
  (load[:tools] || {}).filter_map do |name, t|
    cp = t[:changepoint_at]
    next unless cp && (now - Time.parse(cp)) < within

    { name: name.to_s, at: cp, window_rate: Array(t[:window]).sum.to_f / [Array(t[:window]).length, 1].max }
  end
rescue StandardError
  []
end

.effective_rate(opts = {}) ⇒ Object

Blended success rate: when proxy_distrust > 0 and judge samples exist, mix judge_rate into the handler-ok rate. distrust=1 → pure judge (or 0.5 if no judge data). distrust=0 → pure proxy.



456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
 
# File 'lib/pwn/ai/agent/metrics.rb', line 456

public_class_method def self.effective_rate(opts = {})
  name = opts[:name].to_s
  data = load[:tools] || {}
  t = data[name.to_sym]
  return 0.5 unless t

  calls = [t[:calls].to_f, 1.0].max
  proxy = t[:ok].to_f / calls
  win = Array(t[:window])
  proxy = win.sum.to_f / win.length if win.length >= 3
  distrust = 1.0 - proxy_trust
  jrate = judge_rate(name: name)
  if distrust > 0.05 && !jrate.nil?
    # P1 — scale judge weight by judge confidence so a thin local
    # heuristic ORM cannot fully replace proxy when distrust is high.
    # effective_distrust = distrust * mean(confidence), floor 0.15 when
    # we do have judge samples so the signal still moves the needle.
    jconf = judge_confidence(name: name)
    jconf = 0.7 if jconf.nil?
    eff_d = (distrust * jconf).clamp(0.0, 1.0)
    eff_d = [eff_d, 0.15].max if jconf >= 0.3
    ((proxy * (1.0 - eff_d)) + (jrate * eff_d)).clamp(0.0, 1.0).round(3)
  else
    proxy.round(3)
  end
rescue StandardError
  0.5
end

.health_line ⇒ Object 



617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
 
# File 'lib/pwn/ai/agent/metrics.rb', line 617

public_class_method def self.health_line
  board = scoreboard
  gap = nil
  if defined?(Reward) && Reward.respond_to?(:sentinel)
    s = Reward.sentinel
    gap = s[:gap_proxy_judge] if s.is_a?(Hash)
  end
  trend = if defined?(Curriculum) && Curriculum.respond_to?(:repeating_trend)
            Curriculum.repeating_trend
          else
            {}
          end
  traj = (Reward.generator_mix[:trajectory_fraction] if defined?(Reward) && Reward.respond_to?(:generator_mix))
  parked = (Mistakes.operator_inbox(limit: 50)[:count] if defined?(Mistakes) && Mistakes.respond_to?(:operator_inbox))
  "HEALTH tool_ok=#{board[:tool_ok] || '-'} task_ok=#{board[:task_ok] || '-'} " \
    "judge_ok=#{board[:judge_ok] || '-'} pred=#{board[:mean_predicted] || '-'} " \
    "judge-proxy-gap=#{gap || '-'} repeating=#{trend[:status] || '-'} " \
    "w1-traj=#{traj || '-'} parked-needs-human=#{parked || '-'}\n"
rescue StandardError
  ''
end

.help ⇒ Object

Display Usage for this Module



722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
 
# File 'lib/pwn/ai/agent/metrics.rb', line 722

public_class_method def self.help
  puts "USAGE:
    # Run load and return its result
    #{self}.load

    # Run save and return its result
    #{self}.save(
      metrics: 'required - Hash returned by .load / mutated in place'
    )

    # Run record and return its result
    #{self}.record(
      name: 'required - tool name that was dispatched',
      success: 'required - Boolean, did the handler complete without error',
      duration: 'optional - Float seconds the dispatch took (wall time)',
      error: 'optional - String error message when success is false',
      engine: 'optional - Symbol/String AI engine that chose this tool (segments telemetry)'
    )

    # Append one JSONL telemetry row under ~/.pwn/logs/tool_metrics.jsonl.
    #{self}.append_jsonl(
      name: 'required - tool name',
      success: 'optional - Boolean outcome',
      duration: 'optional - Float seconds the dispatch took (wall time)',
      error: 'optional - error string',
      engine: 'optional - engine name',
      timeout: 'optional - timeout used',
      bytes_out: 'optional - output byte count'
    )

    # Run summary and return its result
    #{self}.summary(
      limit: 'optional - cap number of tools returned (default 25)',
      engine: 'optional - only that engine\\s sub-bucket (falls back to global when absent)'
    )

    # Run to context and return its result
    #{self}.to_context(
      limit: 'optional - cap number of tools included (default 8)',
      engine: 'optional - restrict to one engine\\s telemetry'
    )

    # P4 helper — Registry.rank calls this so β·advantage is scaled down
    #{self}.proxy_trust

    # Run ucb and return its result
    #{self}.ucb(
      name: 'optional - binary or identifier name',
      c: 'optional - c value consumed by #ucb'
    )

    # Run thompson and return its result
    #{self}.thompson(
      name: 'optional - binary or identifier name'
    )

    # Run advantage and return its result
    #{self}.advantage(
      name: 'optional - binary or identifier name'
    )

    # Run prm advantage and return its result
    #{self}.prm_advantage(
      name: 'optional - binary or identifier name'
    )

    # Run prm n and return its result
    #{self}.prm_n(
      name: 'optional - binary or identifier name'
    )

    # P18 — called by Reward.prm after session annotate so live routing
    #{self}.record_step_reward(
      name: 'required - binary or identifier name',
      reward: 'optional - reward value consumed by #record_step_reward'
    )

    # P20 — fold ORM judge (0..1) into per-tool telemetry so UCB /
    #{self}.record_judge(
      name: 'required - binary or identifier name',
      score: 'optional - score value consumed by #record_judge',
      confidence: 'optional - confidence value consumed by #record_judge',
      source: 'optional - source value consumed by #record_judge'
    )

    # P1 — mean judge confidence for a tool (nil when no samples)
    #{self}.judge_confidence(
      name: 'optional - binary or identifier name'
    )

    # Mean judge score for a tool (nil when no ORM samples yet)
    #{self}.judge_rate(
      name: 'optional - binary or identifier name'
    )

    # Blended success rate: when proxy_distrust > 0 and judge samples
    #{self}.effective_rate(
      name: 'optional - binary or identifier name'
    )

    # Run changepoints and return its result
    #{self}.changepoints(
      cause: 'optional - :env_drift and does NOT count toward [REPEATING].',
      within_secs: 'optional - within secs value consumed by #changepoints'
    )

    # W3 — plan_first emits p(success); Loop.run calls this with the
    #{self}.record_calibration(
      engine: 'optional - engine value consumed by #record_calibration (defaults to :global))',
      brier: 'optional - brier value consumed by #record_calibration',
      predicted: 'optional - predicted value consumed by #record_calibration',
      actual: 'optional - actual value consumed by #record_calibration'
    )

    # Run calibration and return its result
    #{self}.calibration(
      engine: 'required - engine value consumed by #calibration'
    )

    # Run calibration green and return its result
    #{self}.calibration_green?(
      engine: 'optional - engine value consumed by #calibration_green?'
    )

    # Run scale prediction and return its result
    #{self}.scale_prediction(
      predicted: 'optional - predicted value consumed by #scale_prediction',
      engine: 'optional - engine value consumed by #scale_prediction'
    )

    # Run scoreboard and return its result
    #{self}.scoreboard(
      engine: 'optional - engine value consumed by #scoreboard'
    )

    # Run health line and return its result
    #{self}.health_line

    # Run reset and return its result
    #{self}.reset

    # Snapshot success rates per judge source for the LEARNING block.
    #{self}.snapshot(
      day: 'optional - reserved day key for rotation'
    )

    # Record token/cost for a model call.
    #{self}.record_tokens(
      tokens: 'optional - integer token count',
      cost: 'optional - Float USD cost',
      model: 'optional - model id string'
    )

    # Return cumulative token/cost usage.
    #{self}.usage(
      session_id: 'optional - unused reserved session id'
    )

    # Return ai.routing fallback chain from pwn.yaml.
    #{self}.routing(
      n: 'optional - unused reserved index'
    )

    # Print the AUTHOR(S) string for this module.
    #{self}.authors
  "
  constants.sort
end

.judge_confidence(opts = {}) ⇒ Object

P1 — mean judge confidence for a tool (nil when no samples).



415
416
417
418
419
420
421
422
423
424
425
 
# File 'lib/pwn/ai/agent/metrics.rb', line 415

public_class_method def self.judge_confidence(opts = {})
  t = (load[:tools] || {})[opts[:name].to_s.to_sym]
  return nil unless t

  win = Array(t[:judge_conf_window])
  return nil if win.empty?

  (win.sum.to_f / win.length).round(3)
rescue StandardError
  nil
end

.judge_rate(opts = {}) ⇒ Object

Mean judge score for a tool (nil when no ORM samples yet). When per-sample sources exist, LLM ORM outweighs heuristic overlap so the proxy haircut tracks the outcome model.



430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
 
# File 'lib/pwn/ai/agent/metrics.rb', line 430

public_class_method def self.judge_rate(opts = {})
  t = (load[:tools] || {})[opts[:name].to_s.to_sym]
  return nil unless t

  win = Array(t[:judge_window])
  return nil if win.empty?

  src = Array(t[:judge_src_window])
  if src.length == win.length && defined?(Reward) && Reward.respond_to?(:judge_sample_weight)
    num = 0.0
    den = 0.0
    win.each_with_index do |score, i|
      wt = Reward.judge_sample_weight(source: src[i]).to_f
      num += score.to_f * wt
      den += wt
    end
    return (num / den).round(3) if den.positive?
  end
  (win.sum.to_f / win.length).round(3)
rescue StandardError
  nil
end

.load ⇒ Object

Supported Method Parameters

metrics = PWN::AI::Agent::Metrics.load



40
41
42
43
44
45
46
47
 
# File 'lib/pwn/ai/agent/metrics.rb', line 40

public_class_method def self.load
  FileUtils.mkdir_p(File.dirname(METRICS_FILE))
  return { tools: {}, updated_at: nil } unless File.exist?(METRICS_FILE)

  JSON.parse(File.read(METRICS_FILE), symbolize_names: true)
rescue StandardError
  { tools: {}, updated_at: nil }
end

.prm_advantage(opts = {}) ⇒ Object 



333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
 
# File 'lib/pwn/ai/agent/metrics.rb', line 333

public_class_method def self.prm_advantage(opts = {})
  data = load[:tools] || {}
  t = data[opts[:name].to_s.to_sym]
  return 0.0 unless t

  win = Array(t[:prm_window])
  n = win.length
  return 0.0 if n < PRM_MIN_N

  mean = win.sum.to_f / n
  globals = data.values.map { |v| Array(v[:prm_window]) }.select { |w| w.length >= PRM_MIN_N }
  gmean = if globals.empty?
            0.0
          else
            all = globals.flatten
            all.sum.to_f / all.length
          end
  adv = mean - gmean
  # shrinkage toward 0 until PRM_FULL_N
  shrink = [n.to_f / PRM_FULL_N, 1.0].min
  # variance damp: if all equal, halve influence
  uniq = win.uniq
  var_damp = uniq.length <= 1 ? 0.5 : 1.0
  (adv * shrink * var_damp).round(3)
rescue StandardError
  0.0
end

.prm_n(opts = {}) ⇒ Object 



361
362
363
364
365
366
367
368
 
# File 'lib/pwn/ai/agent/metrics.rb', line 361

public_class_method def self.prm_n(opts = {})
  t = (load[:tools] || {})[opts[:name].to_s.to_sym]
  return 0 unless t

  Array(t[:prm_window]).length
rescue StandardError
  0
end

.proxy_trust ⇒ Object

P4 helper — Registry.rank calls this so β·advantage is scaled down when the proxy is untrustworthy.



257
258
259
260
261
262
 
# File 'lib/pwn/ai/agent/metrics.rb', line 257

public_class_method def self.proxy_trust
  d = defined?(Reward) && Reward.respond_to?(:proxy_distrust) ? Reward.proxy_distrust : 0.0
  (1.0 - d.to_f).clamp(0.0, 1.0)
rescue StandardError
  1.0
end

.record(opts = {}) ⇒ Object

Supported Method Parameters

PWN::AI::Agent::Metrics.record( name: 'required - tool name that was dispatched', success: 'required - Boolean, did the handler complete without error', duration: 'optional - Float seconds the dispatch took', error: 'optional - String error message when success is false', engine: 'optional - Symbol/String AI engine that chose this tool (segments telemetry)' )



83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
 
# File 'lib/pwn/ai/agent/metrics.rb', line 83

public_class_method def self.record(opts = {})
  name     = opts[:name].to_s
  success  = opts[:success] ? true : false
  duration = opts[:duration].to_f
  error    = opts[:error]
  engine   = opts[:engine].to_s
  return if name.empty?

  metrics = load
  metrics[:tools] ||= {}
  key = name.to_sym
  t = metrics[:tools][key] ||= blank_bucket
  bump(bucket: t, success: success, duration: duration, error: error)
  unless engine.empty?
    t[:engines] ||= {}
    e = t[:engines][engine.to_sym] ||= blank_bucket
    bump(bucket: e, success: success, duration: duration, error: error)
  end
  save(metrics: metrics)
  append_jsonl(opts.merge(name: name, success: success, duration: duration, error: error, engine: engine))
  t
end

.record_calibration(opts = {}) ⇒ Object

Supported Method Parameters

PWN::AI::Agent::Metrics.record_calibration(predicted:, actual:, brier:, engine:)

W3 — plan_first emits p(success); Loop.run calls this with the realised outcome. Tracked per-engine so calibration of the local LoRA vs frontier is comparable.



513
514
515
516
517
518
519
520
521
522
523
524
 
# File 'lib/pwn/ai/agent/metrics.rb', line 513

public_class_method def self.record_calibration(opts = {})
  m = load
  m[:calibration] ||= {}
  eng = (opts[:engine] || :global).to_s.to_sym
  c = m[:calibration][eng] ||= { n: 0, brier_sum: 0.0, p_sum: 0.0, a_sum: 0.0 }
  c[:n]         += 1
  c[:brier_sum] += opts[:brier].to_f
  c[:p_sum]     += opts[:predicted].to_f
  c[:a_sum]     += opts[:actual].to_f
  save(metrics: m)
  c
end

.record_judge(opts = {}) ⇒ Object

P20 — fold ORM judge (0..1) into per-tool telemetry so UCB / Thompson / advantage can prefer judge-grounded rates over the inflated handler-ok proxy when proxy_distrust is high.



392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
 
# File 'lib/pwn/ai/agent/metrics.rb', line 392

public_class_method def self.record_judge(opts = {})
  name = opts[:name].to_s
  return if name.empty?

  score = opts[:score].to_f.clamp(0.0, 1.0)
  conf  = opts.key?(:confidence) ? opts[:confidence].to_f.clamp(0.0, 1.0) : 0.7
  src   = opts[:source].to_s
  src   = 'heuristic' if src.empty?
  m = load
  m[:tools] ||= {}
  t = m[:tools][name.to_sym] ||= blank_bucket
  t[:judge_window] = (Array(t[:judge_window]) + [score]).last(40)
  t[:judge_conf_window] = (Array(t[:judge_conf_window]) + [conf]).last(40)
  t[:judge_src_window] = (Array(t[:judge_src_window]) + [src]).last(40)
  t[:judge_sum] = t[:judge_window].sum.to_f
  t[:judge_n] = t[:judge_window].length
  save(metrics: m)
  t
rescue StandardError
  nil
end

.record_step_reward(opts = {}) ⇒ Object

P18 — called by Reward.prm after session annotate so live routing can bias toward tools that recently advanced goals (+1 step_reward).



372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
 
# File 'lib/pwn/ai/agent/metrics.rb', line 372

public_class_method def self.record_step_reward(opts = {})
  name = opts[:name].to_s
  return if name.empty?

  rew = opts[:reward].to_f.clamp(-1.0, 1.0)
  m = load
  m[:tools] ||= {}
  t = m[:tools][name.to_sym] ||= blank_bucket
  t[:prm_window] = (Array(t[:prm_window]) + [rew]).last(40)
  t[:prm_sum] = t[:prm_window].sum
  t[:prm_n] = t[:prm_window].length
  save(metrics: m)
  t
rescue StandardError
  nil
end

.record_tokens(opts = {}) ⇒ Object 



172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
 
# File 'lib/pwn/ai/agent/metrics.rb', line 172

public_class_method def self.record_tokens(opts = {})
  n = opts[:tokens].to_i
  cost = opts[:cost].to_f
  model = opts[:model].to_s
  m = load
  m[:usage] ||= { tokens: 0, cost: 0.0, calls: 0, by_model: {} }
  m[:usage][:tokens] += n
  m[:usage][:cost] += cost
  m[:usage][:calls] += 1
  m[:usage][:by_model][model] ||= { tokens: 0, cost: 0.0, calls: 0 }
  m[:usage][:by_model][model][:tokens] += n
  m[:usage][:by_model][model][:cost] += cost
  m[:usage][:by_model][model][:calls] += 1
  save(metrics: m)
  m[:usage]
end

.reset ⇒ Object 



639
640
641
642
 
# File 'lib/pwn/ai/agent/metrics.rb', line 639

public_class_method def self.reset
  FileUtils.rm_f(METRICS_FILE)
  { tools: {}, updated_at: nil }
end

.routing(opts = {}) ⇒ Object 



194
195
196
197
198
199
200
 
# File 'lib/pwn/ai/agent/metrics.rb', line 194

public_class_method def self.routing(opts = {})
  _n = opts[:n]
  chain = (PWN::Env.dig(:ai, :routing) if defined?(PWN::Env))
  Array(chain)
rescue StandardError
  []
end

.save(opts = {}) ⇒ Object

Supported Method Parameters

PWN::AI::Agent::Metrics.save( metrics: 'required - Hash returned by .load / mutated in place' )



54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
 
# File 'lib/pwn/ai/agent/metrics.rb', line 54

public_class_method def self.save(opts = {})
  metrics = opts[:metrics] ||= { tools: {} }
  metrics[:updated_at] = Time.now.utc.iso8601
  FileUtils.mkdir_p(File.dirname(METRICS_FILE))
  # 4.4 — flock + atomic rename
  path = METRICS_FILE
  tmp  = File.join(File.dirname(path), ".#{File.basename(path)}.#{Process.pid}.tmp")
  body = JSON.pretty_generate(metrics)
  File.open(tmp, File::WRONLY | File::CREAT | File::TRUNC, 0o644) do |f|
    f.flock(File::LOCK_EX)
    f.write(body)
    f.flush
    f.fsync
  end
  File.rename(tmp, path)
  metrics
ensure
  FileUtils.rm_f(tmp) if defined?(tmp) && tmp && File.exist?(tmp)
end

.scale_prediction(opts = {}) ⇒ Object 



563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
 
# File 'lib/pwn/ai/agent/metrics.rb', line 563

public_class_method def self.scale_prediction(opts = {})
  p = opts[:predicted].to_f.clamp(0.0, 1.0)
  cal = calibration(engine: opts[:engine])
  return p.round(3) if cal[:n].to_i < 8 || cal[:overconfidence].nil?

  oc = cal[:overconfidence].to_f
  return p.round(3) if oc <= 0.02

  temp = (1.0 + (4.0 * oc.clamp(0.0, 1.0))).clamp(1.0, 8.0)
  q = p.clamp(1.0e-6, 1.0 - 1.0e-6)
  logit = Math.log(q / (1.0 - q))
  scaled = 1.0 / (1.0 + Math.exp(-(logit / temp)))
  scaled.round(3)
rescue StandardError
  opts[:predicted].to_f.clamp(0.0, 1.0)
end

.scoreboard(opts = {}) ⇒ Object 



580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
 
# File 'lib/pwn/ai/agent/metrics.rb', line 580

public_class_method def self.scoreboard(opts = {})
  rows = summary(limit: 50)
  tool_ok = if rows.empty?
              nil
            else
              w = rows.sum { |r| r[:calls].to_f }
              w.positive? ? (rows.sum { |r| r[:success_rate].to_f * r[:calls].to_i } / w).round(3) : nil
            end
  task_ok = nil
  if defined?(Learning) && Learning.respond_to?(:outcomes)
    rec = Learning.outcomes(limit: 200).reject do |r|
      r[:success].nil? || r[:status].to_s == 'unverified' || r[:verdict].to_s == 'unknown' ||
        (r[:decision_version] && r[:training_score].nil?)
    end
    if rec.any?
      hits = rec.count { |r| r[:success] == true }
      task_ok = (hits.to_f / rec.length).round(3)
    end
  end
  cal = calibration(engine: opts[:engine])
  judge_ok = cal[:mean_actual]
  if judge_ok.nil? && defined?(Reward) && Reward.respond_to?(:sentinel)
    s = Reward.sentinel
    judge_ok = s[:judge] if s.is_a?(Hash)
  end
  {
    tool_ok: tool_ok,
    task_ok: task_ok,
    judge_ok: judge_ok,
    mean_predicted: cal[:mean_predicted],
    overconfidence: cal[:overconfidence],
    n: cal[:n].to_i
  }
rescue StandardError
  { tool_ok: nil, task_ok: nil, judge_ok: nil, n: 0 }
end

.snapshot(opts = {}) ⇒ Object 



159
160
161
162
163
164
165
166
167
168
169
170
 
# File 'lib/pwn/ai/agent/metrics.rb', line 159

public_class_method def self.snapshot(opts = {})
  _day = opts[:day]
  learn = defined?(Learning) ? Learning.stats : {}
  {
    success_rate: learn[:success_rate],
    success_rate_orm: learn[:success_rate_orm],
    success_rate_heur: learn[:success_rate_heur],
    brier: (learn.dig(:calibration, :brier) if learn.is_a?(Hash)),
    proxy_distrust: learn[:proxy_distrust],
    tools: summary(limit: 8)
  }
end

.summary(opts = {}) ⇒ Object

Supported Method Parameters

rows = PWN::AI::Agent::Metrics.summary( limit: 'optional - cap number of tools returned (default 25)', engine: 'optional - only that engine's sub-bucket (falls back to global when absent)' )



134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
 
# File 'lib/pwn/ai/agent/metrics.rb', line 134

public_class_method def self.summary(opts = {})
  limit  = opts[:limit] || 25
  engine = opts[:engine].to_s
  tools  = load[:tools] || {}
  rows = tools.map do |name, t|
    b = engine.empty? ? t : (t.dig(:engines, engine.to_sym) || t)
    calls = b[:calls].to_i
    ok    = b[:ok].to_i
    rate  = calls.positive? ? (ok.to_f / calls).round(3) : 0.0
    avg   = calls.positive? ? (b[:total_duration].to_f / calls).round(3) : 0.0
    {
      name: name.to_s,
      calls: calls,
      success_rate: rate,
      judge_rate: judge_rate(name: name),
      effective_rate: effective_rate(name: name),
      avg_duration: avg,
      last_error: b[:last_error],
      last_at: b[:last_at]
    }
  end
  rows.reject { |r| r[:calls].zero? }
      .sort_by { |r| [-r[:calls], -r[:success_rate]] }.first(limit)
end

.thompson(opts = {}) ⇒ Object

Supported Method Parameters

p = PWN::AI::Agent::Metrics.thompson(name: 'shell')

C1 — Thompson sample from Beta(ok+1, fail+1). Naturally balances exploit/explore; used by Registry.rank as the tie-breaker.



284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
 
# File 'lib/pwn/ai/agent/metrics.rb', line 284

public_class_method def self.thompson(opts = {})
  t = (load[:tools] || {})[opts[:name].to_s.to_sym] || blank_bucket
  # P20 — when judge samples exist and distrust > 0, tilt Beta toward
  # judge_rate so Thompson explore/exploit tracks ORM not handler-ok.
  ok = t[:ok].to_f
  fail = t[:fail].to_f
  jn = Array(t[:judge_window]).length
  if jn >= 3 && proxy_trust < 0.95
    jr = judge_rate(name: opts[:name]).to_f
    # pseudo-counts from judge window, mixed by distrust
    d = (1.0 - proxy_trust).clamp(0.0, 1.0)
    jok = jr * jn
    jfail = (1.0 - jr) * jn
    ok = (ok * (1.0 - d)) + (jok * d)
    fail = (fail * (1.0 - d)) + (jfail * d)
  end
  beta_sample(alpha: ok + 1.0, beta: fail + 1.0)
rescue StandardError
  0.5
end

.to_context(opts = {}) ⇒ Object

Supported Method Parameters

ctx = PWN::AI::Agent::Metrics.to_context( limit: 'optional - cap number of tools included (default 8)', engine: 'optional - restrict to one engine's telemetry' )



208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
 
# File 'lib/pwn/ai/agent/metrics.rb', line 208

public_class_method def self.to_context(opts = {})
  limit  = opts[:limit] || 8
  engine = opts[:engine]
  rows   = summary(limit: limit, engine: engine)
  return '' if rows.empty?

  # P4 — when Reward.sentinel says proxy is hacked, haircut displayed
  # success rates so the model does not trust the lie in the prompt.
  distrust = defined?(Reward) && Reward.respond_to?(:proxy_distrust) ? Reward.proxy_distrust : 0.0
  scope = engine.to_s.empty? ? 'historical' : "engine=#{engine}"
  scope = "#{scope}, proxy_distrust=#{distrust.round(2)}" if distrust.positive?
  lines = rows.map do |r|
    # P20 — display effective_rate (judge-blended) when available;
    # fall back to distrust haircut on raw proxy.
    rate = r[:success_rate].to_f
    eff  = r[:effective_rate]
    adj  = if eff
             eff.to_f
           else
             rate - ((rate - 0.5) * distrust)
           end
    err = r[:last_error] ? " last_err=#{r[:last_error][0, 60]}" : ''
    jtag = r[:judge_rate] ? " judge=#{(r[:judge_rate].to_f * 100).round(1)}%" : ''
    tag = distrust.positive? || r[:judge_rate] ? ' (adj)' : ''
    "  - #{r[:name]}: calls=#{r[:calls]} success=#{(adj * 100).round(1)}%#{tag}#{jtag} avg=#{r[:avg_duration]}s#{err}"
  end
  warn_line = distrust.positive? ? "WARNING: reward proxy diverges from judge — success rates haircut by distrust=#{distrust.round(2)}; prefer judge-scored exemplars over raw rates.\n" : ''
  # P0 ops — surface W1 generator_mix when diet is unhealthy so the
  # online controller (and the model) prefer underfilled sources.
  mix_line = ''
  if defined?(Reward) && Reward.respond_to?(:generator_mix)
    begin
      m = Reward.generator_mix
      unless m[:healthy]
        mix_line = "W1 MIX: n=#{m[:n]} traj=#{m[:trajectory_fraction]} " \
                   "urgent=#{Array(m[:urgent]).join(',')} " \
                   "suppress=#{Array(m[:suppress]).join(',')} " \
                   "rec=#{m[:recommendation]}\n"
      end
    rescue StandardError
      mix_line = ''
    end
  end
  health = health_line
  "#{warn_line}#{mix_line}#{health}TOOL EFFECTIVENESS (#{scope}, adapt tool choice accordingly)\n#{lines.join("\n")}\n\n"
end

.ucb(opts = {}) ⇒ Object 



264
265
266
267
268
269
270
271
272
273
274
275
276
 
# File 'lib/pwn/ai/agent/metrics.rb', line 264

public_class_method def self.ucb(opts = {})
  name = opts[:name].to_s
  c    = (opts[:c] || 1.4).to_f
  data = load[:tools] || {}
  t    = data[name.to_sym] || blank_bucket
  n    = [t[:calls].to_f, 1.0].max
  total = [data.values.sum { |v| v[:calls].to_f }, 1.0].max
  # P20 — mean from effective_rate (judge-blended when distrust high)
  mean  = effective_rate(name: name)
  mean + (c * Math.sqrt(Math.log(total) / n))
rescue StandardError
  1.0
end

.usage(opts = {}) ⇒ Object 



189
190
191
192
 
# File 'lib/pwn/ai/agent/metrics.rb', line 189

public_class_method def self.usage(opts = {})
  _sid = opts[:session_id]
  load[:usage] || { tokens: 0, cost: 0.0, calls: 0 }
end