Results for “relative-strength”

Skeptical pushback on visual placement — both density and quality. Reads the annotated outline plus the visuals manifest and asks (a) whether the article hits the density target from editorial-principles-visuals.md, (b) whether each [VISUAL:...] earns its place, (c) whether sections without one would benefit. One revision pass on FAIL (BLOG_AGENT_VISUALS_REVISION_BUDGET, default 1).

Strengthen interfaces against edge cases, errors, internationalization issues, and real-world usage scenarios that break idealized designs.

Evaluates the robustness and alignment with human judgment of reference-based and reference-free evaluation metrics for machine translation and summarization, particularly under adversarial conditions.

Stress-test a feature idea for an existing product by surfacing risky assumptions across Value, Usability, Viability, and Feasibility using multi-perspective devil's advocate thinking.

Multi-phase dialectical stress-test for HIGH-STAKES decisions only (architecture choices, irreversible product calls, strategic bets). Heavy-cost skill — do NOT use for routine questions, brainstorming, or simple tradeoffs. User must explicitly invoke or describe a decision they call "high-stakes", "irreversible", or "needs stress-testing". Triggers: "stress test this decision", "dialectic on", "challenge this thesis", "should I really".

High-performance vector similarity search engine for RAG and semantic search. Use when building production RAG systems requiring fast nearest neighbor search, hybrid search with filtering, or scalable vector storage with Rust-powered performance.

Expert guidance for GRPO/RL fine-tuning with TRL for reasoning and task-specific model training

Compress an oversized SKILL.md to under 200 lines without losing effectiveness. Load when a skill exceeds 200 lines, when AGENTS.md triggers compression after a skill edit, or when the user asks to compress, shrink, slim down, or optimize a skill. Also triggers on "this skill is too long", "reduce skill size", "make this skill shorter". Applies to all skills including meta skills — the 200-line rule has no exceptions. Preserves hard gates, gotchas, output format, routing triggers, and at least one example. When genuinely CORE content cannot be compressed away, invokes split-skill instead of degrading the skill.

Evaluate industry dynamics and market attractiveness by analyzing competitive rivalry, supplier power, buyer power, threat of substitutes, and threat of new entrants.

Grit: General Robust Image Task Benchmark

Use when measuring volatility with ATR, setting ATR-based stop-losses, calculating position sizes by risk tolerance, or comparing volatility across timeframes.

Agent validation v4.3.0 — Make agents act effectively by disabling harmful actions, lowering gates, and injecting cross-run learning

Industry and market research. Market sizing, structure, competition, regulation, supply chains, pricing, M&A. Produces a single compact, source-cited research report with Porter's Five Forces, HHI / CR4, PESTLE, and trade flows. Supports UK, US, EU, Australia, and global scope, with multi-geography comparison.

Tests approaches at extreme scales to reveal fundamental limits and strengths hidden at normal scales.

Linkable-Asset Skill — data studies that pull links (Lesson 7)

Fix PPO training early-stop issues. Trigger when: (1) impossible drawdown values (>100%), (2) training stops too early, (3) need adaptive recovery instead of hard stop.

High-performance vector similarity search engine for RAG and semantic search. Use when building production RAG systems requiring fast nearest neighbor search, hybrid search with filtering, or scalable vector storage with Rust-powered performance.

Train LLMs with reinforcement learning using verl (Volcano Engine RL), supporting RLHF, GRPO, PPO, and other algorithms for scalable post-training with flexible infrastructure backends.

Fine-tune and align language models using reinforcement learning with TRL, including SFT, DPO, PPO, GRPO, and reward model training.

Run replication, robustness, and specification-sensitivity workflows for Stata projects. Use when a researcher wants to reproduce a result, rerun a pipeline, compare specifications, audit a do-file sequence, or check whether a claim is stable.

Enforces an algorithm-first discipline: state Big-O, data structure, and algorithm family before writing loops, queries, or recursion, catching O(n^2), N+1, and brute-force defaults.

Algorithm-first discipline: state Big-O, data structure, and algorithm family BEFORE writing loops, queries, or recursion. Catches O(n^2), N+1, and brute-force defaults.

Validate skill outputs against assertion manifests to detect regressions, enforce quality gates, and bootstrap starter manifests from historical runs.

Simulate an insider threat or post-breach attacker to identify lateral movement paths, privilege escalation vectors, and sensitive data exposure within a corporate network.

当观察到或经历某种极端转化现象，或试图主动促成根本性观念转变时

Algorithm-first discipline: state Big-O, data structure, and algorithm family BEFORE writing loops, queries, or recursion. Catches O(n^2), N+1, and brute-force defaults.

v4.1.0 reward function redesign to fix overtrading and DSR dominance

proptest — property-based testing for Rust (Hypothesis-style). Generates thousands of random inputs to find counterexamples to invariants, with shrinking to minimal failing case. Critical for crypto, parsers, state machines, and finance code where edge cases matter. Includes strategies (gen functions), regression tracking, and integration with cargo test. Alternative: quickcheck (simpler API, less powerful). USE WHEN: user mentions "proptest", "property-based test", "proptest!", "Strategy", "Arbitrary", "shrink", "quickcheck", "regression file", "fuzz-light", "proptest-derive" DO NOT USE FOR: Generic Rust unit tests - use `testing/rust-testing` DO NOT USE FOR: Fuzzing with mutator (cargo-fuzz, AFL) - use bitcoin/testing/fuzz DO NOT USE FOR: KMP property testing - use `testing/kotest` (has built-in property)

Enforces an algorithm-first discipline: state Big-O, data structure, and algorithm family before writing loops, queries, or recursion to catch O(n²), N+1, and brute-force defaults.

ROACH v1.0 — Striker-only scanner. Stalker mode disabled entirely. Tests whether Stalker adds any value or is pure drag. Only trades violent FIRST_JUMP explosions backed by 1.5x volume. Will have long stretches of silence — that patience is the edge. All hardened gates preserved. XYZ banned. Leverage 7-10x. DSL exit managed by plugin runtime via runtime.yaml.

Train reinforcement learning models using torchforge, Meta's PyTorch-native RL library for scalable, algorithm-focused experimentation with GRPO, DAPO, and custom loss functions.

High-performance vector similarity search engine for RAG and semantic search. Use when building production RAG systems requiring fast nearest neighbor search, hybrid search with filtering, or scalable vector storage with Rust-powered performance.

Reverse-engineers custom JavaScript-based WASM virtual machines and risk-control engines by identifying DSL VM patterns, extracting opcodes, analyzing constant tables, and tracing exported functions through static analysis and runtime injection.

Algorithm-first discipline: state Big-O, data structure, and algorithm family BEFORE writing loops, queries, or recursion. Catches O(n^2), N+1, and brute-force defaults.

Use when implementing risk-adjusted rewards, discussing Sharpe ratio in RL training, or tuning reward components for risk awareness