// verdict · checked 2026-08-11 · Lovable Build mode (formerly Agent mode)
Lovable
Lovable
Lovable is an agent. It passes 5 of the five criteria in our AI agent test. This is a dated assessment of the shipped product, not a claim about every possible setup.
Ships you a CRUD app with confidence. The confidence is the product.
- CAPABILITY
- 5/5 AGENT
- PROOF
- OPERATOR INSPECTABLEYou can run it and read the steps yourself — the record exists for whoever runs it
- TRACE ACCESS
- OPERATOR-OWNEDWhoever runs it holds the record; nothing is published
- LEASH
- Not measuredno execution record scored — measure it yourself
- COMMUNITY
- Loading votes…cast yours on the full list
| LOOPSruns until doneyes | Build mode inspects logs, runtime output and network activity and iterates on fixes until the issue is resolved or clarified; step count is set at runtime by findings, not a fixed plan. |
| CHOOSESpicks its own toolsyes | Docs state 'Lovable decides when subagents can help; you do not need to choose when they run', and it selects codebase search, on-demand file reads and log inspection mid-run. |
| ACTSchanges the worldyes | Changes are applied directly to project files and surfaced as diffs; the agent writes code, runs tools, verifies results, and Publish deploys a live hosted app. |
| RECOVERSfixes its own errorsyes | Docs: Build mode observes build errors/test failures, inspects logs, runtime output and network activity itself, reproduces issues, and iterates on fixes mid-task without the user feeding errors in. |
| UNSUPERVISEDnobody watchingyes | Build mode handles execution end to end with minimal supervision: review comes after the run, not per step, and the stop button is an interrupt rather than an approval gate. |
Capability decisions combine documented behaviour and execution evidence. The proof label on each row shows which of the two decided it, and Leash is reported separately, only when it was measured from an execution record. Primary source: docs.lovable.dev. Tested against: Lovable Build mode (formerly Agent mode). Our confidence in this dossier: 8/10.
Can you see it work?
the weakest of the five grades aboveOPERATOR INSPECTABLE — You can run it and read the steps yourself — the record exists for whoever runs it. Trace access: OPERATOR-OWNED — Whoever runs it holds the record; nothing is published.
Details view lists each step, tools used, files, screenshots and results; no export.
Where we looked: docs.lovable.dev/features: agent-mode, subagents (activity card: files inspected, searches, tools, findings), browser-testing (steps, URLs, results, screenshots), audit-logs; 3rd-party chat exporters · see the evidence → · re-checked by a second pass built to overturn it
-
VERDICT OVERTURNED
UNSUPERVISED no → yes filed by editorial internal audit · none — filed by the editors against their own verdict · 2026-08-11
Build mode "handles execution end to end with minimal supervision" and the docs place review after the run, not at each step. Rule 05 turns on capability, and the stop button is an interrupt, not an approval gate. The old NO cited the mode working and answered no anyway.
read the case →
- 2026-08-113/5 WRAPPER WITH AMBITION → 4/5 AGENTRECOVERS corrected to yes: Docs: Build mode observes build errors/test failures, inspects logs, runtime output and network activity itself, reproduces issues, and iterates on fixes mid-task without the user feeding errors in.editorial correction — no case was filed
- 2026-08-112/5 WRAPPER WITH AMBITION → 3/5 WRAPPER WITH AMBITIONCHOOSES corrected to yes: Build mode docs: agent searches codebase, reads files on demand, fetches web, inspects logs, and spawns subagents chosen at runtime.editorial correction — no case was filed
- 2026-08-114/5 AGENT → 5/5 AGENTUNSUPERVISED corrected to yes: Build mode "handles execution end to end with minimal supervision" — review after the run, not per step. Rule 05 turns on capability, and the old NO cited the mode working.Verdict overturned — case #0004 →
Badges
earned, never for sale
These render from the live dataset, so a badge cannot outlive the verdict it claims. If the verdict changes the badge changes with it, and an unearned one returns 409 rather than an image. No sponsor can buy one, at any price.
Think this is wrong?
bring evidence, not opinionName the criterion and link a trace, a doc, or a recorded run that shows the behaviour. A verdict changed by evidence is the best thing that can happen to this site, and the change goes on the public record with your reason attached. Vendors are welcome, and a vendor's own filing is labelled as such rather than buried.