ajbm-skills

USE WHEN improve prompt, write prompt, craft prompt, analyze prompt, prompt engineering, subagent prompts. Analyze, craft, and improve prompts using 19 research-backed techniques. Modes - Analyze (critique existing), Craft (build from scratch), Teach (explain techniques), Quick Fix (fast improvements).

Use when creating offers, pitches, pricing, guarantees, value propositions, or marketing copy - guides through Alex Hormozi's $100M Offers methodology including Value Equation optimization, MAGIC naming, guarantee frameworks, value stacking, and scarcity/urgency to create offers so compelling prospects feel irrational saying no

USE WHEN write tweet, Twitter thread, LinkedIn post, viral hook, edit tweet, analyze tweet, improve engagement, social media copy. Applies hook fundamentals (direct reader targeting, benefit-first, avoid overused patterns), engagement triggers (social proof, scarcity, low-friction CTA), and format discipline (one idea per tweet, short paragraphs). Includes in-genre examples and iterative editing guidance.

USE WHEN agent alignment, delegation alignment, agent handoff, multi-agent delegation, agent-to-agent task transfer. Structured AI-to-AI alignment protocol that prevents context compression loss, assumption stacking, and constraint evaporation during task delegation between agents.

USE WHEN negotiate, negotiation, salary, deal, difficult conversation, confrontation, give feedback, persuade, convince, influence, get buy-in, tactical empathy, voss, never split the difference, prepare for conversation, practice negotiation, roleplay negotiation, spar. Negotiation and communication expert using Chris Voss methodology. Two modes: Analyze (produces negotiation dossier) and Spar (roleplay counterpart with inline coaching).

Use when writing, reviewing, or reporting tests. Enforces trustworthy testing through the Testing Trophy strategy, 15 evidence-based principles, disciplined mocking, AI oracle prevention, anti-pattern detection, and honest execution reporting. Applies to unit, integration, and e2e test work.

Access Twitter/X data via the `twitter` CLI tool — fetch bookmarks, search tweets, read timelines, view profiles, post tweets, and more. Cookie-based auth, no API key needed. USE WHEN: twitter, tweets, bookmarks, x.com, tweet search, twitter search, post tweet, my bookmarks, twitter feed, liked tweets, user timeline, twitter article, quote tweet, reply to tweet, twitter list, who do I follow, my twitter, tweet thread. Use this skill whenever the user wants to read from or write to Twitter/X, even if they don't explicitly mention a CLI tool. This is the primary interface for all Twitter/X operations.

Multi-agent debate with visible transcripts where specialized agents respond to each other. USE WHEN council, debate, perspectives, weigh options, deliberate, multiple viewpoints, collaborative critique. Collaborative-adversarial — use Red Team for purely adversarial attack.

Decompose to axioms, challenge inherited assumptions, reconstruct from verified truths. USE WHEN first principles, fundamental, root cause, decompose, deconstruct, challenge assumptions, rebuild from scratch, question everything.

Structured multi-angle exploration — 2-8 scientific lens passes over the same problem to surface requirements single-pass analysis misses. USE WHEN iterative depth, deep exploration, multi-angle analysis, multiple perspectives, examine from angles, surface hidden requirements, lens analysis, deeper understanding.

Adversarial analysis using parallel agents to find fatal flaws in arguments. Produces steelman plus counter-argument. USE WHEN red team, attack idea, counterarguments, critique, stress test, poke holes, devil's advocate, find weaknesses, break this, parallel analysis, adversarial validation.

Hypothesis-test-analyze cycles for systematic problem-solving — the meta-pattern underlying all the other thinking modes. USE WHEN think about, figure out, try approaches, experiment with, iterate on, improve, optimize, define goal, generate hypotheses, design experiment, measure results, analyze results, full cycle, quick diagnosis, structured investigation, science, hypothesis.

Multi-mode analytical thinking — first principles decomposition, iterative depth analysis, multi-agent council debates, adversarial red teaming, and scientific hypothesis testing. USE WHEN first principles, decompose, deconstruct, challenge assumptions, iterative depth, multi-angle, deep exploration, council, debate, perspectives, weigh options, deliberate, multiple viewpoints, red team, critique, stress test, adversarial validation, parallel analysis, devil's advocate, find weaknesses, think about, figure out, experiment, iterate, hypothesis, define goal, design experiment, quick diagnosis, structured investigation.

USE WHEN extract wisdom, analyze video, analyze podcast, analyze article, extract insights, key takeaways, what did I miss, extract from YouTube, content analysis, insight report. Content-adaptive wisdom extraction — detects what domains exist in the content and builds custom sections (not static IDEAS/QUOTES). Produces tailored insight reports from videos, podcasts, interviews, and articles at five selectable depth levels.

USE WHEN transferring a PAI skill to a standard Claude Code plugin, porting skills between environments, strip PAI specifics, adapt PAI skill for marketplace, extract skill from PAI. Guides the transformation of PAI-embedded skills into standalone plugin-compatible skills — strip PAI-specific scaffolding, decouple from the PAI Algorithm and memory system, apply Anthropic skill best practices.

USE WHEN brainstorm, generate options, break out of default thinking, need diverse alternatives, explore creative angles, the first answer feels too obvious, name something, come up with hooks, find unconventional approaches. Applies Verbalized Sampling (Zhang et al., 2024) — generate multiple low-probability variants internally before selecting output. Counteracts the centroid bias of alignment-trained models. Reports measured gains of 1.6–2.1× diversity and ~25% quality improvement.

USE WHEN setup automatic linting, add linter hook, configure code formatting on completion, enable lint on save. Auto-detects project type (JS/Python/Rust/Go/Deno/Biome), installs linter if missing, creates config with sensible defaults, wires a Stop hook in .claude/settings.json.

Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes - four-phase framework (root cause investigation, pattern analysis, hypothesis testing, implementation) that ensures understanding before attempting solutions

Use when implementing any feature or bugfix before implementation changes. Enforces test-first behavior: write a failing test, verify failure, implement minimal code, and prove green.

USE WHEN spec, requirements, interview, flesh out idea, plan feature, business idea, design review, ideation, document draft, devil's advocate, stress test, brainstorm, clarify, quick spec, scope this, help me think through, what am I missing. Crystallize ideas into actionable specs through structured elicitation, assumption surfacing, and constraint enforcement. Workflows: QuickClarify, DevSpec, BusinessIdea, DocumentDraft, DesignReview, Ideation, DevilsAdvocate.

USE WHEN distill, extract skill, capture patterns, teach from conversation, learn from session, skill from conversation. Analyzes conversations to extract user guidance patterns — corrections, questions, quality gates, analysis modes — and collaboratively distills them into permanent, replayable skills that teach Claude behavioral dispositions.

Analyze Langfuse traces, design evaluations, and iterate on AI applications. Use when user mentions langfuse, traces, spans, generations, evals, scores, datasets, or experiments. Codex uses a local Python Langfuse CLI to query real telemetry and return actionable findings.