A public collection of Claude Code skills by Abraham Vallez — covering engineering management, product development, and team practices.
npx skills add abrahamvallez/skillsUses skills CLI — supports Claude Code and 40+ other AI agents (Cursor, Copilot, Codex, etc.). The CLI will prompt you to select which skills to install.
| Skill | Category | What it does | Key facts |
|---|---|---|---|
1on1-coach |
Management | Analyzes a 1on1 transcript and generates 3 output documents for the manager | Auto-detects language · 7 meeting types · Fathom / Otter / plain text |
bokata-research |
Product Dev | Runs context research before mapping or slicing a feature | 3 research phases · outputs feature-context.md |
bokata-feature-mapper |
Product Dev | Maps actors, features, and user tasks from a PRD or description | User Story Mapping · bundling heuristics · Phase 0 discovery |
bokata-ac-analyst |
Product Dev | Generates Gherkin acceptance criteria from a Features Backbone | Rule-first · happy path + edge cases + permissions |
bokata-feature-slicer |
Product Dev | Decomposes a feature into a Walking Skeleton + Increments Backlog | 16+ slicing strategies · Walking Skeleton buildable in 1–3 days |
Category: Management
Analyzes a 1on1 meeting transcript and produces three output documents:
| File | Audience | Contents |
|---|---|---|
outputs/shareable/{date}-{name}-summary.md |
Team member | Neutral recap, key points, verbatim quotes, action items |
outputs/coaching/{date}-{name}-analysis.md |
Manager (private) | Meeting effectiveness, Radical Candor quadrant, risks, improvement areas |
outputs/coaching/{date}-{name}-coaching.md |
Manager (private) | Socratic questions to reflect and prepare for the next meeting |
Meeting types: check-in · career · unblocking · feedback · conflict · onboarding · mixed
Language: auto-detected from the transcript. Override with lang=es, lang=en, etc.
Compatible with: Fathom, Otter.ai, and any plain-text transcript format.
Usage:
Analyze this 1on1: transcripts/2026-04-28-carlos.txt
Run 1on1-coach on my Fathom transcript from today's check-in with Alex
Process transcripts/2026-04-10-paula.txt lang=en
Category: Product Dev · Bokata Framework
Runs three research phases to produce feature-context.md — a context document that enriches all downstream Bokata skills.
| Phase | What it produces |
|---|---|
| Feature Research | Domain vocabulary, actors, existing patterns |
| Criteria Research | Business rules, permissions, state transitions per task |
| Slicer Research | Tech stack, architecture constraints, available libraries |
Run this once per initiative, before bokata-feature-mapper or bokata-feature-slicer.
Usage:
Run bokata-research for this initiative:
[paste PRD or description]
Category: Product Dev · Bokata Framework
Maps a PRD or initiative description into a structured Features Backbone using User Story Mapping methodology.
- Features in
[Actor] [Verb] [Object]format - User Tasks in
[Verb] [Object]format (actor inherited from feature) - Phase 0 discovery: asks clarifying questions or documents assumptions
- Bundling heuristics: detects tasks merged into one that should be split
Output: ## Features Backbone with features, user tasks, system tasks, and a dependency map.
Usage:
Run bokata-feature-mapper:
[paste PRD, description, or conversation context]
Run bokata-feature-mapper with this enriched context:
[paste PRD]
## Feature Research Summary
[paste bokata-research output]
Category: Product Dev · Bokata Framework
Generates Gherkin acceptance criteria (Given/When/Then) for a set of User Tasks using Feature Mapping and Example Mapping.
- Rule-first: defines business rules before writing scenarios
- Covers happy path, error states, edge cases, and permission boundaries
- Output scenarios are cross-linked to backbone IDs
Usage:
Run bokata-ac-analyst for these User Tasks:
- Create Match
- Join Match
- Leave Match
Run bokata-ac-analyst:
[paste ## Features Backbone section]
## Criteria Research Summary
[paste bokata-research output]
Category: Product Dev · Bokata Framework
Takes a single feature and decomposes it into a Walking Skeleton + Increments Backlog using Vertical Slicing (the Hamburger Method).
- Walking Skeleton: thinnest end-to-end slice that touches all layers — buildable in 1–3 days, ships to production
- Increments Backlog: prioritized list of follow-on slices building on top of the skeleton
- Uses 16+ breakdown strategies: Zero/One/Many, Dummy to Dynamic, SPIDR, Workflow Simplification, and more — see
breakdown-strategies.md
Usage:
Run bokata-feature-slicer:
[paste Feature definition with User Tasks]
Run bokata-feature-slicer with --show-steps:
[paste Feature definition]
## Slicer Research Summary
[paste bokata-research output]
Use --show-steps, --show-increments, or --show-all to surface intermediate phases.
Output structure:
docs/<initiative>/slices/<FEAT-ID>-<name>.md ← Walking Skeleton + Backlog
docs/<initiative>/walking-skeleton-plan.md ← Consolidated plan across features
Agents are orchestration wrappers that chain Bokata skills together and handle file I/O automatically. Skills can also be invoked standalone.
Runs the full features pipeline for an initiative: scaffolds folders, invokes bokata-research → bokata-feature-mapper → bokata-ac-analyst, assigns deterministic IDs, writes docs/<initiative>/features.md.
Run bokata-mapper-specialist for initiative "basket-chess". The PRD is in docs/PRD.md.
Slices a single feature: reads features.md, invokes bokata-feature-slicer, writes docs/<initiative>/slices/<FEAT-ID>-<name>.md, updates walking-skeleton-plan.md.
Run bokata-slicer-specialist. Slice "Player Executes Game Turn" from initiative basket-chess.
Slice BKC-FEAT-f16d from initiative basket-chess.
Skills and agents architecture inspired by eferro/skill-factory.
MIT — see LICENSE.