Build, test, and review code with purpose-built development skills
Version: v3.12.0 | Archetype: Developer | Skills: 3 specialized + 15 universal
The Developer archetype brings software engineering expertise to your AI coding workflow. Unlike generic coding assistants, developer agents understand:
- Test-driven development — Run tests, interpret results, suggest fixes
- Code quality standards — Lint code with project-aware configuration
- Review discipline — Structured PR reviews with categorized feedback
Domain knowledge that compounds: Developer agents build persistent expertise in your codebase — project conventions, common failure patterns, test strategies, and architectural decisions. Unlike tools that start fresh each session, your agent accumulates context that makes each code review sharper, each test run more targeted, and each fix more informed.
For evaluators: If your team needs an AI that understands the full development lifecycle — not just code generation — the Developer archetype provides specialized skills for building, testing, and maintaining quality code.
Developer agents come with 3 archetype-specific skills plus the universal AGET skills.
| Skill | Description |
|---|---|
| aget-run-tests | Execute test suites (pytest, jest, go test) with auto-detection. Reports pass/fail counts, captures failures with locations, suggests next steps. |
| aget-lint-code | Run linting with project configuration (ruff, eslint, golangci-lint). Categorizes issues by severity, identifies auto-fixable problems. |
| aget-review-pr | Review pull requests for quality and standards. Produces structured feedback: blocking issues, suggestions, nitpicks, and praise. |
All AGET agents include session management, knowledge capture, and health monitoring:
aget-wake-up/aget-wind-down— Session lifecycleaget-create-project/aget-review-project— Project managementaget-record-lesson/aget-record-observation— Learning captureaget-check-health/aget-check-kb/aget-check-evolution— Health monitoringaget-propose-skill/aget-create-skill— Skill developmentaget-save-state/aget-file-issue— State and issue management
Developer agents use a formal vocabulary of 10 concepts organized into 4 clusters:
| Cluster | Concepts |
|---|---|
| Code Management | Code, Code_Change |
| Testing | Test, Unit_Test, Integration_Test, Test_Result |
| Build Management | Build, Dependency |
| Code Review | Pull_Request, Code_Review |
This vocabulary enables precise communication about development artifacts and processes.
See: ontology/ONTOLOGY_developer.yaml
# 1. Clone the template
git clone https://github.com/aget-framework/template-developer-aget.git my-dev-agent
cd my-dev-agent
# 2. Configure identity
# Edit .aget/version.json:
# "agent_name": "my-dev-agent"
# "domain": "your-domain"
# 3. Verify setup
python3 -m pytest tests/ -v
# Expected: All tests passing# In Claude Code CLI
/aget-run-tests # Execute project tests
/aget-lint-code # Check code quality
/aget-review-pr 42 # Review PR #42| Aspect | Generic Coding AI | Developer Agent |
|---|---|---|
| Test execution | Basic pytest run | Framework detection, failure analysis, suggested fixes |
| Linting | Single tool | Project-aware configuration precedence |
| PR review | Line-by-line comments | Structured review with blocking/suggestion/nitpick categories |
| Vocabulary | Informal | Formal ontology (Code_Change, Test_Result, Pull_Request) |
| Domain memory | Starts fresh each session | Accumulates codebase expertise over time |
| Directory | Purpose | Owner |
|---|---|---|
.claude/skills/ |
Slash command definitions | Framework + Agent |
.claude/agents/ |
Subagent definitions | Agent |
.claude/rules/ |
Path-scoped context rules | Agent |
Skills are provided by the template. Agents and rules directories are scaffolded for your customization.
| Attribute | Value |
|---|---|
| Framework | AGET v3.12.0 |
| Archetype | Developer |
| Skills | 18 total (3 archetype + 15 universal) |
| Ontology | 10 concepts, 4 clusters |
| License | Apache 2.0 |
- AGET Framework — Core framework documentation
- Archetype Guide — All 12 archetypes explained
- Getting Started — Full onboarding guide
| Archetype | Best For |
|---|---|
| Worker | Task execution with progress reporting |
| Reviewer | Quality assurance and feedback |
| Architect | System design and tradeoff analysis |
AGET Framework | Apache 2.0 | Issues