adk-docs
guidelines for creating, reviewing, updating, and searching ADK documentation - use when users ask about writing, maintaining, or auditing ADK bot docs
Works with
---
name: adk-docs
description: guidelines for creating, reviewing, updating, and searching ADK documentation - use when users ask about writing, maintaining, or auditing ADK bot docs
license: MIT
---
# ADK Documentation Management
Use this skill when helping users create, review, update, or maintain documentation for their ADK projects. Users write guides and docs for their own bots and features — code examples should come from their project and the official ADK repo.
## When to Use This Skill
Activate this skill when users:
- Want to create documentation for their bot ("Document my order workflow", "Write a guide for my custom actions")
- Need to review existing project docs for accuracy ("Check if my docs are still correct")
- Want to update docs after changing their bot ("I refactored the checkout flow, update the docs")
- Need to check if docs are in sync with their code ("Are my docs current?")
- Want to search their project documentation ("Find docs about my payment handler")
- Ask about documentation standards or templates
- Mention `/adk-document`
## Available Documentation
| File | Description |
| ------------------------------------------------------------ | ----------------------------------------------------------------- |
| [references/doc-standards.md](./references/doc-standards.md) | Document types, templates, quality checklists, and health metrics |
## Available Commands
| Command (and suggested subcommand) | Description |
| ------------------------------------------------ | ----------------------------------------------------- |
| `/adk-document create [topic]` | Create documentation for a feature in the user's bot |
| `/adk-document review [doc-path]` | Review project docs for accuracy and completeness |
| `/adk-document update [doc-path] [what-changed]` | Update project docs after code changes |
| `/adk-document sync [optional-doc-path]` | Check if project docs are in sync with the bot's code |
| `/adk-document search [search-term]` | Search project documentation for specific topics |
## Shared Principles
### 1. AI-Optimized Structure
- Clear section headers (`##`, `###`, `####`) so ripgrep can find sections
- Table of contents at top with anchor links
- Right-sized for document type (see doc-standards reference)
- Keyword-rich section names — no vague "Advanced Topics" or "Other"
### 2. Code-First Approach
- Every concept needs a working code example from actual project code
- Include file paths with line numbers for verification
- Primary sources for examples (in priority order):
1. The user's own ADK project (look for `agent.config.ts` in the workspace) — BEST
2. The official ADK repo examples (clone or find locally)
3. ADK runtime usage (`@botpress/runtime` packages)
- Never invent or speculate examples — all code must be verifiable
### 3. Critical Distinctions
- ADK primitives (from `@botpress/runtime`) vs Botpress SDK primitives
- `this.send()` in conversations vs `client.createMessage()` in workflows
- Messages (persistent, stored) vs Events (ephemeral, not stored)
- Agnostic APIs vs channel-specific features
### 4. No Speculation
- Do NOT add Common Mistakes or Best Practices sections unless the user explicitly provides them
- Use `❌ WRONG` / `✅ CORRECT` only when documenting actual reported errors
- Workflows and examples must be verified from actual code, not imagined
### 5. Writing Style
- **Direct and actionable** — "Use `this.send()`" not "You might want to consider..."
- **Technically accurate** — test examples against actual ADK code
- **Assumes intelligence** — provide context and guidance, not hand-holding
- **No marketing fluff** — straight to the technical substance
## How to Answer
When a user asks about documentation without invoking a specific command:
1. **Load the `adk` skill (& more, if needed)** for ADK context and knowledge
2. **Identify what they need** — creating, reviewing, updating, syncing, or searching
3. **Point them to the right command** or help directly using the principles above
4. **Reference doc-standards** for template and quality guidance
## Documentation Location
Documentation is written in the user's own project. Ask the user where they want docs saved if not obvious (common locations: `./docs/`, `./guides/`, or project root).
## Discovering Code Sources
Discover the user's project and the official ADK repo for examples:
```javascript
// Find ADK projects in the workspace
Glob({ pattern: '**/agent.config.ts' })
// Find ADK runtime usage in the user's project
Grep({ pattern: 'from [\'"]@botpress/runtime', output_mode: 'files_with_matches' })
// Look for existing project documentation
Glob({ pattern: './{docs,guides}/**/*.md' })
// Find official ADK examples if available locally
Glob({ pattern: '**/adk/examples/**/*.ts' })
```
## Cross-References
- **`adk` skill** — Core ADK knowledge (actions, workflows, conversations, etc.)
- **`adk-evals` skill** — Testing and eval documentation
- **`adk-frontend` skill** — Frontend integration documentation
- **`adk` skill `integrations.md` reference** — Integration lifecycle documentationMore Writing & Documentation skills
paper-context-resolver
lllllllama/rigorpilot-skills
Rigor Paper Context helper for README-first deep learning repo reproduction. Use only when the README and repository files leave a narrow reproduction-critical gap and the task is to resolve a specific paper detail such as dataset split, preprocessing, evaluation protocol, checkpoint mapping, or runtime assumption from primary paper sources while recording conflicts. Do not use for general paper summary, repo scanning, environment setup, command execution, title-only paper lookup, or replacing README guidance by default.
repo-intake-and-plan
lllllllama/rigorpilot-skills
Rigor Intake helper for README-first deep learning repo reproduction. Use when the task is specifically to scan a repository, read the README and common project files, extract documented commands, classify inference, evaluation, and training candidates, and return the smallest trustworthy reproduction plan to the main orchestrator. Do not use for environment setup, asset download, command execution, final reporting, paper lookup, or end-to-end orchestration.
minimal-run-and-audit
lllllllama/rigorpilot-skills
Rigor Run skill for README-first deep learning repo reproduction. Use when the task is specifically to capture or normalize evidence from the selected smoke test or documented inference or evaluation command and write standardized `repro_outputs/` files, including patch notes when repository files changed. Do not use for training execution, initial repo intake, generic environment setup, paper lookup, target selection, hidden scientific-meaning changes, or end-to-end orchestration by itself.

