coding

Plugin coding in crsiebler/dotfiles. Commands read from the repo's manifest — 4 days ago.

Skills (25)

• apply-ai-standards

  Apply AI, ML, LLM and NLP lifecycle standards to model development, evaluation and serving. Not automatic training, model installation or deployment.

• apply-app-standards

  Apply Electron, Flutter, mobile platform and PowerShell GUI/TUI standards for app lifecycle, native integration and interface behavior. Not device provisioning, app publishing or asset generation.

• apply-blockchain-standards

  Apply smart-contract, blockchain and DApp standards to scoped code or review work, including storage, calls, finality and local verification. Not investment advice, wallet access or transaction signing.

• apply-database-standards

  Apply SQL and PostgreSQL standards for queries, transactions, schemas, performance and operations against the actual engine/version. Not permission to execute queries, migrate or tune production.

• apply-infrastructure-standards

  Apply Azure, Kubernetes, Terraform, Windows, Microsoft 365, Active Directory and PowerShell security standards to scoped infrastructure work. Assessments stay read-only; no implicit tenant, cluster or identity changes.

• apply-integration-standards

  Apply Git repository tooling, GraphQL, Slack and WebSocket standards to API, messaging and service boundaries. Not permission to send messages, install applications or modify access controls.

• apply-language-standards

  Apply project language standards for C++, Elixir/OTP, Go, Java, JavaScript, Kotlin, PHP, PowerShell, Python, Rust, Swift or TypeScript during implementation or review. Not agent selection or a language upgrade.

• apply-web-standards

  Apply framework standards for Angular, Django, Laravel, Next.js, Rails, React, Spring Boot and Vue/Nuxt in existing web projects. Not framework selection or automatic migration.

• archive-run

  Preserve a completed plan or archive a verified delivered run using the bundled deterministic Python helper. Use for explicit authorized run closeout or preservation; not planning, story execution, review, team dispatch or runner recovery.

• assess-controls

  Inspect architecture, security and compliance controls from authorized artifacts and evidence; return findings and proposed validation. No test execution or remediation.

• create-mcp-server

  Design, implement, and verify MCP servers that expose APIs or local capabilities as agent tools. Use when creating a custom MCP server or extending its tool contracts, transport, or provider adapter; not for connecting an existing server or selecting an LLM provider.

• create-skill

  Author, adapt, and evaluate portable skills with representative task evidence and iterative improvement. Use for creating skills or improving their instructions and discovery descriptions.

• develop-code

  Implement features, diagnose and fix bugs, or refactor code while preserving contracts and verifying behavior. Use for development work, including diagnosis-only requests; not standalone planning, conceptual explanations, or read-only code review.

• format-code

  Format and lint intended changed files using the repository's configured tools across languages; inspect automatic changes and report unavailable checks.

• manage-changes

  Ship scoped changes through Git commits, branch pushes and GitHub pull requests with explicit write approval.

• map-codebase

  Creates and updates source-backed repository maps in focused technical documents. Use when explicitly mapping a codebase, documenting repository context, or consolidating overlapping repository documentation.

• operate-systems

  Plan or execute authorized operational changes, deployment, installation, recovery and controlled experiments with explicit target and recovery constraints. Not standing production access.

• prepare-implementation

  Plan autonomous implementation of approved PRDs or bounded requirements with agent-executable acceptance checks, dependency DAG, and upfront resolution of approval gates. Planning only; does not execute the plan.

• resolve-review-feedback

  Assesses pull-request feedback against code, prioritizes actionable changes, verifies addressed concerns at the PR head, and prepares approved replies and thread resolutions. Use for analyzing review comments or resolving addressed feedback.

• review-code

  Review local changes, pull requests, branches, selected components, or bounded codebase audits for evidenced defects and maintenance risks, including security and complexity assessments. Independent assessment with authorized focused verification; no automatic fixes.

• review-story

  Review one staged story using a workflow-supplied candidate, acceptance criteria, verification scope and bounded review session. Use for a requested story delivery gate with a generic reviewer; not ordinary PR review or implementation.

• run-agent-team

  Execute an authorized coding plan autonomously with parallel agents in the normal main session, isolated worktrees, automated verification and bounded review. Use for explicit multi-agent requests; not planning-only work, simple edits, or installing a separate runner.

• run-tests

  Run project-native focused, regression and integration checks across languages; diagnose failures and report the actual verification scope and limits.

• verify-interface

  Verifies rendered UI behavior, accessibility, and browser errors.

• write-requirements

  Define scoped outcomes and verifiable acceptance for bounded tasks or larger PRDs.

Similar plugins