deepagents-builder

Plugin deepagents-builder from marketplace spuli-plugins in spulido99/claude-toolkit. Commands read from the repo's manifest — 1 month ago.

Skills (7)

• DeepAgents Architecture

  Topology reference for deepagents agents: write coupling decides between one frontal agent (assistant pattern, the default) and read-only subagents; bounded contexts packaged as skills; business-capability mapping. Use when deciding how many agents to build, whether to add a subagent, or how to spli

• deepagents-builder

  Entry point to the deepagents-builder plugin for building Python agents with LangChain's deepagents (`create_deep_agent`). Use when the user names deepagents-builder or deepagents, or wants to start an agent project, decide its topology, design its tools or prompts, write or run its evals, review it

• Evals-Driven Development

  Evals-driven development for deepagents agents: JTBD scenarios, evaluators, snapshots, mocks, and security evals. Use when defining what to test, writing or running evals, or debugging a failing agent.

• DeepAgents Evolution

  Maturity model and refactoring patterns for deepagents agents. Use when assessing an existing agent's architecture or planning its refactor to the next level.

• DeepAgents Patterns

  Implementation patterns for deepagents agents: the canonical create_deep_agent API, system prompts, tool and security patterns, and anti-patterns. Use when writing agent code or prompts, looking up the API, or checking for common mistakes.

• DeepAgents Quickstart

  Minimal working setup for a deepagents agent: install, first create_deep_agent call, chat loop. Use when building a first agent or prototyping quickly.

• tool-design

  Eleven principles for AI-friendly agent tools: LangChain @tool functions, MCP tool definitions, deepagents tool catalogs. Use when designing tools, converting an API into tools, or reviewing a tool's name, parameters, errors, or confirmation level.

Slash commands (16)

• /deepagents-builder:add-interactive-chat

  Generate an interactive chat console (chat.py) for testing your agent with tool call logging, thread management, and optional context injection.

• /deepagents-builder:add-scenario

  Add a single eval scenario to an existing dataset interactively or from a production trace.

• /deepagents-builder:add-subagent

  Add a single subagent to an existing agent architecture interactively or by named capability.

• /deepagents-builder:add-tool

  Add a single tool to an existing catalog interactively or from an API endpoint.

• /deepagents-builder:assess

  Run a maturity assessment on your agent architecture. Scores across 4 categories and identifies the path to the next level.

• /deepagents-builder:design-agent

  Design a simple single agent interactively — role, tools, prompt, and model. Generates create_deep_agent(...) code. Escalates to /design-topology if complexity warrants multi-agent.

• /deepagents-builder:design-evals

  Design eval scenarios from JTBD. Scaffolds evals directory and generates scenario datasets for new or existing agents.

• /deepagents-builder:design-tools

  Design an AI-friendly tool catalog from scratch. Detects agent code, discovers requirements, then generates tools following the 11 principles.

• /deepagents-builder:design-topology

  Interactive guide to design agent topology based on write coupling — single frontal agent (assistant pattern) by default, subagents only for read-only parallelizable work.

• /deepagents-builder:eval-status

  Show eval dataset health dashboard — scenario counts, snapshot staleness, last run results.

• /deepagents-builder:eval-update

  Review and accept/reject changed snapshots after intentional agent changes.

• /deepagents-builder:eval

  Run evals against your agent. Default snapshot mode. Use --smoke, --full, --report, or --diagnose for other modes.

• /deepagents-builder:evolve

  Guided refactoring to evolve your agent to the next maturity level. Assesses current state, recommends a refactoring pattern, and walks through implementation.

• /deepagents-builder:new-sdk-app

  Create and setup a new DeepAgents application with scaffolding, dependencies, and example code. Guides through project creation interactively.

• /deepagents-builder:tool-status

  Show tool catalog quality dashboard — counts by domain, quality scores per tool, principle compliance, and eval coverage.

• /deepagents-builder:validate-agent

  Validate a DeepAgents configuration for anti-patterns, security issues, and best practices compliance.

Subagents (6)

• agent-architect

  Designs deepagents agent topology by write coupling: one frontal agent by default, read-only subagents only when they pay off. Use for architecture decisions, splitting domains into skills or subagents, or a tool catalog that has grown too large.

• code-reviewer

  Reviews deepagents agent code for anti-patterns, security issues, and API misuse. Use after agent code is written or changed.

• eval-designer

  Designs eval scenarios for deepagents agents from jobs-to-be-done and scaffolds the evals directory. Use to start an eval suite, add scenarios, or define what to test before the agent exists.

• eval-runner

  Runs eval scenarios against a deepagents agent, manages snapshots, and diagnoses failures. Use to run evals, review changed snapshots, or explain a failing scenario.

• evolution-guide

  Scores a deepagents agent on the 80-point maturity model and guides refactoring to the next level. Use for maturity assessments or step-by-step architecture refactoring.

• tool-architect

  Designs and generates AI-friendly tool catalogs for deepagents agents, from scratch or from an API. Use to create tools, convert REST or OpenAPI endpoints into tools, or add one tool to an existing catalog.

Similar plugins