wicked-garden 12.27.0

package/skills/integration-discovery/refs/cli-detection.md

@@ -0,0 +1,295 @@
+---
+phase_relevance: ["bootstrap"]
+archetype_relevance: ["*"]
+---
+# CLI Detection
+
+How to discover CLI tools available in the user's PATH, decide which to use, and store preferences.
+
+## Detection Method
+
+Use `command -v` for POSIX-portable detection — works on macOS, Linux, and WSL without relying on `which` (which may not be installed or may behave differently).
+
+```bash
+# Check if a single tool is available
+command -v playwright && echo "found" || echo "not found"
+
+# Check multiple tools, pick first available
+for tool in playwright puppeteer-cli cypress selenium-webdriver; do
+  if command -v "$tool" > /dev/null 2>&1; then
+    echo "Using: $tool"
+    break
+  fi
+done
+```
+
+**Why `command -v` over `which`**:
+- POSIX standard — defined by POSIX.1-2008
+- Works in all POSIX shells (bash, zsh, sh, dash)
+- Returns exit code 0 on found, non-zero on not found
+- No additional dependencies
+
+## CLI Categories
+
+Seven categories with specific tools to check:
+
+### AI CLIs
+
+Tools for interacting with AI models and coding assistants from the terminal.
+
+| Tool | Command | Description |
+|------|---------|-------------|
+| claude | `claude` | Anthropic Claude CLI |
+| codex | `codex` | OpenAI Codex CLI |
+| gemini | `gemini` | Google Gemini CLI |
+| opencode | `opencode` | OpenCode AI CLI |
+| pi | `pi` | Inflection Pi CLI |
+
+```bash
+# Detect available AI CLI
+for tool in claude codex gemini opencode pi; do
+  command -v "$tool" > /dev/null 2>&1 && echo "$tool" && break
+done
+```
+
+---
+
+### Browser Automation CLIs
+
+Tools for headless browser automation, testing, and scraping.
+
+| Tool | Command | Description |
+|------|---------|-------------|
+| playwright | `playwright` or `npx playwright` | Microsoft Playwright |
+| puppeteer | `puppeteer` or `npx puppeteer` | Google Puppeteer |
+| cypress | `cypress` or `npx cypress` | Cypress test runner |
+| selenium-webdriver | `selenium` or via pip/npm | Selenium WebDriver |
+| chrome-devtools-protocol | via `node` + CDP | Chrome DevTools Protocol |
+
+```bash
+# Priority order: playwright > puppeteer > cypress > selenium-webdriver
+for tool in playwright puppeteer cypress selenium-webdriver; do
+  if command -v "$tool" > /dev/null 2>&1 || command -v "npx" > /dev/null 2>&1; then
+    if npx "$tool" --version > /dev/null 2>&1; then
+      echo "Using: $tool"
+      break
+    fi
+  fi
+done
+```
+
+---
+
+### Cloud CLIs
+
+Tools for interacting with cloud providers and hosting platforms.
+
+| Tool | Command | Description |
+|------|---------|-------------|
+| aws | `aws` | Amazon Web Services CLI |
+| gcloud | `gcloud` | Google Cloud SDK |
+| az | `az` | Azure CLI |
+| heroku | `heroku` | Heroku CLI |
+| vercel | `vercel` | Vercel CLI |
+| fly | `fly` | Fly.io CLI |
+
+```bash
+# Detect available cloud CLI
+for tool in aws gcloud az heroku vercel fly; do
+  command -v "$tool" > /dev/null 2>&1 && echo "$tool" && break
+done
+```
+
+---
+
+### Observability CLIs
+
+Tools for monitoring, tracing, and metrics collection.
+
+| Tool | Command | Description |
+|------|---------|-------------|
+| datadog-agent | `datadog-agent` | Datadog Agent |
+| newrelic | `newrelic` | New Relic CLI |
+| dynatrace | `dynatrace` | Dynatrace CLI |
+
+```bash
+for tool in datadog-agent newrelic dynatrace; do
+  command -v "$tool" > /dev/null 2>&1 && echo "$tool" && break
+done
+```
+
+---
+
+### Data CLIs
+
+Tools for querying databases and data stores.
+
+| Tool | Command | Description |
+|------|---------|-------------|
+| duckdb | `duckdb` | DuckDB CLI |
+| psql | `psql` | PostgreSQL CLI |
+| mysql | `mysql` | MySQL CLI |
+| mongosh | `mongosh` | MongoDB Shell |
+| redis-cli | `redis-cli` | Redis CLI |
+
+```bash
+for tool in duckdb psql mysql mongosh redis-cli; do
+  command -v "$tool" > /dev/null 2>&1 && echo "$tool" && break
+done
+```
+
+---
+
+### CI/CD CLIs
+
+Tools for interacting with version control and CI/CD platforms.
+
+| Tool | Command | Description |
+|------|---------|-------------|
+| gh | `gh` | GitHub CLI |
+| glab | `glab` | GitLab CLI |
+| circleci | `circleci` | CircleCI CLI |
+
+```bash
+for tool in gh glab circleci; do
+  command -v "$tool" > /dev/null 2>&1 && echo "$tool" && break
+done
+```
+
+---
+
+### Package Managers
+
+Tools for managing project dependencies.
+
+| Tool | Command | Description |
+|------|---------|-------------|
+| npm | `npm` | Node Package Manager |
+| pip | `pip` or `pip3` | Python Package Installer |
+| cargo | `cargo` | Rust Package Manager |
+| go | `go` | Go toolchain |
+| uv | `uv` | Python package manager (fast) |
+| bun | `bun` | Bun JavaScript runtime + package manager |
+
+```bash
+for tool in bun uv npm pip3 pip cargo go; do
+  command -v "$tool" > /dev/null 2>&1 && echo "$tool" && break
+done
+```
+
+---
+
+## Auto-Decide vs Ask Policy
+
+Not all tool selection decisions have the same impact. Use this policy to determine when to decide silently, decide and inform, or ask the user.
+
+### Low Stakes — Auto-Decide Silently
+
+Decisions that are reversible and have no significant downstream impact.
+
+**Examples**:
+- Which package manager to use for a local install (npm vs pip vs uv)
+- Which database CLI for a read-only query (psql vs mysql)
+- Which AI CLI for a one-off summarization
+
+**Behavior**: Pick the first available tool in priority order. Do not mention the choice unless the user asks.
+
+```markdown
+[Auto-decided: using uv for package install — first available package manager]
+```
+
+---
+
+### Medium Stakes — Auto-Decide but Inform
+
+Decisions that affect visible outputs or workflow configuration.
+
+**Examples**:
+- Which issue tracker MCP to use (linear vs jira)
+- Which browser automation tool (playwright vs puppeteer)
+- Which CI/CD CLI to use for a pipeline trigger
+
+**Behavior**: Pick the first available tool in priority order. Briefly note the choice to the user in a single parenthetical or line.
+
+```markdown
+Using Playwright for browser automation (detected in PATH; preferred over Puppeteer and Cypress).
+```
+
+---
+
+### High Stakes — Always Ask User
+
+Decisions that affect production systems, cost money, or are difficult to reverse.
+
+**Examples**:
+- Which cloud provider for a deployment (aws vs gcloud vs az)
+- Which observability platform for metric ingestion
+- Which database for persistent data storage
+
+**Behavior**: List available options with one-line descriptions. Wait for user confirmation before proceeding.
+
+```markdown
+Multiple cloud CLIs detected. Which should I use for this deployment?
+
+1. aws — Amazon Web Services (IAM configured)
+2. gcloud — Google Cloud SDK (project: my-project)
+3. fly — Fly.io (authenticated)
+
+Please choose (number or name):
+```
+
+---
+
+## Storing Preferences in wicked-brain
+
+After the first tool selection decision, store the preference so future decisions are consistent.
+
+### Store Pattern
+
+```
+Skill(skill="wicked-brain:memory", args="store \"cli-preference:{category}: {chosen-tool} (project: {project-path})\" --type preference")
+```
+
+**Examples**:
+
+```
+cli-preference:browser → playwright
+cli-preference:cloud → aws
+cli-preference:package-manager → uv
+cli-preference:database → psql
+```
+
+### Recall Pattern
+
+Before detecting CLIs for a category, check if a preference already exists:
+
+```
+Skill(skill="wicked-brain:memory", args="recall \"cli-preference:{category}\" --filter_type preference")
+```
+
+If a stored preference exists and the tool is still available (`command -v` check), use it without re-running discovery.
+
+---
+
+## Usage Pattern for Other Domains
+
+Other domains and skills call integration-discovery for CLI selection rather than implementing ad-hoc detection themselves.
+
+**Detection snippet** (copy into any skill or command):
+
+```bash
+# Pick first available browser automation CLI
+for tool in playwright puppeteer cypress selenium-webdriver; do
+  command -v "$tool" > /dev/null 2>&1 && BROWSER_CLI="$tool" && break
+done
+# If not found, recommend: npm i -D playwright && npx playwright install
+```
+
+**Output format** — include in your response when a CLI selection is made:
+
+```
+CLI Detection: playwright found (medium stake — informing user of choice)
+Recommendation: npx playwright test
+Install fallback: npm i -D playwright && npx playwright install
+```
+

package/skills/integration-discovery/refs/discovery-sources.md

@@ -0,0 +1,233 @@
+---
+phase_relevance: ["bootstrap"]
+archetype_relevance: ["*"]
+---
+# Discovery Sources
+
+Where to look for capabilities and how to check each source.
+
+## Source Overview
+
+```
+┌──────────────────────────────────────────────────────────────────────────┐
+│                         Capability Sources                                │
+├─────────────────┬──────────────┬─────────────────┬────────────────────── ┤
+│  MCP Servers    │  CLI Tools   │  Skills         │  Agents               │
+│  (external)     │  (PATH)      │  (methodology)  │  (specialized workers)│
+├─────────────────┼──────────────┼─────────────────┼───────────────────────┤
+│  ListMcpRes...  │ command -v   │  Skill tool     │  Task tool            │
+│  Tool           │ {tool}       │  description    │  agent list           │
+└─────────────────┴──────────────┴─────────────────┴───────────────────────┘
+```
+
+## MCP Servers
+
+**What they provide**: External integrations (Jira, GitHub, Slack, analytics, etc.)
+
+**How to discover**:
+```
+ListMcpResourcesTool
+```
+
+**What to look for**:
+- Server names (often match the service: `atlassian`, `github`, `slack`)
+- Resource descriptions
+- Available tools/operations
+
+**Key insight**: MCP servers are user-configured. Don't assume availability—always check.
+
+### MCP Discovery Pattern
+
+```markdown
+## MCP Discovery
+
+Available servers:
+- `atlassian` - Jira/Confluence (project management, documentation)
+- `context7` - Library documentation retrieval
+
+Matched to needs:
+- Project tracking → atlassian (Jira)
+- Code docs → context7
+- Analytics → NOT AVAILABLE
+```
+
+## PATH / CLI Tools
+
+**What they provide**: Installed binaries for browser automation, cloud platforms, databases, CI/CD, package management, and AI models.
+
+**How to discover**:
+```bash
+command -v {tool}   # returns exit 0 if found, non-zero if not
+```
+
+**Key insight**: CLI tools are environment-specific. Don't assume availability — always check with `command -v`.
+
+**Detection categories**:
+
+| Category | Key Tools |
+|----------|-----------|
+| AI CLIs | claude, codex, copilot, gemini, opencode, pi |
+| Browser | playwright, puppeteer, cypress |
+| Cloud | aws, gcloud, az, heroku, vercel, fly |
+| Observability | datadog-agent, newrelic, dynatrace |
+| Data | duckdb, psql, mysql, mongosh, redis-cli |
+| CI/CD | gh, glab, circleci |
+| Package managers | uv, bun, npm, pip, cargo, go |
+
+### CLI Discovery Pattern
+
+```markdown
+## CLI Discovery
+
+Task requires browser automation.
+
+Detection (priority order):
+- playwright: FOUND (/usr/local/bin/playwright)
+- puppeteer: not checked (playwright found first)
+- cypress: not checked
+
+Recommendation: Use playwright.
+Stake level: medium — informing user of choice.
+```
+
+**Decision policy**: Low stakes → auto-decide silently. Medium stakes → auto-decide and inform. High stakes → always ask user. See [cli-detection.md](cli-detection.md) for full policy.
+
+**Preference storage**: After first selection, store in wicked-brain:memory under `cli-preference:{category}`.
+
+---
+
+## Skills
+
+**What they provide**: Methodology, expertise, guidance, processes
+
+**How to discover**: Check the Skill tool's available skills list in system prompt, or invoke skills by name.
+
+**Key skill families** (wicked-garden):
+
+| Family | Purpose | Example Skills |
+|--------|---------|----------------|
+| **engineering** | Engineering & architecture | solution-architect, api-documentarian |
+| **qe** | Quality engineering | semantic-reviewer, test-strategist |
+| **platform** | Security/compliance | security-engineer, privacy-expert |
+| **product** | Product management | requirements-analyst, ux-designer |
+| **data** | Data engineering | data-engineer |
+| **wicked-crew** | Workflow orchestration | implementer, researcher, reviewer |
+
+### Skill Discovery Pattern
+
+```markdown
+## Skill Discovery
+
+Task involves security review.
+
+Relevant skills:
+- platform (security-engineer, privacy-expert)
+- qe (test-strategist for security test scenarios)
+- engineering (review skill inline / crew:reviewer for code quality)
+
+Recommendation: Use wicked-garden:platform:security-engineer as primary reviewer.
+```
+
+## Agents
+
+**What they provide**: Specialized subagents that can work autonomously
+
+**How to discover**: Check Task tool's `subagent_type` options in system prompt.
+
+**Agent categories**:
+
+| Category | Agents | Best For |
+|----------|--------|----------|
+| **Explore** | Codebase exploration | Finding files, understanding structure |
+| **Plan** | Implementation planning | Designing approaches |
+| **Bash** | Command execution | Git, npm, system commands |
+| **wicked-*:*** | Specialized domain agents | Specific expertise |
+
+### Agent Discovery Pattern
+
+```markdown
+## Agent Discovery
+
+Task: "Review PR for security issues"
+
+Relevant agents:
+- `wicked-garden:platform:security-engineer` - Security expertise
+- `wicked-garden:crew:reviewer` - Code quality review
+- `Explore` - Understand codebase context
+
+Recommendation: Spawn security-engineer for focused review.
+```
+
+## Built-in Tools
+
+**What they provide**: Core operations always available
+
+**Always available**:
+
+| Tool | Purpose |
+|------|---------|
+| Read, Write, Edit | File operations |
+| Bash | Command execution |
+| Glob, Grep | File/content search |
+| WebFetch, WebSearch | Web access |
+| Task | Spawn subagents |
+| Skill | Invoke skills |
+
+**Don't need to discover these**—they're always present.
+
+## Discovery Checklist
+
+When scouting for a task:
+
+```markdown
+## Discovery Checklist
+
+Task: [description]
+
+### 1. MCP Servers
+- [ ] Run ListMcpResourcesTool
+- [ ] Match servers to task needs
+- [ ] Note what's missing
+
+### 2. CLI Tools
+- [ ] Check wicked-brain:memory for stored cli-preference:{category}
+- [ ] If no preference, run `command -v` on priority-ordered tool list
+- [ ] Apply auto-decide vs ask policy based on stake level
+- [ ] Store selection in wicked-brain:memory after first decision
+
+### 3. Skills
+- [ ] Identify relevant skill families
+- [ ] Check if installed/available
+- [ ] Note methodology gaps
+
+### 4. Agents
+- [ ] Identify specialized agents needed
+- [ ] Check availability
+- [ ] Plan orchestration
+
+### 5. Built-in Tools
+- [ ] Confirm required tools (usually all available)
+```
+
+## Discovery vs. Assumption
+
+| Approach | Risk | Better |
+|----------|------|--------|
+| Assume Jira available | Fails if not configured | Check MCP first |
+| Use generic agent | Misses specialized help | Check for domain agents |
+| Skip skill check | Miss methodology guidance | Check relevant skills |
+| Hardcode tool names | Breaks with different setup | Discover by capability |
+
+## Caching Discovery Results
+
+For long-running tasks, cache discovery at the start:
+
+```markdown
+## Session Capabilities (discovered at start)
+
+**MCP**: atlassian, github, context7
+**Skills**: wicked-* family installed
+**Agents**: Full wicked-garden specialist set
+
+Reference this throughout session instead of re-discovering.
+```