Data & analysis
LYGO Ops Detector — local AETHONΔ9 discourse heuristics for evasion, half-truth certainty, saturation bait, coordination language, and policy-refusal signals in operator-supplied text, plus weighted public metadata (account_based_in mismatch, VPN-hint, HTTPS-cited incidents, same-geo+scripted batch). Opt-in only. Stdlib CLI; file reads require --i-consent; eval writes under tests/. Not for doxing or nationality guilt. Dual-threshold (operational 0.65 vs calibration). Pairs with lygo-flame-ward. Triggers: lygo ops detector, aethon d9, evasion index (explicit).
LYGO Ops Detector — local AETHONΔ9 discourse heuristics for evasion, half-truth certainty, saturation bait, coordination language, and policy-refusal signals in operator-supplied text, plus weighted public metadata (account_based_in mismatch, VPN-hint, HTTPS-cited incidents, same-geo+scripted batch). Opt-in only. Stdlib CLI; file reads require --i-consent; eval writes under tests/. Not for doxing or nationality guilt. Dual-threshold (operational 0.65 vs calibration). Pairs with lygo-flame-ward. Triggers: lygo ops detector, aethon d9, evasion index (explicit).
Local, deterministic discourse-signal heuristics.
Not a person profiler. Not sole evidence. Not for doxing.
| Capability | Default |
|---|---|
| Network | None |
| Shell / subprocess | None |
| Read local files | Only --text-file / --assoc-file + --i-consent |
| Write | Eval report under tests/ only |
| Auto-publish / social | Never |
Before analyzing email, DMs, private logs, or association lists:
--i-consent on file inputs).--text; do not paste secrets or third-party PII into shared chats.Invoke only when the user explicitly wants one of:
Do not auto-trigger on generic “analyze this email/log/thread” without ops-detector intent.
| Channel | Measures | Does not measure |
|---|---|---|
| Evasion | Burden-shift, ad hominem, vague claims, authority inflation, gaslight, deflection templates | Person identity |
| Association | Coordination/secrecy language in association strings you provide | Social graph doxing; bare job titles |
| Institutional | Policy-as-shield / refusal-to-comment language only | Affiliation, faith, fraternity, lodge, profession labels |
Unit of analysis = text under review, not a human “subject” or investigation target.
Each signal has in-scope / out-of-scope examples (--show-boundaries). Ordinary disagreement without proof-avoidance templates should stay low.
| Bar | Meaning |
|---|---|
Operational ops_score >= 0.65 or high evasion | Documented strong multi-signal bar for human review |
| Calibration (lower, e.g. 0.05) | Short-suite ranking only — not production performance |
python scripts/eval_ops_detector.py tests/labeled_discourse_suite.json --sweep
# report: operational_metrics (0.65) + calibration_metrics (ranking only)
# --out must stay under tests/
cd path/to/lygo-ops-detector
python scripts/self_check.py
python scripts/lygo_ops_detector.py --text "paste discourse here" --json
# optional local files (operator affirms authority):
python scripts/lygo_ops_detector.py --text-file ./snippet.txt --i-consent
python scripts/lygo_ops_detector.py --show-boundaries
Exit codes: 0 = no strong operational pattern; 3 = consent required; 10 = high evasion discourse signals (review claims).
--i-consent.See references/SECURITY.md and references/SKILLSPECTOR_AUDIT.md.
| Ver | Change |
|---|---|
| 1.1.0 | Dynamic eval suite |
| 1.2.1 | Permissions, narrow triggers, dual-threshold honesty |
| 1.2.2 | SkillSpector: discourse-not-subject framing; no bare job/affiliation association cues; --i-consent on files; eval writes constrained to tests/; signal boundaries |
Δ9Φ963 — receipts over hype · discourse not identity · consent before private data.
LYGO Deception Radar — public proof layer for Ops Detector. Rebuilds an anonymized radar feed + HTML dashboard from the public labeled discourse suite only. Shows strong vs weak vs clear bands with operational threshold 0.65. Not for doxing or private mail. No subprocess. Install clawhub:@deepseekoracle/lygo-deception-radar.
LYGO Flame Ward — harden the lattice against disinfo, injected half-truths, corrupted authority, and silent WebAudio device fingerprinting. Default: all sources fabricated until concordance. Seals-first. Ingest gate · flame-scan · endpoint-scan · quarantine · burn-receipt. Local-first, consent-gated. No network, no subprocess, no auto-publish.
LYGO Prompt Implant System (LPIS) v1.1 — analyze authorized prompts locally; P0 gate, P1 vault, P3 sovereign variants, P5 advisory implant. Requires --i-auth...
LYGO SkillSpector — enhanced local pre-install skill risk scanner for OpenClaw/ClawHub packages. Scan, gate (CI max-band), batch, and Markdown report: subprocess/shell, network/HTTP/httpx, secrets, eval/exec, curl|bash, miners, claim mismatches. Use when auditing ClawHub skills, SkillSpector-style gates, safe install checks, malware triage. Pure stdlib. No network, no subprocess, no auto-install. Builder HTML/multi-root pack on SkillHub FULL if running a full stack. Install clawhub:@deepseekoracle/lygo-skill-spector.
LYGO Skill Gate — local pre-install skill risk scanner for OpenClaw/ClawHub packages. Scan any skill folder before you install or trust it: subprocess/shell, network/HTTP, secrets in source, eval/exec, webhook/exfil hints, permission-claim mismatches. Use when auditing ClawHub skills, reviewing SKILL.md safety, safe install checks, malware triage, or SkillSpector-style local gates. Pure stdlib. No network, no subprocess, no auto-install. Install clawhub:@deepseekoracle/lygo-skill-gate.
LYGO CLI Bridge — unified entrypoint: lygo health | map | analyze | mint | radar | next. Wraps kickstart + ops-detector + mint-walkthrough + deception-radar without exposing internal skill layout. In-process imports only; no subprocess; no auto-publish. Install clawhub:@deepseekoracle/lygo-cli-bridge.