# Examples ## Full Lifecycle: Build a Feature This example walks through the complete specification-driven workflow for adding a search feature. ### 1. Start the workflow ``` /acli.run Add full-text search to the product catalog with filters and pagination ``` The orchestrator guides you through all 10 phases automatically. At each human gate, you review the artifacts and approve or request revisions. ### 2. What gets created ``` .specify/ memory/ constitution.md reference-architecture.md quality-standards.md specs/ 001-product-search/ spec.md plan.md tasks.md analysis.md checklists/ security.md accessibility.md performance.md ``` ### 3. Implementation loop For each task, the orchestrator runs: 1. `@development` implements the task 2. `@qa` reviews via `/acli.critique` 3. `@qa` runs tests 4. If issues found: `/acli.respond` to fix, then re-review 5. If clean: mark done, next task --- ## Multi-Platform Setup Target all editors at once: ```bash acli init --platforms copilot cursor claude windsurf open-plugins ``` After init, every `acli install` emits to all platforms: ``` .github/agents/architect.agent.md ← Copilot .cursor/rules/architect.mdc ← Cursor .windsurf/rules/architect.md ← Windsurf .agents/plugins/agent-framework/agents/ ← Open Plugins AGENTS.md ← Claude Code (aggregated) ``` To add a new platform later without reinitialising: ```bash # Edit .agent-framework.json — add "cursor" to "platforms" array acli update # re-emits all agents to new + existing platforms ``` --- ## Installing a Community Agent The Agency roster provides 140+ specialist agents across engineering, design, marketing, and more. ```bash # See what’s available acli agency list acli agency list --division engineering # Find a specific type acli agency search "data engineer" acli agency search "security" # Install by full slug acli agency install engineering-senior-developer # Install by keyword (installs if there’s exactly one match) acli agency install senior-developer # Scope to a division to disambiguate acli agency install developer --division engineering ``` On a project targeting Copilot + Claude, the command above writes: ``` .github/agents/engineering-senior-developer.agent.md .claude/agents/engineering-senior-developer.md ``` --- ## Custom Plugin: Domain-Specific Agent Build a plugin for a specific domain (e.g., database migrations): ```bash acli extensions create db-migrations \ --description "Agent and skill for database migration workflows" \ --author "Your Name" \ --with-agent \ --with-skill ``` This scaffolds: ``` .agents/plugins/db-migrations/ .plugin/plugin.json agents/db-migrations.agent.md ← edit with your migration expertise skills/db-migrations/SKILL.md ← edit with step-by-step guidance rules/ ← add platform rules here ``` Edit the agent file and add domain knowledge, then share it: ```bash acli extensions pack db-migrations # → db-migrations-v1.0.0.zip # Team member installs it: acli extensions add db-migrations --from https://yourhost.com/db-migrations-v1.0.0.zip ``` --- ## Onboarding an Existing Project ``` /acli.onboard ``` The brownfield onboarding scans your codebase and produces: - Constitution with principles derived from your actual code - Reference architecture mapping your real directory structure - Quality standards extracted from your linter/test config --- ## Debugging a Bug ``` /acli.debug Users get a 500 error when uploading files larger than 10MB ``` The debug workflow: 1. Captures the symptom precisely 2. Forms ranked hypotheses 3. Tests each systematically 4. Implements minimal fix with test coverage 5. Suggests prevention measures --- ## Code Review ``` /acli.critique src/services/payment.ts src/services/auth.ts ``` Produces a structured review checking: - Spec alignment (does it match what was specified?) - Plan conformance (correct architecture layer?) - Constitution compliance (follows project principles?) - Security (OWASP Top 10) - Test coverage --- ## Branch Completion ``` /acli.finish squash ``` Runs pre-completion checklist, cleans up debug artifacts, drafts conventional commit messages, and recommends merge strategy. --- ## Creating a Custom Agent ``` /acli.create.agent ``` Interactive prompts guide you through: - Agent name and description - Tools and agent references - Handoff configuration - Instructions template The result is a `.agent.md` file emitted to all configured platforms. --- ## Using Individual Phases ### Constitution ``` /acli.constitution ``` Creates the project constitution — the single source of truth for how the project is built. ### Specification ``` /acli.specify Add OAuth2 login with Google and GitHub providers ``` Creates a technology-agnostic spec with user stories, acceptance criteria, and P0–P3 priorities. ### Clarification ``` /acli.clarify ``` Identifies ambiguities in the spec and asks targeted questions (max 5) with suggested answers. ### Planning ``` /acli.plan ``` Creates the technical plan with file structure, ADRs, and testing strategy. ### Quality Checklists ``` /acli.checklist all /acli.checklist security ``` Generates domain-specific checklists that become gates during implementation. ### Task Breakdown ``` /acli.tasks ``` Creates a dependency-ordered task list with test requirements for each task. ### Consistency Analysis ``` /acli.analyze ``` Cross-references spec, plan, and tasks to find gaps and conflicts. ### Implementation ``` /acli.implement /acli.implement start from TASK-005 ``` Implements tasks one at a time with the Dev -> Critique -> Test review loop. ### 2. What gets created ``` .specify/ memory/ constitution.md -- project principles (created once) reference-architecture.md -- architecture reference quality-standards.md -- quality rules specs/ 001-product-search/ spec.md -- feature specification plan.md -- technical plan tasks.md -- task list analysis.md -- consistency analysis checklists/ security.md accessibility.md performance.md ``` ### 3. Implementation loop For each task, the orchestrator runs: 1. `@development` implements the task 2. `@qa` reviews via `/acli.critique` 3. `@qa` runs tests 4. If issues found: `/acli.respond` to fix, then re-review 5. If clean: mark done, next task ## Onboarding an Existing Project ``` /acli.onboard ``` The brownfield onboarding scans your codebase and produces: - Constitution with principles derived from your actual code - Reference architecture mapping your real directory structure - Quality standards extracted from your linter/test config ## Debugging a Bug ``` /acli.debug Users get a 500 error when uploading files larger than 10MB ``` The debug workflow: 1. Captures the symptom precisely 2. Forms ranked hypotheses 3. Tests each systematically 4. Implements minimal fix with test coverage 5. Suggests prevention measures ## Code Review ``` /acli.critique src/services/payment.ts src/services/auth.ts ``` Produces a structured review checking: - Spec alignment (does it match what was specified?) - Plan conformance (correct architecture layer?) - Constitution compliance (follows project principles?) - Security (OWASP Top 10 checks) - Test coverage ## Branch Completion ``` /acli.finish squash ``` Runs pre-completion checklist, cleans up debug artifacts, drafts conventional commit messages, and recommends merge strategy. ## Creating a Custom Agent ``` /acli.create.agent ``` Interactive prompts guide you through: - Agent name and description - Platform target (VS Code or GitHub Copilot) - Tools and agent references - Handoff configuration - Instructions template The result is a `.agent.md` file in `.github/agents/`. ## Using Individual Phases ### Constitution (project setup) ``` /acli.constitution ``` Creates the project constitution -- the single source of truth for how the project is built. All agents respect it. ### Specification (what to build) ``` /acli.specify Add OAuth2 login with Google and GitHub providers ``` Creates a technology-agnostic spec with user stories, acceptance criteria, and P0-P3 priorities. ### Clarification (resolve unknowns) ``` /acli.clarify ``` Identifies ambiguities in the spec and asks targeted questions (max 5) with suggested answers. ### Planning (how to build) ``` /acli.plan ``` Creates the technical plan with file structure, ADRs, and testing strategy. ### Quality Checklists ``` /acli.checklist all /acli.checklist security ``` Generates domain-specific checklists that become gates during implementation. ### Task Breakdown ``` /acli.tasks ``` Creates a dependency-ordered task list with test requirements for each task. ### Consistency Analysis ``` /acli.analyze ``` Cross-references spec, plan, and tasks to find gaps and conflicts. ### Implementation ``` /acli.implement /acli.implement start from TASK-005 ``` Implements tasks one at a time with the Dev -> Critique -> Test review loop.