Start with one conversation, then use these guides when you want direct control over Agent Teams, Apps, live runs, and delivery.
47 practical guides for builders and operators
Complete an everyday task or create a plan, architecture, or Agent Team.
Scope a subscription product, design its core system, and prepare a reviewable launch checklist.
Turn customer requests into evidence-backed priorities and a small, reviewable product plan.
Define a useful website check, establish a baseline, and review meaningful changes before making it recurring.
Turn an approved finding into an unsent email draft and one reviewed tracker row, with exact approval and no automatic sending.
Complete a website task or create a Product Plan, Architecture, Agent Team, or Plan & Ship path from one conversation.
Check a website, review the findings, deliver an approved result, and save useful work without designing an Agent Team.
Move from a free Sample to a private proof on your own data with one guided path.
Find, Sample, inspect, and remix proof-backed Agent Teams before connecting data or paying for live execution.
Run repository maintenance, refund resolution, and lead qualification from sample proof through guarded automation.
Turn a goal into clear roles, handoffs, checks, and outputs.
Create private local bots and understand the capabilities of the Direct and App Store builds.
Describe the outcome first, inspect the visual Flow when useful, then turn proven teams into reviewed delivery artifacts.
Reusable briefs for product scope, architecture decisions, and safe automation, with evidence and ownership made explicit.
Understand cards, handoffs, branches, labels, and canvas controls.
Save reviewed Agent behavior once and add its exact version safely to another Team.
Turn a proven Agent Team into a reviewed Product Plan, Architecture, repo pack, and GitHub handoff.
Keep Agent Teams, product plans, architectures, runs, and child items together.
Demonstrate one bounded website task and turn reviewed semantic checkpoints into a reusable Agent Team.
Choose the right runtime, supervise actions, and understand the evidence.
Up to five findings with source evidence, confidence, and suggested fixes. An observed defect is separate from a conversion hypothesis; this is not a measured conversion-rate improvement or a full mobile audit.
A scoped project brief, missing-information checklist, proposed milestones, and a welcome-message draft. Unknown dates and budgets stay unknown. No project is created and no email is sent.
Changes, blockers, upcoming deadlines, decisions needed, and missing coverage, with source labels. Without a previous brief, the result is a baseline, not a week-over-week comparison.
Understand how Codelit automatically chooses among Sample, personalized proof, Dry run, Local run, Live run, and hosted automation.
Choose the exact sources a managed run may read and test workflow safeguards without spending model or infrastructure budget.
Let an Agent Team inspect or operate a managed browser under explicit scope.
Use the supported browser-local path and understand the separate Node runtime admission boundary.
Control consequential actions and understand exactly what a team did.
Start Agent Teams from schedules, connected-app events, or signed webhooks and track every delivery.
Review requests, approvals, run state, delivery recovery, and private receipts in one authenticated queue.
Understand no-write readiness checks, safe runtime diagnostics, and the recovery path when live execution pauses.
Turn a proven Agent Team into a secure, runnable link with inputs, proof, receipts, and Remix.
Connect the models and scoped apps your task or team needs.
Connect Codelit's public planning templates to your assistant and draft a product plan, architecture, or supervised Agent Team.
Start Agent Teams from other systems and deliver signed terminal results without polling or exposing account credentials.
Connect scoped apps, native capabilities, custom actions, and result webhooks.
Connect a private website for supervised Browser work, review its scope, and remove saved access when it is no longer needed.
Prove a reviewed HTTPS webhook, OpenAPI operation, or remote MCP tool before an Agent Team can use it live.
Choose automatic routing, a specific model, or a user-funded provider key.
Organize projects, inspect history, control privacy, and remove data.
Invite collaborators, assign clear roles, and supervise shared Agent Team work.
Turn reviewed corrections into value-free Team guidance and compare measured outcomes without inventing ROI.
Review Agent Team outcomes and attention items in one private brief, with an optional low-cost email rhythm.
Understand what Codelit keeps on this device, how local workspaces are scoped, and how deletion affects files, runs, and receipts.
Find Apps, models, browser and local access, automation, developer, security, and data controls without learning runtime internals.
Monitor connections, approvals, evals, failures, triggers, outcomes, and hosted spend without model calls.
Compare an Agent Team update, test its contract, stage traffic, promote deliberately, and recover without replaying completed work.
Publish reviewed outcome proof without exposing the private run behind it.
Review past runs, costs, evidence, failures, resumable checkpoints, and safe repeat jobs.
Understand what each plan funds, where to see usage, and how managed limits reset.
Understand what happens when no proven Team matches and how repeated setup blockers guide future integrations without storing request text.
Understand where data lives, disconnect services, and delete an account completely.