daemon-blockint-tech

Guides experimentation engineering for A/B and multivariate tests—hypothesis framing, primary/secondary/guardrail metrics, design (A/B, A/B/n, MVT), sample size and power, duration and seasonality, allocation and randomization units, SRM checks, instrumentation, analysis plans (frequentist and Bayesian workflow level), stopping rules, readouts, rollout/kill decisions, and experiment registry hygiene (sound statistical practice only). Use for "A/B test", "A/B testing engineer", "experiment design", "sample size calculator", "statistical significance", "power analysis", "split test", "experiment readout", "guardrail metrics", "SRM", "randomization unit", "experiment duration", "multivariate test", "holdout", or "experiment analysis". Not for analytics stack build (analytics-data-engineer), marketing copy (cmo-advisor), roadmap-only (cpo-advisor), ML model experiments (ml-research-engineer-safeguards), or privacy/legal review (compliance-engineer).

Guides hands-on actuarial analyst work for insurance, reinsurance, and pension—reserving and loss development (IBNR, triangles, chain-ladder diagnostics), pricing and rate indication support (experience, trend, credibility, basic GLM at spec level), data validation and model I/O review, reporting packs and workpapers, assumption application under actuary direction, and statutory tie-outs at analyst depth. Use when the user mentions actuarial analyst, loss development, IBNR, reserve analysis, rate indication, pricing support, actuarial workpaper, triangle analysis, credibility, experience study, actuarial reporting, or reserve roll-forward—not actuary sign-off (actuary), consulting engagements (actuarial-consulting), assumption governance (assumption-setting), ALM strategy (asset-liability-management), P&C legal depth (property-casualty-insurance), charts only (data-visualization), or ETL-only pipelines (data-scrubbing).

Guides actuarial consulting engagements—client scoping and SOW design, stakeholder communication (CFO, risk, boards, regulators at overview level), due diligence and M&A actuarial support, reserving/pricing/capital review programs, model validation and opinion support, regulatory interaction prep, and deliverable governance (memos, exhibits, management presentations). Use when the user mentions actuarial consulting, actuarial engagement, reserve opinion, due diligence actuarial, model validation engagement, actuarial memo, SOW actuarial, regulatory actuarial, M&A reserves, or actuarial review—not deep technical modeling execution (actuary), P&C line education only (property-casualty-insurance), legal advice (commercial-counsel), or generic management consulting without actuarial lens (business-consultant).

Guides actuarial work for insurance and reinsurance—pricing and rate adequacy, reserving and IBNR, loss development and triangles, mortality/morbidity and lapse assumptions, experience studies and credibility, capital and risk metrics at overview level, product design tradeoffs (life, health, P&C, annuity), and regulatory reporting concepts (NAIC, IFRS 17, Solvency II overview—not legal advice). Use when the user mentions actuary, actuarial, IBNR, loss development, reserve analysis, mortality table, pricing insurance, experience study, IFRS 17, loss ratio, combined ratio, credibility, or asks for assumption documentation and model governance for insurance products—not generic FP&A (financial-analyst), investment banking valuation (comps-analysis, dcf-model), legal policy interpretation (commercial-counsel), clinical trials, software-only implementation (senior-software-engineer), or broad GRC without actuarial models (compliance-engineer).

Guides advanced long-term actuarial mathematics (SOA ALTAM)—survival models, life insurance and annuity APVs, premiums and reserves (equivalence principle, Thiele), multiple decrement and Markov states, yield-curve discounting, mortality improvement, longevity risk, profit testing, and mortality graduation. Tool-agnostic, concept-first. Use when the user mentions advanced long-term actuarial mathematics, ALTAM, survival model, life insurance reserve, annuity valuation, equivalence principle, Thiele equation, multiple decrement, force of mortality, longevity risk, mortality improvement, actuarial present value, or net premium reserve—not ASTAM/P&C (advanced-short-term-actuarial-mathematics), workpapers only (actuarial-analyst), appointed actuary (appointed-chief-actuary), assumption governance (assumption-setting), ALM detail (asset-liability-management), or exam-only deliverables.

Guides advanced short-term actuarial mathematics aligned with SOA ASTAM and P&C/health-adjacent modeling—severity and frequency distributions, aggregate and compound loss models, Bühlmann and Bühlmann-Straub credibility, ratemaking and experience rating, short-term reserving at the math level, MLE and goodness-of-fit, and risk measures (VaR, TVaR). Tool-agnostic and concept-first. Use when the user mentions advanced short-term actuarial mathematics, ASTAM, severity model, frequency model, aggregate loss, compound distribution, Bühlmann credibility, experience rating, ratemaking, pure premium, negative binomial frequency, tail factor, TVaR, or short-term actuarial models—not life contingencies (life-health-insurance), Excel workpapers only (actuarial-analyst), appointed actuary sign-off (actuary, appointed-chief-actuary), assumption governance (assumption-setting), P&C legal/operations depth (property-casualty-insurance), or general ML (data-scientist, quantitative-researcher).

This skill should be used when the user asks about AML/CFT, counter-terrorist financing, CFT, terrorist financing, proliferation financing, PF, targeted financial sanctions, asset freeze, TF typology, charitable NPO due diligence, MVTS, FATF Recommendation 6, or STR terrorist financing. Guides CFT/PF risk—TF typologies (self-funding, charitable fronts, MVTS, trade-based TF), PF and dual-use red flags, TFS and asset-freeze workflows (conceptual), CFT STR/SAR narrative structure, NPO due diligence, correspondent and cross-border TF risks, FATF R6–R8 alignment, sanctions screening integration, CFT training and independent review, exam readiness—not full AML program (KYC, CDD, TM, MLRO) → aml-compliance; sanctions vendor config only; law enforcement tactics; legal determinations. Peers: aml-compliance, compliance-engineer, auditor, commercial-counsel, information-security-engineer.

This skill should be used when the user asks about AML compliance, anti-money laundering, KYC, CDD, EDD, PEP screening, sanctions screening, transaction monitoring, SAR, STR, suspicious activity, FATF, BSA, financial crime compliance, travel rule, AML risk assessment, or MLRO. Guides risk-based AML/CFT programs, alert triage, SAR narrative structure (not legal filing), governance, crypto AML, and exam readiness—not ISO/SOC only (compliance-engineer), cloud attestations (cloud-compliance-specialist), contracts (commercial-counsel), IT audit workpapers (auditor), FinOps (finops-analyst), SIEM build (information-security-engineer), or blockchain LE investigations (blockint skills).

Mythos-class source-code reasoning for DevSec — pre-merge, pre-deploy, pre-release. Use when the user says "review this PR for security", "audit this diff", "block bad merges", "secure code review", "DevSec gate", "shift-left security", "pre-deploy security scan", "release security gate", or asks Raven to act as a Mythos-grade auditor inside the CI/CD pipeline. Every finding terminates at a tool oracle (𝒯), classical-ML detector (𝓜), or scored hypothesis (𝓛) with a falsification test — never raw LLM speculation — and ships with the D3FEND Harden technique id it implements. Defender-only: emits findings + remediation candidates, never exploit payloads.

Orchestrate 100+ specialized AI agents across an ensemble of frontier and distilled models to discover, debate, validate, prove, enrich, and triage exploitable bugs end-to-end. Use when the user wants to run the full MDASH pipeline with multi-model ensemble, dynamic agent spawning, or production-scale vulnerability discovery. Every finding terminates at a tool oracle (𝒯), classical-ML detector (𝓜), or scored hypothesis (𝓛), and ships with D3FEND defensive techniques, ATT&CK offensive context, exposure scoring, and DevSecOps routing.

Scaffold a research-grade experiment that tests whether a Looped / Recurrent-Depth Transformer (RDT) is better than a parameter-matched vanilla transformer at vulnerability discovery, when both are continued-pretrained on a security corpus. Use when the user says "test the looped-transformer hypothesis", "run the OpenMythos RDT experiment", "compare loop count to trial budget", "does inference-time loops help security reasoning", or asks for a reproducible research pipeline around the Mythos architecture hypothesis. The skill scaffolds the experiment — it does NOT make claims about Mythos itself, and it explicitly bins OpenMythos as a community speculative reconstruction, not Anthropic's actual model.

Run Project Raven's MDASH (Multi-Model Agentic Security Harness) pipeline against a target codebase or runtime. Use when the user says "hunt 0-days", "look for zero-days", "ARES scan", "variant analysis", "memory corruption hunt", "kernel/driver hunt", "smart-contract 0-day", or asks Raven to anticipate kill chains, generate exploit hypotheses, or validate LLM-claimed vulnerabilities. Every finding ends in a tool-oracle (𝒯), classical-ML detector (𝓜), or scored hypothesis (𝓛) — never raw LLM speculation — and ships with D3FEND defensive techniques, ATT&CK offensive context, and real-world exposure scoring.

Automatically enforce defensive policy against an in-progress 0-day signal, end-to-end, with strict approval gating. Use when the user says "auto-prevent", "auto-block", "auto-quarantine", "policy-driven response", "stop this 0-day now without me clicking through", or asks Raven to execute the Defend plan automatically when criteria are met. Every automatic action terminates at a tool oracle (𝒯) with mandatory approval-gate verification, references a real D3FEND Isolate / Evict / Harden technique id, and emits a tamper-evident audit trail.

Stand up Raven's defensive posture against an in-the-wild 0-day campaign. Use when the user says "defend against `<CVE>` / `<campaign>`", "harden us against this 0-day", "deploy mitigations for the new exploit", "isolate exposure", or asks to compose D3FEND Harden + Isolate + Deceive techniques into an active defensive plan. Every action terminates at a tool oracle (𝒯) or approval-gated mitigation module, never raw LLM speculation, and every recommendation is bound to a real D3FEND technique id from the 271-entry MITRE catalog plus ATT&CK offensive context and exposure scoring. Built for production DevSecOps at enterprise scale.

Run Raven's online 0-day detection stack against live telemetry, files, binaries, or network flows. Use when the user says "detect 0-days", "scan for novel threats", "run anomaly detection", "screen this artifact for unknown malware", "novelty detection", or asks for a continuous monitor against unknown-bad. Every detection terminates at a classical-ML detector (𝓜) or a tool-oracle rule (𝒯) — LLM verdicts alone are never accepted — and every alert ships with a D3FEND Detect-tactic technique id, ATT&CK offensive context, and exposure scoring. Built for production DevSecOps at enterprise scale on private codebases.

Generate, verify, and apply patches for a confirmed 0-day finding. Use when the user says "fix this 0-day", "patch this CVE", "generate the fix", "virtual-patch this", "rollback unsafe change", "apply remediation", or hands over a confirmed root cause from `raven-zero-day-investigator`. Every patch terminates at a tool oracle (𝒯) — static verifier, test suite, formal check, or restore primitive — never raw LLM speculation, and every fix is tagged with a D3FEND Harden / Restore technique id.

Compress natural language memory files (CLAUDE.md, todos, preferences) into caveman format to save input tokens. Preserves all technical substance, code, URLs, and structure. Compressed version overwrites the original file. Human-readable backup saved as FILE.original.md. Trigger: /caveman:compress FILEPATH or "compress memory file"

Compress natural language memory files (CLAUDE.md, todos, preferences) into caveman format to save input tokens. Preserves all technical substance, code, URLs, and structure. Compressed version overwrites the original file. Human-readable backup saved as FILE.original.md. Trigger: /caveman:compress FILEPATH or "compress memory file"