The Built-in Agent Toolkit (2026)
WorkDaddy agents don’t just chat — they execute. 17 built-in tools cover the full loop: real shell commands, surgical file edits, live browser checks, web research, creative production, and human-approval gates.
Under every WorkDaddy agent is the same engine that powers serious coding agents — extended for commerce. This page documents the 17 built-in tools that ship with the app in 2026: what each does, why it matters for an e-commerce team, and how the draft-first policy keeps every powerful capability accountable. If MCP servers and connectors are what agents reach out to, these tools are what agents are.
Core Coding Tools
Bash Shell
Agents run real shell commands in a sandboxed workspace — builds, scripts, data crunching, and deployments execute for real, not in theory.
Learn more →File Editing (Read / Write / Edit / Patch)
Precise file reads, writes, edits, and patch application — theme code and content change surgically, with every diff reviewable.
Learn more →Code Search (Glob / Grep)
Fast pattern and content search across the whole workspace — agents find the right file before touching anything.
Learn more →Language Server (LSP)
Go-to-definition, references, symbols, and call hierarchy — agents navigate storefront code like an IDE, not like a text scanner.
Learn more →Repository Tools
Clone repositories and generate structural overviews — agents onboard onto any storefront codebase in minutes.
Learn more →Research
Web Fetch
Fetch any page as text, markdown, or HTML — competitor pages, docs, and references become working context on demand.
Learn more →Web Search (Exa / Parallel)
Live web search backed by Exa and Parallel — market research and trend scans use today's web, not last year's training data.
Learn more →Orchestration
Subagent Tasks
Agents delegate to background subagents and track their status — big jobs fan out in parallel instead of running one long queue.
Learn more →Todo Planning
Structured task lists keep multi-step work honest — you see the plan, the progress, and what's left at every moment.
Learn more →Skills
Reusable SKILL.md playbooks load on demand — your team's know-how becomes an executable asset every agent can follow.
Learn more →Human-in-the-Loop Approvals
Draft-first by design — agents ask questions, request env vars, and wait for approval before anything irreversible ships.
Learn more →Automation
Browser Automation
A built-in Chrome DevTools browser panel — agents click through your store like a shopper and catch what static checks miss.
Learn more →Interactive Terminal
A full PTY-backed terminal — long-running processes, dev servers, and interactive CLIs run under agent control with live output.
Learn more →Scheduled Tasks
Cron-style scheduling built in — daily digests, nightly audits, and weekly reports run on time without anyone remembering to ask.
Learn more →Creative
Image Generation
gpt-image-2 generation inside the agent workflow — creative gets produced, reviewed, and placed without leaving the loop.
Learn more →Document Toolkit (Word / Excel / PPT)
Native docx, xlsx, and pptx generation — weekly reports, buying sheets, and decks come out as real office files, ready to share.
Learn more →The toolkit, by layer
The tools compose into a working loop — find, change, verify, ship:
Core coding: bash shell, file read/write/edit/patch, glob & grep search, language-server navigation, repository tools
Research: live web search (Exa / Parallel) and full-page web fetch
Orchestration: background subagents, structured todo plans, reusable skills, human-approval gates
Automation: a real browser panel, an interactive PTY terminal, and cron-style scheduled tasks
Creative & interface: image generation, native Word/Excel/PowerPoint output, and hands-free voice mode
Why built-in beats bolted-on in 2026
Because the tools share one workspace and one approval policy, agents chain them without glue code: search finds the file, LSP traces the usage, edit changes it, the browser verifies the storefront, and the scheduler makes it recur. Every step is logged and reviewable — which is exactly what you want when the workspace is your revenue.
Frequently asked questions
What tools do WorkDaddy agents have in 2026?
All 17 built-in tools: bash, file editing, code search, LSP, web fetch, web search, subagents, todo planning, skills, approval gates, browser automation, image generation, Office document output, an interactive terminal, scheduled tasks, voice mode, and repository tools.
Can WorkDaddy agents really run code and commands?
Yes — agents execute real shell commands and edit real files in a sandboxed workspace. Irreversible operations are approval-gated, and every change is presented as a reviewable diff.
How do built-in tools differ from MCP integrations?
Built-in tools are the agent’s own hands — local execution, editing, browsing, and research. MCP servers and connectors are the outside world: SaaS platforms, data sources, and services. Real work usually chains both.
Do I need to configure anything to use these tools?
No — the toolkit ships with the app. Only a few surfaces need keys (voice mode and image generation use OpenAI APIs) or OS permissions (computer use on macOS).
Explore more integration directories
Put the team to work on your store
Connect your storefront, analytics, and email stack, set a goal, and let the agents run the work end to end. Start on the free plan with your own model key.