Canopy / Guides
Build with agents. Stay in control.
Practical workflows for getting a project running, directing agents, and checking the work before it ships.
Choose the job in front of you
Start here
Start a first project
Install an agent CLI, open a real repository, and check one small result.
Open guide → Start hereGet the app running
Find the project's own command, local URL, and first useful error.
Open guide → Start hereDirect several agents
Give each agent a result, checkout, and review owner.
Open guide → Start hereReview a pull request
Follow the requested behavior through the diff, live app, and latest checks.
Open guide → Start hereDecide whether to launch
Bring access, data, payment, recovery, and ownership into one release decision.
Open guide → Start hereUnderstand token usage
Separate repeated context and cache categories before changing models.
Open guide →Need one quick answer? Browse common questions → · Want a subject area? Browse topics →
All 72 guides
Browse alphabetically or filter by title, description, or topic. Filtering happens in this browser.
Agent session, terminal tab, branch, or worktree: what is the difference? A practical map of the four things people mix up when running multiple coding agents, with a two-agent example and checkout verification commands. AI-built app works locally but fails on Vercel Preview Find the first difference between a working Canopy run and a broken Vercel Preview: commit, build, environment, runtime request, or backend target. AI-generated pull request review checklist and reviewer prompt A practical checklist for intent, changed files, risk, tests, UI behavior, comments, and the final diff, with a copyable reviewer brief. An AI coding-agent task brief you can copy and use A bounded brief with context, acceptance checks, constraints, and a handoff format for Claude Code, Codex, or another coding CLI. Can a noncoder build software with AI coding agents? A realistic first-project path for noncoders: choose a bounded task, run the result, review observable behavior, and know when a developer must check the change. Can I launch my AI-built app? A founder's decision sheet A go-or-hold review for an AI-built web app: running behavior, access, data, payments, recovery, and the exact release version. Can two coding agents use the same development database? Find which database each agent worktree really uses, choose shared or isolated state deliberately, and review competing migrations before integration. Coding agent hit a usage limit mid-task? Save the work first Identify the actual limit, preserve the checkout and partial result, then resume, wait, or hand off without repeating work or changing accounts blindly. Coding agent stuck or waiting for input? Check the live session Distinguish a pending decision, active tool, running dev server, completed turn, and failed agent before interrupting or restarting. Debug an agent-built page using the browser console and network log Collect the exact failing action, console message, request status, service log, and checkout before asking an agent to fix a broken preview. Did your coding agent weaken a test to make CI pass? Review changed assertions, skips, mocks, and browser setup against the original behavior before accepting an agent's green check. Did your coding-agent skill actually load? A five-minute test Separate skill discovery, invocation, and useful execution with an explicit test, a natural-language test, and one negative control. Git worktrees for parallel coding agents: commands and cleanup Create a separate branch and directory per coding agent, check status, integrate reviewed work, and remove worktrees safely. How do I ask a coding agent for a demo I can inspect? A founder-friendly request and acceptance checklist that turns an agent's 'done' message into a running page, observable behavior, and a reviewable diff. How do I check an AI-built page with a keyboard and screen reader? Review an agent-built form and dialog with keyboard, accessible-name, error, and screen-reader checks before accepting the PR. How do I find the command that runs an AI-built app? A beginner workflow for finding the project directory, runtime, run script, local URL, and first useful error before asking an agent to fix setup. How do I keep a coding agent in the right app in a monorepo? Identify the owning package, shared dependencies, checkout, and run command before an agent edits one feature in a multi-app repository. How do I resolve a merge conflict between two coding agents? Find the conflicting branches, compare each agent's intent, make one integration decision, and test the combined behavior. How do I run a Trigger.dev worker beside my app in Canopy? Set up Trigger.dev once, save its dev command beside the website and API, run a dashboard test, and debug the right process. How do I run Supabase locally with my AI-built app? Start the Supabase CLI stack, connect the local app to its actual project URL and keys, and verify a test record stays off production. How do I stop a coding agent rebuilding a feature that already exists? Make the agent locate the active implementation and its callers before editing, then review the smallest change against the running app. How do I test a Trigger.dev email task without emailing real users? Trace one local signup through the Development worker and a test email destination before an agent runs a side-effecting task. How do I test an AI-built UI before merging it? Review the running page across viewport sizes, interaction states, console and network errors, keyboard use, and the latest PR diff. How do I turn a Figma design into a running page with a coding agent? Hand an agent one frame, states, assets, and component constraints; run the page in Canopy and review behavior, visuals, and the final diff. How many coding agents should you run at once? A practical way to size parallel agent work by independent tasks, review capacity, worktrees, local resources, and provider usage. How to evaluate an AI coding workspace on a real project A repeatable trial for terminal, IDE, and multi-agent workspaces that follows one change from setup through running behavior, review, resume, and accepted result. How to give a coding agent useful visual feedback Turn a vague 'this looks wrong' into a page, viewport, element, expected behavior, and screenshot an agent can act on. How to hand an AI-built app to a developer for review A concrete handoff for founders: reproduce the app, show the exact change, list verified behavior and unknowns, and ask a developer for a risk decision. How to inspect an open-source AI IDE before installing it A repeatable Canopy release check: match installer to tag, verify the asset digest, read source and license boundaries, and test what connects to the network. How to reduce coding-agent token usage without losing the result Scope tasks, manage growing context, choose models deliberately, and compare accepted work rather than chasing a low token count. How to review an AI-generated access-control PR Use a permission matrix and cross-user tests to review an agent-written authentication or authorization change before merge. How to review an AI-generated pull request A review workflow for agent-written code: inspect the diff, ask for a risk map, address comments, fix CI, and merge deliberately. How to run multiple coding agents without losing track A two-agent workflow with task ownership, separate worktrees, service ports, attention checks, and a final review decision. How to set a practical budget for parallel coding agents Decide how many agents to launch, assign each an output and stop point, then track plan usage, estimates, review load, and accepted work. How to set up Canopy and run your first AI coding agent Install an agent CLI, open a project in Canopy, run the app, and review your first change. Includes a downloadable practice project. How to stop an endless AI code review loop When each review pass finds another issue, triage findings against the same acceptance checks, verify the latest commit, and decide what blocks this PR. Is my local AI-built app using the production database? Trace the browser, API, worker, and database targets before an agent tests writes or migrations in a locally running app. Keep building a Bolt app with local coding agents in Canopy Connect Bolt to GitHub, isolate an agent change, check the app's backend and run commands, merge in GitHub, and verify Bolt and the live site. Keep coding-agent context useful between sessions Make a handoff that preserves the task, decisions, branch, and evidence without carrying an entire chat transcript forever. MCP server configured but unavailable to your coding agent? Diagnose MCP scope, connection, authentication, tool selection, and cross-CLI setup with a read-only documentation server as the example. Measure AI coding cost per accepted change A repeatable worksheet for comparing model choices by accepted work, review time, retries, token categories, and provider charges. My coding agent committed an API key. What now? A staged response to an agent-written secret leak: stop propagation, revoke the credential, restore services, inspect exposure, and decide whether Git history needs cleanup. My coding agent edited the wrong branch or worktree. What now? Identify the actual checkout, preserve every change, move the intended work safely, and verify the correct branch before restarting the agent. Research a codebase question without changing code Choose an answer, source-backed investigation, or implementation task deliberately, then keep findings available for the next decision. Review a coding-agent skill before your team shares it Inspect a borrowed SKILL.md, its scripts, tool access, and test behavior before making a repeatable workflow available to every agent. Review an AI-built Stripe Checkout before taking payments A founder's test-mode review of Checkout, signed webhooks, delayed payments, duplicate events, and fulfillment before launch. Review an AI-generated Supabase migration before deployment Check the target project, schema diff, existing data, RLS, staging result, and actual restore option before an agent-written migration reaches production. Run a website, API, and background worker in one workspace Configure project run commands once, then start, inspect, and restart your local development stack from Canopy. Run OpenCode with a local Ollama model in Canopy Set up a local model route, account for OpenCode V1 and V2 differences, verify context and tool use, then test an offline coding task in Canopy. Share agent work with a team without sharing model accounts A practical boundary map for shared project context, CLI profiles, team collaboration, model access, and billing in Canopy. Share project instructions across Claude Code and Codex Place durable facts, CLI-specific instructions, reusable skills, and MCP connections where each agent can actually use them. Switch from Claude Code to Codex without losing the task A cross-CLI handoff when usage limits, model choice, or review needs change: preserve the goal, branch, evidence, and next action. The coding agent says done, but the app is still broken A recovery workflow: reproduce the failure, inspect local services and logs, point the agent at evidence, and verify the new diff. Trigger.dev runs stuck in dev? Check for a second agent worker Trace a queued local task across the app, Trigger.dev project, dashboard, and worker processes when two agent worktrees are open. v0 not showing your local agent's GitHub changes? Check the branches Trace a local Canopy agent change through GitHub, v0's base and working branches, Pull Changes, preview deployments, and production permissions. Vercel says the Git author must be a team member: what do I check? Diagnose a Vercel PR preview blocked before its build by checking the commit author, connected Git account, project team, and access request. What to do when an AI agent opens a 40-file pull request Inventory an oversized agent diff, separate independent work from dependencies, and rebuild reviewable PRs without losing the original change. When a coding-agent session suddenly burns through usage Pause the session, separate context growth from repeated work and stale estimates, then restart with a smaller verified scope. Which coding-agent CLIs work in Canopy, and what differs? A dated, source-backed comparison of launch, resume, session signals, shared context, and file attribution across seven CLIs. Which coding-agent session introduced this regression? Trace a reproducible failure through the running build, Git history, pull request, and related agent session before assigning a fix. Why a new agent worktree cannot build: env files and dependencies Diagnose missing .env files, local packages, Python environments, ports, and shared services before blaming the coding agent. Why did Claude Code forget my task after compaction? Recover a Claude Code task after /compact by checking the actual branch, diff, tests, and lost instructions before giving the agent one verified next step. Why did my coding agent change the lockfile for a small fix? Trace an unexpected package-lock or pnpm-lock diff to its command and dependency decision, then review and reproduce the PR safely. Why did my Trigger.dev task send two emails? Trace duplicate email through task run IDs, retry attempts, app triggers, and provider events before replaying or changing idempotency settings. Why do my agent tests pass locally but fail in GitHub Actions? Compare the exact PR commit, failing CI step, runner, install, environment, and test command before changing code or rerunning a job. Why does my AI-built app lose data when I refresh? Find whether a Save button updates only UI state, browser storage, or a real backend, then test the persistence the product actually needs. Why does my browser show the old app after my coding agent changed the code? Trace a stale local preview through the checkout, running server, actual port, hot reload, browser cache, and data request before asking an agent to edit again. Why does my coding agent keep asking for permission? Diagnose repeated coding-agent approval prompts by checking the action, CLI rules, working directory, and session before changing permissions. Why does my coding agent show so many sent tokens? Understand large input totals, growing conversation context, cache reads and writes, output tokens, and the checks to make before changing models. Why won’t my dev server start in Canopy? Trace a failed website, API, or worker launch through its checkout, command, first error, port, dependency, and observed URL. Why won’t my phone open my local dev site? Use the computer's LAN address, check the dev server's bind host and port, then trace API calls and HTTPS-only browser features on the phone. Work on a Lovable app locally in Canopy without losing Git sync A branch, environment, preview, and publishing checklist for using local coding agents on a Lovable project connected to GitHub.