Agent Catalog¶

Overview of all available Claude Code agents in the Kamerplanter project for autonomous feature implementation, requirements engineering, code reviews and documentation.

Quick Reference¶

Agents by Category¶

Agent Details¶

agent-catalog-generator¶

Model: haiku | Tools: Read, Write, Glob, Grep

Role: Technical writer who systematically reads all agent definitions and generates a compact, developer-friendly catalog.

Workflow: 1. Read all agent definitions from .claude/agents/*.md 2. Parse YAML frontmatter and Markdown body 3. Group and sort agents by category 4. Generate documentation catalog with tables, tabs and details 5. Insert RAG-optimized metadata

Output: docs/de/development/agent-catalog.md — This catalog with overview, tables, categorization and decision guidance

agrobiology-requirements-reviewer¶

Model: sonnet | Tools: Read, Write, Glob, Grep

Role: Agrobiology expert with 20+ years of practical experience in indoor growing, hydroponics, houseplants and protected cultivation, who checks requirements for biological correctness and completeness.

Workflow: 1. Read and classify requirement documents (indoor/hydroponics/outdoor) 2. Check biological correctness (light, temperature, VPD, substrates) 3. Work through completeness checklists (houseplants, hydroponics, IPM) 4. Evaluate data source availability 5. Create report with findings (errors/incomplete/inaccurate)

Output: spec/requirements-analysis/agrobiology-review.md — Detailed report with botanical findings, correction proposals and data sources

cannabis-indoor-grower-reviewer¶

Model: sonnet | Tools: Read, Write, Glob, Grep

Role: Cannabis cultivation specialist who reviews specifications for legality, feasibility and best practice compliance.

Workflow: 1. Identify all cannabis-specific requirements 2. Check legal basis (CanG licensing requirements, PflSchG pre-harvest intervals) 3. Evaluate genetic data, phenotypes, terpenes 4. Validate growing parameters (photoperiod, VPD, EC) 5. Report with legal checks and best-practice deviations

Output: spec/requirements-analysis/cannabis-grower-review.md — Legality check, growing parameter validation, risk assessment

casual-houseplant-user-reviewer¶

Model: sonnet | Tools: Read, Write, Glob, Grep

Role: Disorganized houseplant owner without a green thumb, who checks requirements for everyday usability and beginner friendliness.

Workflow: 1. Read all requirements from a layperson's perspective 2. Identify dealbreaker features (e.g. photo recognition) 3. Mark frustrating and overwhelming aspects 4. Estimate effort per week 5. Conduct competitor comparison (Planta, Greg) 6. Report with dealbreakers, optimizations, effort analysis

Output: spec/requirements-analysis/casual-houseplant-user-review.md — Layperson perspective, dealbreakers, effort analysis, competitor comparison

code-security-reviewer¶

Model: sonnet | Tools: Read, Edit, Bash, Glob, Grep

Role: Application security engineer who checks implemented backend and frontend code for OWASP Top 10 vulnerabilities and fixes them.

Workflow: 1. Analyze backend and frontend code (discovery) 2. Systematically check OWASP A01 through A10 3. Validate tenant isolation, RBAC, JWT, secrets 4. Immediately fix critical vulnerabilities (P0/P1) 5. Security report with resolved and open items

Output: spec/requirements-analysis/code-security-review.md — Security audit, P0/P1/P2/P3 findings, compliance matrix

e2e-testcase-extractor¶

Model: sonnet | Tools: Read, Write, Glob, Grep

Role: Elite QA architect who systematically derives E2E test cases from specifications from the end-user perspective (browser view).

Workflow: 1. Read specifications and decompose into testable scenarios 2. Derive test cases for each UI element (happy path + errors) 3. Structure test cases: preconditions → test steps → expected results 4. Add RAG metadata (frontmatter, tags) 5. Generate structured test case files

Output: spec/test-cases/TC-{REQ-ID}.md — RAG-optimized test case documents with traceability

frontend-design-reviewer¶

Model: sonnet | Tools: Read, Write, Glob, Grep

Role: Frontend designer with 15+ years of experience in responsive design, kiosk systems and touch interfaces for demanding work environments.

Workflow: 1. Classify all requirements by operating context 2. Check responsive design (breakpoints, fluid grids) 3. Evaluate kiosk mode (64–72px touch targets, simplification) 4. Validate mobile on-site scenarios 5. Report with design findings and wireframe proposals

Output: spec/requirements-analysis/frontend-design-review.md — Responsive matrix, kiosk detail assessment, touch target audit, wireframes

frontend-usability-optimizer¶

Model: sonnet | Tools: Read, Write, Edit, Bash, Glob, Grep

Role: UX engineer who optimizes React/MUI components according to mobile-first principles for better usability (labels, help texts, validation, layout).

Workflow: 1. Read code and identify usability problems 2. Add descriptive texts and help texts 3. Optimize field ordering and validation 4. Implement mobile-first progressive enhancement 5. Perform UI-NFR compliance check 6. Run tests and output summary

Output: Optimized code in src/frontend/ + summary with usability improvements and compliance checks

fullstack-developer¶

Model: opus | Tools: Read, Write, Edit, Bash, Glob, Grep

Role: Senior full-stack developer who fully implements features (backend FastAPI + frontend React) in strict compliance with all NFRs.

Workflow: 1. Read requirements specifications 2. Backend: ArangoDB models, APIs, services, tests 3. Frontend: React components, Redux slices, tests 4. Comply with all NFRs and UI-NFRs 5. Write and validate unit tests

Output: Production-ready implementation in backend + frontend with tests; the unit-test-runner can pass green

gemini-graphic-prompt-generator¶

Model: sonnet | Tools: Read, Write, Glob, Grep

Role: Visual design director who generates precise, production-ready Google Gemini prompts for icons, illustrations and marketing materials.

Workflow: 1. Analyze design requirements 2. Apply Kamerplanter corporate design (colors, style) 3. Write Gemini prompts with design details 4. Specify light/dark mode variants 5. Define quality criteria (transparency, resolution)

Output: Gemini image generation prompts for designers/marketing — e.g. for icons, illustrations

growing-phase-auditor¶

Model: sonnet | Tools: Read, Write, Edit, Glob, Grep, Bash, WebSearch, WebFetch

Role: Horticultural scientist who checks growth phase data (bloom_months, harvest_months, etc.) in seed YAML for biological correctness and chronological consistency.

Workflow: 1. Read all seed YAML files 2. Validate against 5 check rules (gaps, biology, plausibility) 3. Correction proposals with WebSearch verification 4. Correct YAML files directly with the Edit tool 5. Create report with findings and corrections

Output: Corrected YAML files + spec/requirements-analysis/growing-phase-audit.md with findings and verification

ha-integration-requirements-engineer¶

Model: sonnet | Tools: Read, Write, Glob, Grep

Role: Home Assistant specialist with 8+ years of HACS integration development, who derives concrete HA integration requirements from REQ documents.

Workflow: 1. Read REQ documents 2. Apply three-sided model (export A, import B, control C) 3. Define entity taxonomies 4. Design coordinator data structures 5. Specify API requirements

Output: spec/ha-integration/HA-CUSTOM-INTEGRATION.md — Entity mappings, coordinators, services, events

ha-integration-sync¶

Model: opus | Tools: Read, Write, Edit, Bash, Glob, Grep

Role: Home Assistant integration developer who synchronizes the kamerplanter-ha custom integration with the current backend API without altering existing domain logic.

Workflow: 1. Capture backend API endpoints 2. Analyze HA API client (api.py) 3. Perform delta analysis (new/changed endpoints) 4. Adapt coordinator, sensor and service code 5. Deploy HA integration

Output: Updated HA integration files + deployment instructions

it-security-requirements-reviewer¶

Model: sonnet | Tools: Read, Write, Glob, Grep

Role: IT security architect with 15+ years of experience who checks requirements for security, data protection and GDPR compliance.

Workflow: 1. Read all requirements 2. Create security index (data, access, interfaces) 3. Validate data protection requirements (GDPR Art. 15–21) 4. Reconcile OWASP ASVS against specs 5. Report with security gaps and GDPR recommendations

Output: spec/requirements-analysis/it-security-review.md — Security assessment, GDPR audit, recommendations

mkdocs-documentation¶

Model: sonnet | Tools: Read, Write, Edit, Bash, Glob, Grep

Role: Technical writer and documentation engineer who creates and maintains user-friendly, multilingual documentation in MkDocs Material format.

Workflow: 1. Write documentation in MkDocs Material format 2. German and English in parallel (i18n) 3. Use Mermaid diagrams for visualizations 4. Use mkdocstrings for API docs from docstrings 5. Test with local mkdocs preview

Output: docs/de/ and docs/en/ — User and admin documentation

outdoor-garden-planner-reviewer¶

Model: sonnet | Tools: Read, Write, Glob, Grep

Role: Ambitious hobby gardener with 15 years of outdoor gardening and an 80m² community garden plot, who checks requirements for practicality and everyday relevance.

Workflow: 1. Read requirements from a hobby gardener's perspective 2. Identify garden life workflows 3. Assess outdoor everyday practicality 4. Validate crop rotation, overwintering, companion planting features 5. Report with practical notes and improvement proposals

Output: spec/requirements-analysis/outdoor-garden-planner-review.md — Hobby gardener perspective, practical feedback

plant-info-document-generator¶

Model: sonnet | Tools: Read, Write, Glob, Grep, WebSearch, WebFetch

Role: Agricultural botanist with 20+ years of practical experience (nursery, indoor growing, allotment garden) who researches and documents detailed plant profile documents.

Workflow: 1. Analyze user input (plant name, list) 2. Research scientific names 3. Capture taxonomy, phases, nutrients, IPM, companion planting 4. Verify sources (RHS, USDA, DWD) 5. Output structured document for data import

Output: Detailed plant information documents for data import

png-to-transparent-svg¶

Model: haiku | Tools: Read, Write, Bash, Glob

Role: Image processing specialist who converts PNG images with checkerboard backgrounds into clean SVGs with true transparency.

Workflow: 1. Analyze PNG input (check alpha channel) 2. Detect and remove checkerboard pattern 3. Generate cleaned PNG with true alpha transparency 4. Vectorize PNG to SVG using vtracer 5. Save SVG file

Output: Transparent SVG files → assets/icons/ or user-defined target folder

pr-to-develop¶

Model: sonnet | Tools: Read, Bash, Glob, Grep

Role: Release engineer who prepares GitHub pull requests from feature branches to develop with CI validation and meaningful documentation.

Workflow: 1. Branch analysis (commits since develop) 2. Analyze changed files (backend/frontend/spec) 3. Identify REQ/NFR numbers 4. Create PR title and detailed description 5. Set labels and wait for CI

Output: GitHub pull request to develop with title, description, labels, CI status

requirements-contradiction-analyzer¶

Model: sonnet | Tools: Read, Write, Glob, Grep, Bash

Role: Requirements engineer who systematically analyzes requirement documents using RAG for contradictions between functional and non-functional requirements.

Workflow: 1. Read all requirement documents (RAG retrieval) 2. Classify requirements (FR/NFR) 3. Build requirements index 4. Systematically check for contradictions 5. Generate report with contradiction findings

Output: spec/requirements-analysis/contradiction-analysis.md — Contradictions, inconsistencies, recommendations

seed-data-validator¶

Model: sonnet | Tools: Read, Write, Glob, Grep, Bash, WebSearch, WebFetch

Role: Data quality engineer who checks YAML seed data for completeness, consistency, referential integrity and botanical plausibility.

Workflow: 1. Read YAML seed files 2. Check schema compliance 3. Validate referential integrity 4. Assess botanical plausibility (optionally with agrobiology reviewer) 5. Multi-source verification (manufacturer, SDS, retailer) 6. Report with data quality findings

Output: spec/requirements-analysis/seed-data-validation.md — Data quality report with findings

selenium-test-generator¶

Model: opus | Tools: Read, Write, Edit, Glob, Grep, Bash

Role: QA engineer and Selenium expert who generates NFR-008-compliant Python Selenium tests with page object pattern and screenshot checkpoints.

Workflow: 1. Read NFR-008 and test case documents 2. Analyze requirements (testable) 3. Create page objects 4. Generate test classes with pytest 5. Integrate screenshot and logging functionality 6. Validate tests locally

Output: tests/e2e/test_*.py — NFR-008-compliant Selenium tests with report generation

selenium-test-reviewer¶

Model: sonnet | Tools: Read, Write, Edit, Glob, Grep, Bash

Role: QA specialist who reviews existing Selenium tests for quality, maintainability, flakiness and NFR-008 compliance and optimizes them.

Workflow: 1. Analyze existing Selenium tests 2. Check against NFR-008 requirements 3. Identify flakiness problems 4. Optimize code quality and maintainability 5. Create report with recommendations

Output: Optimized Selenium tests + quality report

smart-home-ha-reviewer¶

Model: sonnet | Tools: Read, Write, Glob, Grep

Role: Home Assistant integration reviewer who checks the kamerplanter-ha custom integration against specification and best practices.

Workflow: 1. Analyze HA integration code 2. Check against HA-CUSTOM-INTEGRATION.md 3. Validate entity taxonomies 4. Review coordinator data flow 5. Report with compliance and optimizations

Output: spec/ha-integration/HA-REVIEW.md — HA integration review with findings

target-audience-analyzer¶

Model: sonnet | Tools: Read, Write, Glob, Grep

Role: Market and UX analyst who systematically analyzes and documents the target audiences, personas and market potential of the application.

Workflow: 1. Analyze requirements and contexts 2. Identify potential target audiences 3. Create personas with goals, pain points, motivation 4. Estimate market size and potential 5. Report with personas and market analysis

Output: spec/requirements-analysis/target-audience-analysis.md — Target audience analysis, personas, market potential

tech-stack-architect¶

Model: sonnet | Tools: Read, Write, Glob, Grep

Role: Tech architect who validates the defined tech stack (FastAPI, React, ArangoDB, Redis, Kubernetes) against requirements and justifies technology decisions.

Workflow: 1. Read requirements and NFRs 2. Define tech stack (backend, frontend, databases, infra) 3. Reconcile requirements against stack 4. Evaluate alternative technologies 5. Report with validation and justifications

Output: spec/stack.md or technology decision record — Tech stack validation with justifications

unit-test-runner¶

Model: sonnet | Tools: Read, Edit, Bash, Glob, Grep

Role: QA engineer and test specialist who runs unit tests (pytest, vitest) and static analysis (Ruff, ESLint, TypeScript), analyzes errors and proposes fixes.

Workflow: 1. Run backend Ruff (lint + format) 2. Run frontend TypeScript and ESLint 3. Run backend unit tests (pytest) 4. Run frontend unit tests (vitest) 5. Analyze test failures and implement minimal fixes 6. Perform regression check 7. Report with test results and merge status

Output: Test results + test report — green tests or failure analysis with fixes

Decision Guide¶

Notes for Developers¶

• Start an agent: /agent <agent-name> in the Claude Code chat
• Reports: Analysis agents write to spec/requirements-analysis/, Selenium reports to test-reports/, test cases to spec/test-cases/, documentation to docs/
• Model choice: opus = highest quality (complex features), sonnet = best value (review, analysis), haiku = fast & cheap (simple tasks)
• Tool availability: Not all agents have all tools — e.g. png-to-transparent-svg has no Edit tool

Agent Dependencies and Workflow Suggestions¶

graph LR
    A["Specification<br/>(REQ/NFR)"] --> B["tech-stack-architect<br/>requirements-contradiction-analyzer"]
    B --> C["agrobiology-requirements-reviewer<br/>cannabis-indoor-grower-reviewer<br/>casual-houseplant-user-reviewer<br/>it-security-requirements-reviewer"]
    C --> D["e2e-testcase-extractor"]
    D --> E["fullstack-developer"]
    E --> F["unit-test-runner"]
    F --> G["code-security-reviewer"]
    G --> H["pr-to-develop"]
    H --> I["GitHub Actions / CI"]

    E -.-> J["frontend-design-reviewer<br/>frontend-usability-optimizer"]
    E -.-> K["ha-integration-sync"]
    E -.-> L["growing-phase-auditor"]
    L -.-> M["seed-data-validator"]

    D -.-> N["selenium-test-generator"]
    N -.-> O["selenium-test-reviewer"]
    O -.-> P["unit-test-runner"]

    A -.-> Q["ha-integration-requirements-engineer"]
    Q -.-> K
    K -.-> R["smart-home-ha-reviewer"]

    A -.-> S["mkdocs-documentation"]
    A -.-> T["plant-info-document-generator"]
    A -.-> U["target-audience-analyzer<br/>outdoor-garden-planner-reviewer"]

    V["Icon/graphics requirements"] -.-> W["gemini-graphic-prompt-generator"]
    W -.-> X["png-to-transparent-svg"]

Legend: - Solid arrows: Standard workflow (spec → implementation → QA → PR) - Dotted arrows: Parallel or optional workflows (design, HA, documentation, graphics)

Category Overview¶

Analysis & Review (11 agents): Evaluate requirements and code against quality criteria. No implementation.

Development (4 agents): Write or optimize productive code and integration code.

Testing & QA (6 agents): Generate, validate or run tests. Ensure quality.

Design & Graphics (3 agents): UI/UX optimization and asset generation.

Documentation (3 agents): Create and maintain documentation and plant profiles.

Catalog updated: 2026-03-23 Agents documented: 27 Generated by: agent-catalog-generator