product-ai-public

KO | EN

gitlite — search

#python #typescript #golang #machine-learning #rust #linux #java #nodejs #kubernetes #vue #security #openai #mcp #open-source

product-ai-public

(★ 16)

Curated skills, commands, and agents for AI-assisted product management (and life)

product-ai-public 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Product AI Public

Curated prompts, commands, skills, and agents for AI-assisted product management. A subset of my personal "second brain" repo, published for others to adapt. See [How I AI](https://elezea.com/how-i-ai/) for a full series of posts on how this system works in practice.

## What's here

These prompts, context files, and commands work with Claude, ChatGPT, OpenCode, Cursor, or any LLM-based coding/writing assistant.

### Context files

* [`product-philosophy.md`](01-context/product-philosophy.md) -- How I think about product work: problem-first, outcomes over outputs, empowered teams
* [`avoid-ai-patterns.md`](01-context/avoid-ai-patterns.md) -- Words, phrases, and structural patterns that make AI writing obvious
* [`project-template.md`](01-context/project-template.md) -- Template for creating a dedicated "project brain" folder for major initiatives
* [`stable-facts-template.md`](01-context/stable-facts-template.md) -- Template for the stable facts file used by cross-session memory

### Prompts

**Product management**

* [`draft-review-prd.md`](02-prompts/pm/draft-review-prd.md) -- Draft or review PRDs using a structured Product Opportunity Assessment framework
* [`review-okrs.md`](02-prompts/pm/review-okrs.md) -- Draft or review OKRs with outcome-over-output discipline and scoring rubrics
* [`debate-product-idea.md`](02-prompts/pm/debate-product-idea.md) -- Simulate a high-stakes product strategy debate with multiple perspectives
* [`review-spec.md`](02-prompts/pm/review-spec.md) -- Review engineering specs to ensure they faithfully translate product requirements

**Development**

* [`brainstorming-planning.md`](02-prompts/dev/brainstorming-planning.md) -- Build ideas through guided discovery, then produce developer-ready specs

**Meta**

* [`generate-writing-style.md`](02-prompts/meta/generate-writing-style.md) -- Analyze writing samples and produce a reusable system prompt for style emulation

### Skills

* [`pm-thinking`](.opencode/skills/pm-thinking/SKILL.md) -- Apply problem-first, outcome-focused product thinking to any product task
* [`interactive-explainer`](.opencode/skills/interactive-explainer/SKILL.md) -- Build self-contained interactive HTML visualizations of processes, algorithms, or systems
* [`linear-walkthrough`](.opencode/skills/linear-walkthrough/SKILL.md) -- Generate narrative walkthroughs of code, documents, or systems for learning and onboarding
* [`discord-bot`](.opencode/skills/discord-bot/SKILL.md) -- Discord bot development guide: Developer Portal setup, Gateway vs Interactions architecture, Cloudflare Workers integration, common gotchas, and verification checklist
* [`check-skill`](.opencode/skills/check-skill/SKILL.md) -- Audit any SKILL.md against best practices: 5 skill killers, anatomy completeness, scoping, and context binding
* [`diagnosing-mac-performance`](.opencode/skills/diagnosing-mac-performance/SKILL.md) -- Triage a slow Mac: snapshot CPU/memory/swap, identify runaway processes and memory pressure, and recommend fixes ordered by disruption
* [`workers-vitest-pool`](.opencode/skills/workers-vitest-pool/SKILL.md) -- Set up or upgrade vitest tests for Cloudflare Workers: known-good stack (vitest 4 + pool-workers 0.15 + Node 22) and the v2-to-v4 migration recipe that fixes the `.sqlite-shm` isolation bug
* [`system-file-guardrails`](.opencode/skills/system-file-guardrails/SKILL.md) -- Safety checks for editing AI system files (skills, commands, agents): pre-edit reference checks, post-edit validation, post-session coherence
* [`finding-original-cds`](.opencode/skills/finding-original-cds/SKILL.md) -- Pick the best original (non-remaster) CD pressing on Discogs; a personal-hobby example of working around MCP tool gaps by hitting the public API directly

### Agents

* [`code-review`](.opencode/agent/code-review.md) -- Evidence-based code review with tool-assisted validation and provability classification
* [`blind-validator`](.opencode/agent/blind-validator.md) -- Independently re-fetch and verify every source cited in a draft document
* [`challenger`](.opencode/agent/challenger.md) -- Adversarial reasoning reviewer that finds logic errors, weak explanations, unsupported claims, and gaps that mechanical checklists miss
* [`fix-assessor`](.opencode/agent/fix-assessor.md) -- Read code, map change surface area, check CI/tests, and score feasibility criteria for a proposed fix
* [`editor`](.opencode/agent/editor.md) -- Ruthless style editor that runs ten scoped passes over a document to catch AI patterns, overused words, banned sentence structures, and style-guide violations

### Commands

* [`/assess-fix`](.opencode/command/assess-fix.md) -- Assess whether a Jira ticket's fix can be implemented by a technical PM and LLM coding agent, with a Go/No-Go verdict and implementation brief
* [`/code-review`](.opencode/command/code-review.md) -- Review changes using three parallel code-review agents, then validate with a fourth
* [`/meeting`](.opencode/command/meeting.md) -- Transform a meeting transcript into structured notes with decisions and action items
* [`/doc-review`](.opencode/command/doc-review.md) -- Review generated files for accuracy, completeness, and style
* [`/new-project`](.opencode/command/new-project.md) -- Scaffold a new project brain folder
* [`/session-start`](.opencode/command/session-start.md) -- Load recent session context and corrections at the start of a work session
* [`/session-end`](.opencode/command/session-end.md) -- Write a session handoff note before ending a work session

### Docs

* [`cross-session-memory.md`](04-docs/cross-session-memory.md) -- A four-layer memory system that gives AI agents continuity between sessions
* [`multi-agent-investigation.md`](04-docs/multi-agent-investigation.md) -- Architecture for multi-agent customer escalation investigations with parallel dispatch and blind validation

### Personal

* [`books/CLAUDE.md`](05-personal/books/CLAUDE.md) -- Agent instructions for a per-book reading-notes folder: progress-aware, spoiler-safe, web-search-first for factual claims
* [`books/the-will-of-the-many.md`](05-personal/books/the-will-of-the-many.md) -- Example book notes file following the pattern (characters, magic system, worldbuilding, plot threads)
* [`antiphon/CONTEXT.md`](05-personal/antiphon/CONTEXT.md) -- Operating instructions for a personal Last.fm music recommender: live listening data via MCP, Spotify search links, mood buckets, and signal-cited picks
* [`antiphon/templates/`](05-personal/antiphon/templates/) -- Empty starter files for the recommender pattern: user config, the three-state mood library, and an anti-rec list
* [`therapy-session-summary.md`](02-prompts/meta/therapy-session-summary.md) -- Summarize therapy session transcripts into structured notes with IFS (Internal Family Systems) support

## How to use these

Most of these files are **system prompts** or **prompt templates** you can paste into any LLM conversation. Some are designed for specific tools:

* **`01-context/`** files work well as persistent context (system prompts, custom instructions, or project knowledge files)
* **`02-prompts/`** files are task-specific prompts you paste when needed
* **`.opencode/command/`** files are [OpenCode](https://opencode.ai) slash commands (but the prompt content works anywhere)
* **`.opencode/skills/`** files are on-demand skills that get loaded when specific topics come up
* **`.opencode/agent/`** files define specialized subagents for specific tasks

## Adapting to your context

These prompts reference a few conventions you'll want to adapt:

* References to `01-context/*.md` files point to personal context documents (writing style, philosophy, etc.). Create your own equivalents.
* Some prompts mention MCP servers or tool integrations. Replace with whatever tools you use.
* The writing style and product philosophy reflect my preferences. Adjust to match yours.

## About

I'm [Rian van der Merwe](https://elezea.com), a product manager who uses AI tools extensively in my daily work. This repo contains the prompts and context files I've found most useful, published so others can adapt them.

For more on my approach to product management and AI, see [`product-philosophy.md`](01-context/product-philosophy.md).

## License

MIT. Use these however you like.

product-ai-public

product-ai-public 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

product-ai-public

product-ai-public 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README