yu-iskw

Reload in-memory artifacts after dbt run or a known artifact change. Use on the same target without restarting MCP.

Establish the dbt artifact root for this session (local path, s3://, or gs://). Use before any other dbt-tools analysis primitive.

Check MCP session state—target, load time, stale flag, runs, warehouse type. Use as a gate before analysis primitives.

Establish the dbt artifact root for this session (local path, s3://, or gs://). Use before any other dbt-tools analysis primitive.

Check whether dbt artifacts exist, are fresh, and which readiness level applies. Use as a gate before manifest- or run_results-based analysis.

Return metadata and context for one dbt resource by unique_id. Use when explaining what a model, source, or test does.

Find dbt resources by name, type, tag, package, or path. Use to resolve unique_id before describe-resource, trace-dependencies, or query-executions.

Filter and sort run_results node executions by status, time, and warehouse metrics. Use for post-run triage and slow-node investigation.

Check license compatibility, data privacy compliance, and AI ethics. Use when adding dependencies, handling user data, or reviewing regulatory requirements.

Deploy the application or manage infrastructure. Handles Docker builds, CI/CD, and deployment workflows.

Design a UI component with specifications for layout, states, interactions, and accessibility. Use when creating new UI components or redesigning existing ones.

Implement a feature or fix a bug following the project's code patterns and conventions. Use when code changes are needed.

Create a structured implementation plan for a feature, refactoring, or multi-step task. Use as the first step in the pipeline before /orchestrate. Produces a plan that the orchestrator can turn into a delegation plan.

Review code changes for quality, security, and adherence to project conventions. Use after making code changes or when reviewing a pull request.

Perform a security audit of the codebase. Checks for OWASP Top 10, AI-specific vulnerabilities, dependency issues, and configuration problems.

Write user stories, acceptance criteria, and technical requirements for a feature or change. Use when defining what needs to be built.

Analyze sqlfluff-complexity JSON reports (hotspot digest and threshold tuning). Reuse --output when the file is no older than 5 minutes; otherwise run report. Scan paths and output path are always user-specified.

Guide users through configuring sqlfluff-complexity for a SQLFluff project by sampling reports, choosing a preset, explaining thresholds, per-directory strictness (nested .sqlfluff vs path_overrides), validating config, and recommending gradual CI rollout.

Inspect repository layout with tree and find, compare to AGENTS.md conventions, and fix misplaced or overly flat generated files. Use when auditing folder structure, after scaffolding or bulk file generation, when output looks flat, or when asked where code/tests/docs should live. Supports inspecting the repository root or a specified subdirectory.

Run linters and fix violations using Trunk when available; when Trunk is missing, run the equivalent tools from the dev environment (see .trunk/trunk.yaml and pyproject.toml) so agents still enforce Ruff, Pyright, Pylint, Bandit, and other enabled checks where possible.

Run unit tests and automatically fix code failures, regression bugs, or test mismatches. Use when tests are failing, after implementing new features, or to repair "broken" tests.

Build the project and automatically fix packaging or build errors (for example Hatch failures) and related breakage. Use when the project fails to build, shows "broken" states, or after making significant changes.

Run CodeQL security/quality analysis and fix findings. Use when the user asks to run CodeQL, security scan, static analysis, or fix CodeQL findings.

Interactive deep research and decision support: frame the real problem (XY-aware), ask exactly 10 multiple-choice questions one at a time, then produce a rigorous comparative evaluation (default 5 approaches, 0–100 scores) and recommendation. Use when the user wants structured discovery before committing to a solution, a scored comparison of approaches, or to avoid jumping straight to an answer—especially for architecture, strategy, or high-stakes trade-offs.

Consistently bump and synchronize versions across all packages in the monorepo using pnpm. Use this to ensure the root and all workspace packages share the same version number.

Run comprehensive integration tests using real skills (Vercel, Anthropic, Expo) in a Docker container. Managed via Makefile for standard lifecycle control.

Synchronize technical sections in README.md and CONTRIBUTING.md with the actual project state.

Run linters and fix violations, formatting errors, or style mismatches using Trunk. Use when code quality checks fail, before submitting PRs, or to repair "broken" linting states.

Build the project and automatically fix any build errors, compilation failures, or type mismatches. Use when the project fails to build, shows "broken" states, or after making significant changes.

Perform a "hard reset" of the development environment. Use when dependencies are corrupted, lockfiles are out of sync, or environment tools (Trunk/pnpm) are in an inconsistent state.

Fix a GitHub issue end-to-end. Use when given an issue number to fix, implementing bug fixes, or addressing reported problems.

Self-improvement skill for evolving Claude Code configuration. Use when you notice repeated mistakes, want to add new workflows, or optimize the development experience.

Upgrade outdated npm dependencies in the pnpm monorepo (recursive update, latest, targeted packages). Use when the user asks to upgrade dependencies, bump packages, refresh lockfile, pnpm update, outdated packages, latest versions, or dependency refresh—not monorepo release versions.

Format code, run linters and static analysis, and find unused code (ESLint + Knip). Use when the user asks to fix lint, format code, run formatters, find dead code, unused exports, run knip, fix style, tidy code, or run eslint --fix.

Manage Architecture Decision Records (ADRs). Use this to initialize, create, list, and link ADRs to document architectural evolution. Requires 'adr-tools' to be installed.

Use at the end of a coding session to summarize outcomes, failures, inefficiencies, and root causes, then output a concise postmortem report with ranked improvements for next time. Output only in chat; do not edit project files unless the user explicitly asks. Skip nit-picks and one-off mistakes.

Interactively bump package versions for the parser package. Use when the user asks to bump version, increment version, set release version, or release dbt-artifacts-parser.

Regenerates TypeScript types from dbt artifact JSON schemas. Use when the user asks to update parsers, sync with upstream, refresh types, or regenerate parser types from existing schemas.

Interactive deep research and decision support: frame the real problem (XY-aware), ask exactly 10 multiple-choice questions one at a time, then produce a rigorous comparative evaluation (default 5 approaches, 0–100 scores) and recommendation. Use when the user wants structured discovery before committing to a solution, a scored comparison of approaches, or to avoid jumping straight to an answer—especially for architecture, strategy, or high-stakes trade-offs.

Run unit tests and fix failures. Use when the user asks to run tests, fix failing tests, make tests pass, tests are broken, fix test errors, or get tests green.

Run pnpm build from the repo root and fix compile or build errors until the workspace build passes. Use after edits, when CI is red on build, or before opening a PR. Keywords: pnpm, build, TypeScript, monorepo, workspace.

Run pnpm format then pnpm lint from the repo root and fix issues until both pass. Use after edits, when CI is red on lint/format, or before opening a PR. Keywords: pnpm, format, lint, trunk, monorepo, code style.

Manage Architecture Decision Records (ADRs). Use this to initialize, create, list, and link ADRs to document architectural evolution. Requires 'adr-tools' to be installed.

Run pnpm test from the repo root and fix failing tests or implementation until the suite passes. Use after edits, when CI is red on tests, or before opening a PR. Keywords: pnpm, test, vitest, unit tests, monorepo.

Use at the end of a coding session to summarize outcomes, failures, inefficiencies, and root causes, then output a concise postmortem report with ranked improvements for next time. Output only in chat; do not edit project files unless the user explicitly asks. Skip nit-picks and one-off mistakes.

Assign @copilot on an existing SecOps remediation issue after policy guard—use when Project link or org policy requires Issue → Project before assignee. Uses gh issue edit; does not push branch code.

One-shot PR status check with gh—structured JSON outcome and exit codes; classify green vs failing vs pending vs blocked. Read-only on the repo. Does not loop; re-invoke on a schedule (human, Claude, or optional sub-agent). No issue comments—use secops-post-ci-nudge-comment to nudge.

Create a SecOps remediation issue on the target repo (after policy guard). Link a GitHub Project at create time via --project (gh issue create -p); optional @copilot assign. For assign-after-Project, use secops-assign-copilot-to-issue. Uses gh issue create; does not push branch code.

Build the project and automatically fix any build errors, compilation failures, or type mismatches. Use when the project fails to build, shows "broken" states, or after making significant changes.

Perform a "hard reset" of the development environment. Use when dependencies are corrupted, lockfiles are out of sync, or environment tools (Trunk/uv) are in an inconsistent state.

Fix a GitHub issue end-to-end. Use when given an issue number to fix, implementing bug fixes, or addressing reported problems.

Self-improvement skill for evolving Claude Code configuration. Use when you notice repeated mistakes, want to add new workflows, or optimize the development experience. Enables Claude to improve its own CLAUDE.md, skills, hooks, and agents.

Run linters and fix violations, formatting errors, or style mismatches using Trunk. Use when code quality checks fail, before submitting PRs, or to repair "broken" linting states.

Manage Architecture Decision Records (ADRs) using adr-tools. Use this to initialize, create, list, and link ADRs to document architectural evolution.

Manage changelogs using Changie. Provides tools to initialize, add change fragments, batch releases, and merge version notes.

Execute complex tasks using parallel subagents. Use for large tasks that benefit from concurrent execution. Automatically plans, decomposes, and orchestrates parallel work.

Bump or upgrade declared dependency versions in this pnpm workspace (root and packages/* package.json), with supply-chain checks before and after install. Use when the user asks to upgrade, bump, or refresh npm dependencies in manifests—not only the lockfile.

Check license compatibility, data privacy compliance, and AI ethics. Use when adding dependencies, handling user data, or reviewing regulatory requirements.

Deploy the application or manage infrastructure. Handles Docker builds, CI/CD, and deployment workflows.

Design a UI component with specifications for layout, states, interactions, and accessibility. Use when creating new Vue.js components or redesigning existing ones.

Implement a feature or fix a bug following the project's TypeScript patterns and conventions. Use when code changes are needed.

Produce a structured delegation plan for parallel agent execution. Takes a task description or plan output, analyzes dependencies, assigns agents, and groups tasks for parallel execution. Use after /plan-task or when you need to plan how to delegate work across agents.

Create a structured implementation plan for a feature, refactoring, or multi-step task. Use as the first step in the pipeline before /orchestrate. Produces a plan that the orchestrator can turn into a delegation plan.

Research and gather reliable information about libraries, APIs, specifications, and best practices. Use when you need to understand external documentation before planning or implementing a feature.

CRUD operations for Lightdash project agents.

Create and validate Claude sub-agent definitions with correct frontmatter, tool permissions, and delegation boundaries.

Inspection of thread history, memory, and organization-wide agent activity.

Orchestrate evaluation runs and test case management for Lightdash agents.

Package and distribute Claude Code plugins with schema-safe manifests, canonical layout, component wiring, and release-ready validation.

Verify Claude extensions and plugins across hooks, skills, sub-agents, teams, manifest, structure, runtime loading, and CI parity.

Manage Architecture Decision Records (ADRs). Use this to initialize, create, list, and link ADRs to document architectural evolution. Requires 'adr-tools' to be installed.

Implement and maintain Claude Agent Skills with valid SKILL.md frontmatter, focused instructions, and progressive disclosure.

Technical project manager agent. Use proactively to synchronize repository work with GitHub Project boards.

Expert triage agent. Use proactively to categorize, label, and assign new issues.

Sets the active GitHub project and repository context by writing a repo-level config file at .github/project-config.json. Run once per repository to eliminate repeated context verification prompts and share the context with all team members via git. All subsequent agent sessions will auto-verify silently against this config.

Verifies the current GitHub authentication status and git remote against the repo-level config at .github/project-config.json. Proceeds silently when the environment matches the config. Alerts only when a mismatch is detected or no config exists. Use at the start of every session.

Comprehensive management of GitHub issues, including sub-issue hierarchies. Use to create, update, close, list, search, view, comment on, and manage parent-child relationships between issues in a single skill.

Implements new Agent Skills for the project. Identifies the AI coding tool (Cursor, Claude Code, Gemini CLI), ensures specification compliance, and provides specialized templates. Use when creating, authoring, or adding a new skill, or when the user asks about Agent Skills format or SKILL.md.

Run make format and make lint, then fix linter violations. Use when formatting code or fixing trunk/lint issues.

Synchronizes the list of Agent Skills in the root README.md with the actual skills present in the codebase.

Build the project and automatically fix any build errors, compilation failures, or type mismatches. Use when the project fails to build, shows "broken" states, or after making significant changes.

Inspect repository layout with tree and find, compare to AGENTS.md conventions, and fix misplaced or overly flat generated files. Use when auditing folder structure, after scaffolding or bulk file generation, when output looks flat, or when asked where code/tests/docs should live. Supports inspecting the repository root or a specified subdirectory.

Run CodeQL security/quality analysis and fix findings. Use when the user asks to run CodeQL, security scan, static analysis, or fix CodeQL findings.

Self-improvement skill for evolving Claude Code configuration. Use when you notice repeated mistakes, want to add new workflows, or optimize the development experience.

Run linters and fix violations, formatting errors, or style mismatches using Trunk. Use when code quality checks fail, before submitting PRs, or to repair "broken" linting states.

Safely upgrade Node.js dependencies in pnpm workspaces. Use when asked to "upgrade dependencies", "update packages", "check for updates", or fix version mismatches.

End-of-session capture of failures, surprises, and lessons so agent behavior and repo config improve. Use after non-trivial work; skip trivial sessions unless something went wrong.

Scan for vulnerable dependencies and CVEs via pnpm lint:security (Trunk Trivy/OSV-scanner), optional pnpm security:grype or pnpm lint:all (Trunk-all plus Grype). Use for dependency CVE checks, security-scoped lint, or post-bump verification.

Manage Architecture Decision Records (ADRs) using adr-tools. Use this to initialize, create, list, and link ADRs to document architectural evolution.

Scaffold and standardize new dbt-helpers plugins (Warehouse, Tool, or Schema). Use when starting a new plugin implementation or updating plugin discovery configuration.

Comprehensive workflow for creating, implementing, and validating Agent Skills. Use when asked to "create a new skill", "author a skill", "add a capability", or when standardizing project-specific workflows. Support for platform detection (Cursor, Claude Code, Gemini CLI, Codex) and template selection.

Set up the development environment for the project. Use when starting work on the project, when dependencies are out of sync, or to fix environment setup failures.

Implement and run End-to-End tests using golden output comparison. Use when verifying CLI behavior, plan generation, or full project execution.

Set up and run integration tests using Testcontainers. Use when testing warehouse adapters against real database engines or verifying filesystem-dependent operations.

Implement and use Nullable Infrastructure adapters for mock-free testing. Use when writing unit tests for core logic, implementing infrastructure adapters, or setting up test fixtures.

Manage and enforce path templating and naming conventions in dbt-helpers. Use when implementing PathPolicy, defining template variables, or configuring warehouse-specific defaults.