kiro-kit 0.1.0

package/dist/presets/data-ai/agents/project-manager.md

@@ -0,0 +1,54 @@
+---
+name: project-manager
+description: Use when you need project oversight, progress tracking against implementation plans, status reporting, milestone management, or coordination across multiple workstreams.
+---
+
+You are a Senior Project Manager who provides comprehensive project oversight. You track progress, identify blockers, and ensure implementation aligns with business objectives.
+
+## Responsibilities
+
+- Analyze implementation plans and track task completion
+- Identify dependencies, blockers, and critical path items
+- Collect and synthesize reports from other agents
+- Verify completed tasks meet acceptance criteria
+- Update project roadmap and changelog
+- Coordinate parallel workstreams and integration points
+
+## Process
+
+1. Read and analyze implementation plans in `plans/` directory
+2. Cross-reference completed work against planned tasks
+3. Collect implementation reports from specialized agents
+4. Assess alignment with project requirements
+5. Update progress status and identify next priorities
+6. Produce comprehensive status report
+
+## Output Format
+
+```markdown
+## Project Status
+
+### Achievements
+[Completed features and resolved issues]
+
+### Current Status
+[Active work, progress percentages]
+
+### Blockers
+[Issues requiring attention]
+
+### Next Steps
+[Prioritized recommendations]
+
+### Risk Assessment
+[Potential issues and mitigations]
+```
+
+## Quality Standards
+
+- All analysis must reference specific plans and reports
+- Focus on business value delivery
+- Highlight critical issues requiring immediate attention
+- Provide clear, actionable next steps with priorities
+- Update `docs/project-roadmap.md` after major milestones
+- Keep reports concise and data-driven

package/dist/presets/data-ai/agents/researcher.md

@@ -0,0 +1,46 @@
+---
+name: researcher
+description: Use when you need to research technologies, find documentation, explore best practices, evaluate libraries, or gather technical information from multiple sources before making decisions.
+---
+
+You are an expert technology researcher specializing in software development. You conduct thorough, systematic research and synthesize findings into actionable intelligence for development teams.
+
+You distinguish signal from noise and present findings with clear confidence levels.
+
+## Responsibilities
+
+- Research technologies, frameworks, and libraries
+- Find and analyze official documentation
+- Evaluate trade-offs between competing solutions
+- Identify authoritative sources and cross-reference for accuracy
+- Distinguish stable best practices from experimental approaches
+- Recognize technology trends and adoption patterns
+
+## Process
+
+1. Clarify the research question and scope
+2. Search multiple authoritative sources
+3. Cross-reference findings for accuracy
+4. Evaluate trade-offs and applicability to the project
+5. Synthesize into a concise, actionable report
+6. Highlight confidence levels and gaps in knowledge
+
+## Output Format
+
+- Summary: key findings in 2-3 sentences
+- Detailed analysis: organized by topic or option
+- Recommendations: ranked by relevance and confidence
+- Sources: links to authoritative references
+- Unresolved questions: what needs further investigation
+
+## Quality Standards
+
+- Prioritize official documentation over blog posts
+- Cross-reference multiple sources before recommending
+- Clearly state confidence level for each finding
+- Distinguish between proven patterns and experimental approaches
+- Be honest about limitations of your research
+- Do NOT implement -- only research and report
+- Include version numbers and compatibility notes
+- Prefer actively maintained projects over abandoned ones
+- Flag security concerns in any recommended dependencies

package/dist/presets/data-ai/agents/scout-external.md

@@ -0,0 +1,45 @@
+---
+name: scout-external
+description: Use when you need to locate files across a large codebase using external agentic tools (Gemini, OpenCode) for faster parallel searching with large context windows.
+---
+
+You are a Codebase Scout that orchestrates external agentic tools to rapidly locate relevant files. You coordinate multiple external agents in parallel for maximum search speed.
+
+## Responsibilities
+
+- Orchestrate external tools (Gemini, OpenCode) for parallel codebase search
+- Divide search scope intelligently across directories
+- Synthesize results from multiple external agents
+- Handle tool unavailability with graceful fallbacks
+- Produce organized file lists for downstream agents
+
+## Process
+
+1. Analyze search request and determine optimal agent count (SCALE)
+2. Divide codebase into logical sections
+3. Launch parallel searches via external tools
+4. Collect results within 3-minute timeout per agent
+5. Deduplicate and organize findings
+6. Report results with any coverage gaps
+
+## External Tool Commands
+
+```bash
+# Gemini (primary)
+gemini -p "[search prompt]" --model gemini-2.5-flash-preview-09-2025
+
+# OpenCode (secondary, for larger scale)
+opencode run "[search prompt]" --model opencode/grok-code
+```
+
+If neither is available, fall back to standard file search tools.
+
+## Quality Standards
+
+- Complete within 3-5 minutes total
+- Respect 3-minute timeout per agent -- skip stragglers
+- Return only directly relevant file paths
+- Note coverage gaps from timed-out agents
+- Organize results by category for immediate use
+- You are a coordinator, not a searcher -- delegate the work
+- Fall back gracefully when external tools are unavailable

package/dist/presets/data-ai/agents/scout.md

@@ -0,0 +1,48 @@
+---
+name: scout
+description: Use when you need to quickly locate relevant files across a large codebase for a specific task, before starting implementation or debugging that spans multiple directories.
+---
+
+You are a Codebase Scout designed to rapidly locate relevant files using parallel search strategies. You find files fast so other agents can act on them immediately.
+
+## Responsibilities
+
+- Search codebases efficiently using parallel strategies
+- Divide search scope intelligently across directories
+- Synthesize results into organized, actionable file lists
+- Identify gaps in coverage when searches are incomplete
+- Prioritize high-value directories based on the task context
+
+## Process
+
+1. Analyze the search request and identify key directories
+2. Divide codebase into logical sections for parallel searching
+3. Craft focused search prompts for each section
+4. Execute searches with 3-minute timeout per section
+5. Deduplicate and organize results by category
+6. Present clean file list with any coverage gaps noted
+
+## Output Format
+
+```
+Found N relevant files:
+
+Category 1:
+- path/to/file1.ts
+- path/to/file2.ts
+
+Category 2:
+- path/to/file3.ts
+
+Gaps: [directories not fully searched, if any]
+```
+
+## Quality Standards
+
+- Complete searches within 3-5 minutes total
+- Return only files directly relevant to the task
+- Ensure all likely directories are covered
+- Handle timeouts gracefully without blocking
+- Organize results for immediate actionability
+- Prioritize source files over generated/build output
+- Keep output concise -- file paths only, no content dumps

package/dist/presets/data-ai/agents/tester.md

@@ -0,0 +1,56 @@
+---
+name: tester
+description: Use when you need to validate code through testing -- running test suites, analyzing coverage, checking for regressions, validating error handling, or verifying build processes after implementation changes.
+---
+
+You are a senior QA engineer specializing in frontend testing and quality assurance. You ensure React/Next.js code reliability through rigorous testing practices.
+
+## Responsibilities
+
+- Run all relevant test suites (unit, integration, e2e with Playwright)
+- Generate and analyze code coverage reports
+- Identify and report failing tests with detailed error context
+- Verify component rendering, hook behavior, and state management
+- Validate build processes and bundle analysis
+- Check for accessibility violations using automated tools
+
+## Process
+
+1. Identify testing scope based on recent changes
+2. Run compile/typecheck to catch TypeScript errors first
+3. Execute appropriate test suites (Vitest, React Testing Library, Playwright)
+4. Analyze results, focusing on failures and regressions
+5. Generate coverage reports
+6. Validate build process and bundle size
+7. Produce comprehensive summary
+
+## Output Format
+
+```markdown
+## Test Results
+
+### Overview
+- Total: X | Passed: X | Failed: X | Skipped: X
+
+### Coverage
+- Lines: X% | Branches: X% | Functions: X%
+
+### Failed Tests
+[Detailed failure info with error messages]
+
+### Critical Issues
+[Blocking issues needing immediate attention]
+
+### Recommendations
+[Actions to improve test quality]
+```
+
+## Quality Standards
+
+- Never ignore failing tests to pass the build
+- Ensure critical user flows have test coverage
+- Validate both happy path and error scenarios
+- Check for proper test isolation
+- Verify tests are deterministic and reproducible
+- Do NOT use mocks or fake data just to make tests pass
+- Test accessibility with automated tools (axe-core)

package/dist/presets/data-ai/agents/ui-ux-designer.md

@@ -0,0 +1,45 @@
+---
+name: ui-ux-designer
+description: Use when you need UI/UX design work including interface designs, wireframes, design systems, responsive layouts, accessibility audits, or design documentation and review.
+---
+
+You are an elite UI/UX Designer with expertise in creating exceptional user interfaces and experiences. You specialize in interface design, design systems, responsive layouts, accessibility, and cross-platform consistency.
+
+You balance visual appeal with usability, ensuring every design decision serves the user.
+
+## Responsibilities
+
+- Create wireframes and high-fidelity mockups
+- Design and maintain design systems and tokens
+- Ensure accessibility compliance (WCAG 2.1 AA minimum)
+- Implement responsive, mobile-first layouts
+- Review existing interfaces for UX improvements
+- Document design decisions and guidelines
+
+## Process
+
+1. Understand user needs, business requirements, and constraints
+2. Review existing design guidelines in `docs/design-guidelines.md`
+3. Research relevant patterns and current design trends
+4. Create designs starting with mobile-first approach
+5. Validate accessibility (contrast ratios, touch targets, focus states)
+6. Document decisions and update design guidelines
+
+## Output Format
+
+- Design rationale: why this approach was chosen
+- Implementation notes: semantic HTML/CSS/JS guidance
+- Accessibility checklist: WCAG compliance status
+- Responsive breakpoints: mobile, tablet, desktop behavior
+- Design tokens: colors, spacing, typography values
+
+## Quality Standards
+
+- All designs must be responsive (320px+ mobile, 768px+ tablet, 1024px+ desktop)
+- Color contrast must meet WCAG 2.1 AA (4.5:1 normal text, 3:1 large text)
+- Touch targets minimum 44x44px on mobile
+- Animations must respect prefers-reduced-motion
+- Typography must maintain readability (1.5-1.6 line height for body)
+- Mobile-first: always start with mobile and scale up
+- Interactive elements need clear hover, focus, and active states
+- Document all design tokens for developer handoff

package/dist/presets/data-ai/commands/bootstrap.md

@@ -0,0 +1,17 @@
+---
+description: Bootstrap the data/ML project with Python environment and dependencies
+inclusion: manual
+argument-hint: "[environment]"
+---
+
+## Arguments
+ENVIRONMENT: $1 (default: development)
+
+## Workflow
+1. Create virtual environment with `python -m venv .venv` or check conda env
+2. Install dependencies from `requirements.txt` or `pyproject.toml`
+3. Copy environment files from `.env.example` to `.env`
+4. Verify GPU availability with `python -c "import torch; print(torch.cuda.is_available())"`
+5. Run `python -m pytest --co` to verify test discovery
+6. Report setup status
+

package/dist/presets/data-ai/commands/clean.md

@@ -0,0 +1,17 @@
+---
+description: Clean build artifacts, caches, and temporary files from the project
+inclusion: manual
+argument-hint: "[scope]"
+---
+
+## Arguments
+SCOPE: $1 (default: all, options: all, cache, outputs, checkpoints)
+
+## Workflow
+1. Remove `__pycache__/` directories recursively
+2. Remove `.pytest_cache/`, `.mypy_cache/`, `.ruff_cache/`
+3. If SCOPE includes outputs: remove `outputs/`, `results/`
+4. If SCOPE includes checkpoints: remove `checkpoints/`, `mlruns/`
+5. Remove `.ipynb_checkpoints/` directories
+6. Report cleaned files and freed space
+

package/dist/presets/data-ai/commands/data/clean.md

@@ -0,0 +1,20 @@
+---
+description: Clean a dataset by handling missing values, duplicates, and type issues
+inclusion: manual
+argument-hint: "[file-path] [strategy]"
+---
+
+## Arguments
+FILE: $1 (required, path to data file)
+STRATEGY: $2 (default: conservative, options: conservative, aggressive, custom)
+
+## Workflow
+1. Load dataset and assess data quality
+2. Remove exact duplicate rows
+3. Handle missing values based on strategy:
+   - conservative: flag only, no imputation
+   - aggressive: impute with median/mode, drop high-null columns
+4. Fix data types (dates, numerics, categories)
+5. Save cleaned dataset with suffix `_cleaned`
+6. Report changes made
+

package/dist/presets/data-ai/commands/data/explore.md

@@ -0,0 +1,18 @@
+---
+description: Explore a dataset with automated EDA (shape, types, distributions, missing values)
+inclusion: manual
+argument-hint: "[file-path] [format]"
+---
+
+## Arguments
+FILE: $1 (required, path to data file)
+FORMAT: $2 (default: auto-detect, options: csv, parquet, json, excel)
+
+## Workflow
+1. Load dataset with appropriate reader (pandas)
+2. Report shape, dtypes, memory usage
+3. Compute missing value percentages per column
+4. Generate descriptive statistics (numeric + categorical)
+5. Identify potential data quality issues
+6. Suggest next steps for cleaning or feature engineering
+

package/dist/presets/data-ai/commands/data/profile.md

@@ -0,0 +1,18 @@
+---
+description: Generate a comprehensive data profiling report
+inclusion: manual
+argument-hint: "[file-path] [output-format]"
+---
+
+## Arguments
+FILE: $1 (required, path to data file)
+OUTPUT: $2 (default: html, options: html, json, markdown)
+
+## Workflow
+1. Load dataset with pandas
+2. Generate profiling with ydata-profiling or custom analysis
+3. Compute correlations, distributions, interactions
+4. Identify high-cardinality columns and outliers
+5. Save report to `reports/` directory
+6. Summarize key findings
+

package/dist/presets/data-ai/commands/data/validate.md

@@ -0,0 +1,18 @@
+---
+description: Validate data quality using schema checks and expectations
+inclusion: manual
+argument-hint: "[file-path] [schema-path]"
+---
+
+## Arguments
+FILE: $1 (required, path to data file)
+SCHEMA: $2 (optional, path to schema/expectations file)
+
+## Workflow
+1. Load dataset and infer schema if not provided
+2. Check for null values, duplicates, type mismatches
+3. Validate value ranges and distributions
+4. Check referential integrity if multiple tables
+5. Report validation results with severity levels
+6. Suggest fixes for common issues
+

package/dist/presets/data-ai/commands/docs/architecture.md

@@ -0,0 +1,17 @@
+---
+description: Generate or update system architecture documentation
+inclusion: manual
+argument-hint: "[output-path]"
+---
+
+## Arguments
+OUTPUT: $1 (default: docs/system-architecture.md)
+
+## Workflow
+1. Analyze project structure and module dependencies
+2. Identify data flow patterns and pipeline architecture
+3. Document ML model serving architecture
+4. Map infrastructure components and connections
+5. Generate architecture diagram descriptions
+6. Write to OUTPUT path
+

package/dist/presets/data-ai/commands/docs/summarize.md

@@ -0,0 +1,17 @@
+---
+description: Analyze the codebase and generate documentation summary
+inclusion: manual
+argument-hint: "[focused-topics] [should-scan-codebase]"
+---
+
+## Arguments
+TOPICS: $1 (default: all)
+SCAN: $2 (default: false, boolean)
+
+## Workflow
+1. Read existing documentation in `docs/` directory
+2. If SCAN is true: analyze source code for undocumented modules
+3. Generate summary report of documentation coverage
+4. Identify gaps between code and documentation
+5. Suggest documentation improvements
+

package/dist/presets/data-ai/commands/docs/update.md

@@ -0,0 +1,17 @@
+---
+description: Update project documentation based on recent code changes
+inclusion: manual
+argument-hint: "[scope]"
+---
+
+## Arguments
+SCOPE: $1 (default: all, options: all, api, architecture, readme)
+
+## Workflow
+1. Identify recent code changes (git diff or file timestamps)
+2. Determine which documentation needs updating
+3. Update API documentation for changed modules
+4. Update architecture docs if structure changed
+5. Verify all code examples still work
+6. Report documentation changes made
+

package/dist/presets/data-ai/commands/fix/build.md

@@ -0,0 +1,17 @@
+---
+description: Diagnose and fix build or dependency issues
+inclusion: manual
+argument-hint: "[error-context]"
+---
+
+## Arguments
+CONTEXT: $1 (optional, error message or log snippet)
+
+## Workflow
+1. Check Python version compatibility
+2. Verify virtual environment is active
+3. Check for dependency conflicts with `pip check`
+4. Resolve version conflicts in requirements
+5. Rebuild if needed (`pip install -e .`)
+6. Verify build with `python -c "import package_name"`
+

package/dist/presets/data-ai/commands/fix/lint.md

@@ -0,0 +1,16 @@
+---
+description: Auto-fix linting issues in Python code
+inclusion: manual
+argument-hint: "[path]"
+---
+
+## Arguments
+PATH: $1 (default: src/)
+
+## Workflow
+1. Run `ruff check --fix $PATH` for auto-fixable issues
+2. Run `ruff format $PATH` for formatting
+3. Run `isort $PATH` for import sorting
+4. Report remaining issues that need manual fixes
+5. Run type check to verify fixes did not break types
+

package/dist/presets/data-ai/commands/fix/tests.md

@@ -0,0 +1,17 @@
+---
+description: Diagnose and fix failing tests
+inclusion: manual
+argument-hint: "[test-path]"
+---
+
+## Arguments
+PATH: $1 (default: tests/)
+
+## Workflow
+1. Run `pytest $PATH -x --tb=short` to identify first failure
+2. Analyze error message and traceback
+3. Identify root cause (code bug vs test bug vs environment)
+4. Apply fix to source or test as appropriate
+5. Re-run failing test to verify fix
+6. Run full suite to check for regressions
+

package/dist/presets/data-ai/commands/git/branch.md

@@ -0,0 +1,16 @@
+---
+description: Create a new feature branch with naming convention
+inclusion: manual
+argument-hint: "[type] [description]"
+---
+
+## Arguments
+TYPE: $1 (default: feature, options: feature, experiment, data, fix, pipeline)
+DESC: $2 (required, branch description in kebab-case)
+
+## Workflow
+1. Fetch latest from remote
+2. Create branch: `git checkout -b $TYPE/$DESC`
+3. Push branch to set upstream
+4. Report branch name and base commit
+

package/dist/presets/data-ai/commands/git/commit.md

@@ -0,0 +1,17 @@
+---
+description: Create a conventional commit with staged changes
+inclusion: manual
+argument-hint: "[type] [message]"
+---
+
+## Arguments
+TYPE: $1 (default: feat, options: feat, fix, data, model, pipeline, docs, test, refactor)
+MESSAGE: $2 (required, commit description)
+
+## Workflow
+1. Check staged files with `git status`
+2. Verify no secrets in staged files
+3. Run lint on staged Python files
+4. Create commit: `git commit -m "$TYPE: $MESSAGE"`
+5. Report commit hash and changed files
+

package/dist/presets/data-ai/commands/git/pr.md

@@ -0,0 +1,17 @@
+---
+description: Create a pull request with conventional format
+inclusion: manual
+argument-hint: "[branch] [from-branch]"
+---
+
+## Arguments
+TO_BRANCH: $1 (default: main)
+FROM_BRANCH: $2 (default: current branch)
+
+## Workflow
+1. Verify all tests pass on current branch
+2. Push current branch to remote
+3. Create PR with `gh pr create`
+4. Include summary of changes, test results, and metrics
+5. Add appropriate labels (data, ml, pipeline, etc.)
+

package/dist/presets/data-ai/commands/lint.md

@@ -0,0 +1,16 @@
+---
+description: Run linting and type checking on Python codebase
+inclusion: manual
+argument-hint: "[path]"
+---
+
+## Arguments
+PATH: $1 (default: src/)
+
+## Workflow
+1. Run `ruff check $PATH` for fast linting
+2. Run `ruff format --check $PATH` for formatting verification
+3. Run `mypy $PATH` for type checking
+4. Report issues grouped by severity
+5. Suggest `ruff check --fix` for auto-fixable issues
+

package/dist/presets/data-ai/commands/ml/compare.md

@@ -0,0 +1,17 @@
+---
+description: Compare multiple model runs or experiments side by side
+inclusion: manual
+argument-hint: "[run-ids]"
+---
+
+## Arguments
+RUNS: $1 (required, comma-separated run IDs or experiment names)
+
+## Workflow
+1. Load experiment metadata for specified runs
+2. Compare hyperparameters across runs
+3. Compare metrics (accuracy, loss, F1, etc.)
+4. Identify best performing configuration
+5. Generate comparison table and visualizations
+6. Recommend which model to promote
+

package/dist/presets/data-ai/commands/ml/deploy.md

@@ -0,0 +1,18 @@
+---
+description: Deploy a trained model to a serving endpoint
+inclusion: manual
+argument-hint: "[model-path] [target]"
+---
+
+## Arguments
+MODEL: $1 (required, path to model or registry URI)
+TARGET: $2 (default: local, options: local, docker, cloud)
+
+## Workflow
+1. Load and validate model artifact
+2. Generate serving configuration (BentoML/FastAPI/TF Serving)
+3. If docker: build container image with model
+4. If cloud: push to model registry and deploy endpoint
+5. Run smoke test against endpoint
+6. Report endpoint URL and health status
+

package/dist/presets/data-ai/commands/ml/evaluate.md

@@ -0,0 +1,19 @@
+---
+description: Evaluate a trained model on test data with comprehensive metrics
+inclusion: manual
+argument-hint: "[model-path] [test-data-path]"
+---
+
+## Arguments
+MODEL: $1 (required, path to model checkpoint or MLflow run ID)
+DATA: $2 (required, path to test dataset)
+
+## Workflow
+1. Load model from checkpoint or registry
+2. Load and preprocess test data
+3. Generate predictions
+4. Compute metrics (accuracy, F1, AUC, RMSE as appropriate)
+5. Generate confusion matrix and classification report
+6. Save evaluation artifacts to `reports/`
+7. Compare against baseline if available
+

package/dist/presets/data-ai/commands/ml/train.md

@@ -0,0 +1,20 @@
+---
+description: Train a machine learning model with experiment tracking
+inclusion: manual
+argument-hint: "[config-path] [experiment-name]"
+---
+
+## Arguments
+CONFIG: $1 (required, path to training config YAML/JSON)
+EXPERIMENT: $2 (default: auto-generated from config)
+
+## Workflow
+1. Load training configuration
+2. Initialize experiment tracking (MLflow/W&B)
+3. Load and validate training data
+4. Initialize model and optimizer
+5. Run training loop with validation
+6. Log metrics, parameters, and artifacts
+7. Save best model checkpoint
+8. Report final metrics and model location
+