Claude Code Skills · page 91

loss-1 judge - read a sample's full dialogue and decide whether the user simulator semantically enacted its Policy Card. check-blind.

loss-2 judge - pairwise quality comparison across the n rungs within one topic; decide monotonicity and endpoint separation. check-blind, D1-D5 only.

Strategy: Inference to the best explanation in the face of anomalies

Remove components one by one, observe system changes to reveal hidden

Map system architecture to ablatable units for ablation studies

Design ablation studies to isolate component contributions in ML systems

Remove components one by one from a system, record the response/impact

Classify assumptions on 2 axes — load-bearing (how much conclusion depends

Extract abstract principles from concrete domain cases. Strips domain-specific

Perform bisociation at multiple abstraction levels

Move between concrete and abstract framings — 3 levels up (Why?) and

Abstract biological principle to design principle. Bridge from biology

Compute Risk Priority Number (RPN = S x O x D), classify failure modes

Enumerate all implementation activities from an experiment design

Understand who the user is — background, resources, constraints, and

Iteratively select maximally informative pairs, execute comparisons,

Structured debate protocol that constructs an advocate, deploys critic

Strategy: Progressive pressure escalation — starts with surface-level

Strategy: Role-play attacks from hostile personas — competing lab researcher,

Tactic: Construct detailed hostile persona, attack artifact from that

Campaign: Logical extreme and boundary testing via reductio ad absurdum

Construct the strongest possible case for a rejected candidate or counter-position.

Aggregate multiple ranking ballots into a consensus ranking using a specified

SOP: Use the AHP (Analytic Hierarchy Process) to determine scoring-dimension weights, outputting a weight vector

SOP: Use the AHRQ PiCMe framework to systematically assess a research gap across 6 dimensions

Classify gap root causes using AHRQ 4-reason framework (insufficient

Strategy: What-If Analysis, Alternative Futures, and Four Ways of Seeing

Generate 2-4 divergent scenarios from the same evidence base, each representing

Generate alternative model formulations by relaxing, replacing, or generalizing

Score each candidate alternative against all criteria to produce a score

Generate alternatives for every known approach — ensure no approach goes

Systematic structure-mapping from source to target domain (Gentner).

Chain analogies to deeper levels (3-5 layers). Each layer reveals new

Extract transferable structural principles from source domains. Orchestrates

Assess analogy depth (surface/structural/systemic). Determines whether

KAOS-style recursive goal decomposition. AND decomposition for sub-goals

SOP: Describe and classify anomalous phenomena that existing theory cannot explain

Tactic: Inductive/abductive path — describe anomalous phenomena, generate

SOP: Design the optimal answering order for sub-questions

Challenge industry best practices' hidden assumptions. Deconstruct benchmarks

Search for positive deviants and extract transferable principles using

Find positive deviants and reframe the problem from deficit-based to

Establish acceptability standards through RAND/UCLA Appropriateness Method

Distill the strongest arguments from each perspective through Argument

Extract and steel-man the core arguments supporting a given opinion cluster.

Detect annotation artifacts and shortcuts in benchmarks

Understand hard boundaries on the user's research — target venues, methodology

Present the GoalTree to the user for confirmation. Ask about reasonableness,

Deep WHY probing inspired by i* Intentionality modeling. Understand the

Present obstacles with their severity assessments and proposed mitigations

Rate each identified obstacle's difficulty — overcomability, time cost,

Standardize assignee names and identify corporate group affiliations

Surface all assumptions, classify by vulnerability (load-bearing × likely-false),

Build assumption dependency graphs and trace cascade failures when root

Tactic: Surface assumptions, sort by dependency, attack root assumptions

Construct the strongest counter-argument against a specific assumption and propose alternatives.

Challenge each assumption's validity — shared cross-repo SOP

Which assumptions are most fragile? — Vulnerability ranking + impact

Measure how much conclusions change when each assumption is negated.

Assumption Destruction Campaign — open new solution spaces by negating,

Systematically identify all assumptions in a method/model — structural, parametric, distributional, and scope assumptions.

Systematic extraction, challenge, and sensitivity analysis of assumptions

Systematically extract all assumptions (stated, implicit, boundary, mathematical, practical) from a method or model.

Classic reductio ad absurdum: negate the core claim, derive logical

One-at-a-time assumption perturbation — extract assumptions, define negations, re-derive conclusions under each negation, measure sensitivity. Identifies which assumptions are load-bearing.

Systematically extract implicit assumptions from methods, frameworks, or arguments. Identifies what is taken for granted without explicit justification.

Compute overall resilience score (0.0-1.0) based on attack results, coverage,

Generate specific attack strategies for a given threat surface, producing

Identify and suspend fundamental assumptions via de Bono PO. Systematically

Gather independent ranking ballots from multiple judges or perspectives

SOTA Performance Baseline Campaign — 5 strategies for systematically

Select appropriate baselines for experimental comparison

Produce final structured baseline report integrating all analysis results

Evaluation Methodology Archaeology Campaign — 5 strategies for systematic

Systematic quality assessment using BetterBench 46-criterion framework

Identify and negate benchmark assumptions. Deconstruct best practices

Identify and catalog all relevant benchmarks in target domain

Systematically scan all known solutions, identify gaps in coverage and

Provides information about the bitwize-music plugin, its version, and its creator. Use when the user asks about the plugin, its purpose, version, or capabilities.

Creates visual concepts for album artwork and generates AI art prompts. Use during planning for concept discussion, or after all tracks are Final for actual artwork generation.

Designs album concepts, tracklist architecture, and thematic planning through 7 structured phases. Use when planning a new album or reworking an existing album concept.

Shows a structured progress dashboard for an album with percentage complete per phase, blocking items, and status breakdown. Use for a quick visual overview of album progress.

Tracks and manages album ideas including brainstorming, planning, and status updates. Use when the user wants to add, review, or organize their album idea backlog.

Copies track content (lyrics, style prompts, streaming lyrics) to the system clipboard. Use when the user needs to paste lyrics or style prompts into Suno or other external tools.

Uploads promo videos and content to Cloudflare R2 or AWS S3. Use when the user wants to host promo content for social media or distribution.

Sets up or edits the plugin configuration file interactively. Use on first-time setup, when config is missing, or when the user wants to change settings.

Searches and retrieves documents from free public sources using automated browser navigation. Use when research needs primary source documents like court filings, government reports, or public records.

Scans lyrics for explicit content and verifies that explicit flags match actual content. Use before Suno generation or release to ensure accurate content ratings.

Create new genre documentation files for the bitwize-music genre library. Use when the user wants to add a genre, says "/genre-creator", "neues Genre erstellen", "Genre hinzufuegen", "add genre", or asks to create genre documentation. Takes a genre name as argument.

Runs plugin health checks (venv packages and skill registration). Use when the user asks to check plugin health, verify setup, or troubleshoot missing skills.

Shows available skills, common workflows, and quick reference for the plugin. Use when the user asks for help, what skills are available, or how to do something.

Places album art files in the correct audio and content directory locations. Use when the user has generated or downloaded album artwork that needs to be saved.

Moves audio files to the correct album location with proper path structure. Use when the user has downloaded WAV files from Suno or other sources that need to be organized.

Moves track markdown files to the correct album location. Use when the user has track files in Downloads or other locations that need to be placed in an album.

Autonomous multi-pass lyric refinement for tightening, cohesion, and album unity. Use after lyrics are written to polish a track or entire album through iterative passes.

Reviews lyrics against a quality checklist before Suno generation. Use before generating tracks to catch rhyme, prosody, pronunciation, and structural issues.

Writes or reviews lyrics with professional prosody, rhyme craft, and quality checks. Use when writing new lyrics, revising existing lyrics, or when the user says 'let's work on a track.

Guides audio mastering for streaming platforms including loudness optimization and tonal balance. Use when the user has approved tracks and wants to master audio files.

Polishes raw Suno audio by processing per-stem WAVs (vocals, backing_vocals, drums, bass, guitar, keyboard, strings, brass, woodwinds, percussion, synth, other) with targeted cleanup, EQ, and compression, then remixing into a polished stereo WAV ready for mastering. Use after audio import and before mastering.

Creates a new album with the correct directory structure and templates. Use IMMEDIATELY when the user says 'make a new album' or similar, before any discussion.