Custom Agent Skills Collection

Custom skills for AI agents that provide specialized domain knowledge and structured methodologies.

How to Use

These skills are designed to be automatically used by an AI agent when a user asks for help with a task covered by one of the skills.

Installation

1. Clone or download skills folder and its contents. Optionally also grab skills-compact/ — compact (minified) variants of some skills for token-economical use.
2. Add skills from the skills folder (or, for a compact skill, from skills-compact/ instead)
   • globally to ~/.agents/skills
   • or to your IDE / CLI Code tool:
     • Cursor: ~/.cursor/skills
     • Cloude Code: ~/.claude/skills/
     • Other IDE or AI agent: check documentation on where skills are located
   • or to a specific project:
     • /.agents/skills
3. [Re}start an AI-agent to use the skill during task execution.

Usage

Example workflow — Business Analysis:

• User: “We’re kicking off a payments feature — help me pin down stakeholders and requirements”
• AI: Loads dpf-business-analysis/SKILL.md from skills and applies its requirements lifecycle patterns
• AI identifies stakeholders, elicits and specifies requirements, surfaces ambiguities, and builds a traceable, prioritized baseline

Example workflow — Crafting a skill:

• User: “I keep pasting the same release-checklist instructions to my agent — turn that into a skill”
• AI: Loads agent-skill-builder/SKILL.md from skills and applies its authoring patterns
• AI checks that a skill is the right layer (vs AGENTS.md/MCP), designs the trigger description, progressive disclosure, and produces a validated SKILL.md + references/

The skills transform the AI into a specialized assistant that follows proven methodologies rather than ad-hoc responses.

Table of Contents

• Custom Agent Skills Collection
  • How to Use
    • Installation
    • Usage
  • Table of Contents
  • Frameworks & Methodologies
    • 🧠 First Principles Framework (FPF Core)
    • 🔬 Episteme Compaction
    • 📖 Pattern Language as Agent Skill (PLAS)
    • 🏗️ Layered Framework Workspace Architecture
  • Business Analysis & Requirements Engineering
    • 📋 Business Analysis & Requirements Engineering
  • Text Analysis & Writing
    • 🔍 Reverse-Engineering Texts
  • Document Processing
    • 📄 PDF to Markdown (pdf2md)
  • Development Tools
    • 🔄 FPF Sync
    • 🔧 Agent Skill Builder
  • Project Management
    • 🗂️ Project Vault
  • Compact Skills
    • 🗜️ Business Analysis & Requirements Engineering (compacted)
    • 🗜️ Project Vault (compacted)
    • 🗜️ Episteme Compaction (compacted)
    • 🗜️ Pattern Language as Agent Skill (compacted)
  • Space Planning & Organization
    • 🚿 Bathroom Planner
    • 🧥 Wardrobe Planner

Frameworks & Methodologies

Structured reasoning frameworks and domain-specific methodologies for systematic problem-solving.

🧠 First Principles Framework (FPF Core)

Root framework and pattern library providing auditable thinking, evidence chains, and systematic problem-solving patterns. Serves as the governing foundation for all DPF and LPF skills. Patterns are loaded individually via a fast index — never the full spec: pattern reference files + INDEX.md from references/, plus orienting context sections in context/.

Use when:

• Guiding reasoning on engineering, research, and management tasks
• Requiring systematic evidence-based problem-solving
• Building structured, verifiable solutions with traceable evidence

Reference: GitHub Repository of the original First Principles Framework by Anatoly Levenchuk

Location: skills/fpf-core/

🔬 Episteme Compaction

Lossless compaction of epistemes: prose ↔ compact typed-slot notation at F4-F5 formality level (~65% median token savings, measured on the instruction-file corpus) with a source-claims vs [bracketed] split and reconstruction back to source. Designed for AI-agent or FPF-literate-human output where token economy and auditability matter — diagnostics, architecture decisions, code reviews, status reports, trust assessments, system compositions. This is compaction, NOT narrativization.

Depends on: fpf-core

Use when:

• Writing diagnostics, ADRs, code reviews, or status reports for AI agent consumers
• Rendering prose claims as typed slots and Γ aggregation with F-G-R-CL trust metrics
• Composing evidence provenance / ADI reasoning blocks
• Reconstructing prose back from a compact episteme (reverse render)
• Any context where token economy AND auditability matter

Do NOT use for: narrativization, casual chat, teaching, non-technical audiences, creative tasks.

Location: skills/episteme-compaction/

Compact variant: skills-compact/episteme-compaction/ (see Compact Skills).

📖 Pattern Language as Agent Skill (PLAS)

Author and improve a Domain Principle Framework (DPF) or Local Practices Framework (LPF) directly in Agent-skill form (SKILL.md + references/). The edition carrier is the skill itself: references/*.md are the canonical pattern bodies — no monolith, no reader-facing publication form.

Depends on: fpf-core, agent-skill-builder

Use when:

• Deciding whether and what to author as a DPF-skill (cold start)
• Authoring a DPF or LPF as an agent skill (FPF-grounded or self-sufficient variant)
• Evaluating, improving, refreshing a DPF or LPF as a skill

Location: skills/pattern-language-as-agent-skill/

Compact variant: skills-compact/pattern-language-as-agent-skill/ (see Compact Skills).

🏗️ Layered Framework Workspace Architecture

Manages the FPF/DPF/LPF workspace structure: where to place frameworks, how to organize skill carriers, dependency chains, and LPF vs Project boundaries.

Depends on: fpf-core, pattern-language-as-agent-skill

Use when:

• Organizing the FPF/DPF/LPF workspace layout
• Deciding where a new framework belongs in the ecosystem
• Setting up skill-carrier structure for a DPF/LPF
• Resolving LPF vs Project boundary questions
• Packaging a DPF/LPF as a distributable skill

Location: skills/dpf-lfw-architecture/

Business Analysis & Requirements Engineering

Domain practical framework for structured business analysis and requirements engineering in engineering projects.

📋 Business Analysis & Requirements Engineering

Covers the full requirements lifecycle: stakeholder identification, elicitation, specification, prioritization, validation, traceability, and change management. Includes business process modeling, use case modeling, and data/security requirements.

Depends on: fpf-core

Use when:

• Identifying and managing stakeholders
• Eliciting, specifying, or prioritizing requirements
• Validating or tracing requirements
• Managing requirements changes
• Modeling business processes or use cases
• Defining data or security requirements

Location: skills/dpf-business-analysis/

Text Analysis & Writing

Skills for analyzing existing texts and extracting reusable structure, argumentation, and rhetorical patterns.

🔍 Reverse-Engineering Texts

Reverse-engineer texts into a reusable structural blueprint (reverse outlining, argument mining, copy teardown).

Use when:

• Analyzing a text’s thesis, hook, and macro-structure
• Extracting argument maps and support/evidence moves
• Mapping rhetoric (Ethos/Pathos/Logos)
• Converting “text → structure/blueprint” to reuse patterns in writing, specs, prompts, or marketing

Location: skills/reverse-engineering-texts/

Document Processing

Skills for converting and extracting content from document formats.

📄 PDF to Markdown (pdf2md)

Extracts PDF content into clean Markdown: text via pdfplumber + image/diagram descriptions via a local vision-language model (LM Studio, Ollama, or any OpenAI-compatible endpoint). Supports recursive batch processing, resume/retry, and custom VL prompts.

Use when:

• Converting presentation decks (slides, diagrams) to text
• Extracting conference talks, workshops, or technical reports from PDF
• Transcribing PDFs where plain text extraction is garbled
• Batch-processing directories of PDFs with automatic skip of already-processed files

Do NOT use for: pure-text PDFs (use pdfplumber directly), filling PDF forms (use the pdf skill), scanned documents without selectable text.

Dependencies: pip install pdfplumber pypdfium2 pillow requests + local VL model (default: qwen/qwen3-vl-8b via LM Studio).

Location: skills/pdf2md/

Development Tools

Tools for creating and managing AI agent skills and FPF/DPF/LPF documentation.

🔄 FPF Sync

Keeps the fpf-core skill in sync with the canonical FPF monolith (FPF-Spec.md). Self-contained tool that pulls the latest spec, decompiles it into atomic reference files, and updates the skill metadata.

Use when:

• The FPF specification has been updated (new commits upstream)
• The fpf-core skill references are stale or show wrong counts
• Running scheduled maintenance on layered framework workspace skills

Dependencies: Python 3.10+, git in PATH.

Location: skills/fpf-sync/

🔧 Agent Skill Builder

Local Practices Framework (LPF) for authoring, reviewing, and evolving agent skills: whether a skill is even the right guidance layer (vs AGENTS.md / MCP / Memory), trigger design (description) and its empirical optimization, progressive disclosure, skill anatomy & frontmatter, atomicity, inter-skill composition, the end-to-end evaluation loop, the weak-model quality gate, placement & portability, evolution, security & trust, and opt-in material compaction. 13 ASB.* patterns — routing SKILL.md + references/.

Depends on: fpf-core, skill-creator, episteme-compaction (some cases)

Use when:

• Creating a new skill or reviewing an existing one
• Deciding between skills vs AGENTS.md vs MCP vs Memory
• Designing atomic, reusable skills with progressive disclosure
• Unsure if a skill is needed at all
• Compacting a skill’s materials to a minimal size (opt-in, on explicit request)

Location: skills/agent-skill-builder/

Project Management

Skills for tracking project state, decisions, risks, and contradictions in a structured markdown vault.

🗂️ Project Vault

Local Practices Framework (LPF) for managing project state in a structured markdown vault: atomic cards for decisions, open questions, risks, contradictions, operational tracks, and work records. Governs working tracks as the mandatory container for all productive activity (research, analysis, synthesis). Routing SKILL.md + 10 PV.* patterns covering vault init, inbox/outbox processing, state updates, external research binding, track-bound work, and report generation.

Depends on: fpf-core

Use when:

• Initializing a project vault from scaffold (PV.Init)
• Processing the inbox (transcripts, PDFs, articles, research) into atomic decision/question/risk/contradiction cards (PV.Inbox)
• Updating project state from dialogue briefings or meeting transcripts (owner chat updates) (PV.StateUpdate)
• Processing external research with two-way binding to project entities (PV.ExternalResearch)
• Managing track-bound productive work (track lifecycle, artifacts, work records) (PV.Track, PV.Artifact, PV.WorkRecord)
• Generating reports and next-meeting agendas from open questions, blockers, and contradictions (PV.Report)
• Sending outgoing feedback or proposals to another system or skill (PV.Outbox)
• Reconciling new information against the open question registry and archiving closed entities

Location: skills/project-vault/

Compact variant: skills-compact/project-vault/ (see Compact Skills).

Compact Skills

Compact (minified) runtime projections of some skills — smaller, token-economical variants for agents that need the method without the full provenance, derivation, and readiness metadata. A compact skill keeps only the runtime allow-list (method + trigger + routing); its references/*.md bodies are episteme blocks with no per-card frontmatter, and it names its canonical source. The full skills/ carrier remains the editable source of truth.

🗜️ Business Analysis & Requirements Engineering (compacted)

Compacted runtime projection of dpf-business-analysis: the same requirements-lifecycle patterns (stakeholder identification, elicitation, specification, prioritization, validation, traceability, change management) plus business-process / use-case / data / security modeling as 12 episteme-block patterns with an FPF grounding slot per card. Readiness (collective): source-faithful.

Depends on: fpf-core

Canonical source: skills/dpf-business-analysis/ (the full, editable DPF carrier).

Location: skills-compact/business-analysis/

🗜️ Project Vault (compacted)

Compacted runtime projection of project-vault: the same vault schema, inbox → state-update → external-research → track/work-record lifecycle, and outbox as 10 episteme-block patterns, plus the unchanged vault scaffold. Readiness (collective): source-faithful.

Depends on: fpf-core

Canonical source: skills/project-vault/ (the full, editable LPF carrier).

Location: skills-compact/project-vault/

🗜️ Episteme Compaction (compacted)

Compacted runtime projection of episteme-compaction: the same typed-slot rendering pipeline (formality gate → core render → aggregation/trust → evidence/reasoning → distinction compliance → admission), reverse render, and round-trip governance as 8 episteme-block ECPF.* patterns. No INDEX.md, relations.md, graph generator, or evals are shipped — the dependency graph and provenance live in the canonical source. Readiness (collective): source-faithful.

Depends on: fpf-core

Canonical source: skills/episteme-compaction/ (the full, editable carrier).

Location: skills-compact/episteme-compaction/

🗜️ Pattern Language as Agent Skill (compacted)

Compacted runtime projection of pattern-language-as-agent-skill: the same DPF/LPF-as-skill authoring patterns — entry route, skill layout, dispatcher, pattern body, governing cues, naming, quality & refresh, self-sufficient variant, and the compacted-projection pattern itself — as episteme-block PLAS.* patterns plus a routing INDEX.md. Readiness (collective): source-faithful.

Depends on: fpf-core, agent-skill-builder, episteme-compaction

Canonical source: skills/pattern-language-as-agent-skill/ (the full, editable LPF carrier).

Location: skills-compact/pattern-language-as-agent-skill/

Space Planning & Organization

Skills for optimizing physical spaces and organizing belongings with systematic methodologies.

🚿 Bathroom Planner

Structured 10-stage methodology for planning bathroom layouts with focus on ergonomics, functionality, and safety.

Use when:

• Planning bathroom furniture placement
• Optimizing bathroom space
• Arranging bathroom fixtures (toilet, sink, bathtub, washing machine)
• Solving bathroom layout challenges

Applicable to: Large bathrooms (10+ square meters) with flexible plumbing.

Location: skills/bathroom-planner/

🧥 Wardrobe Planner

Structured 7-stage methodology for planning wardrobe/closet organization based on First Principles Framework.

Use when:

• Organizing closet space
• Planning wardrobe storage
• Arranging clothing and accessories
• Optimizing closet layout
• Solving wardrobe organization challenges

Location: skills/wardrobe-planner/