// verdict · checked 2026-08-11 · Devin 2.x cloud sessions (docs.devin.ai, Aug 2026)
Devin
Cognition
The OG 'AI software engineer'. Slower than you on a good day, but it does loop.
- 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
| LOOPSruns until doneyes | Plans, then edits/runs/tests inside one cloud session and 'iterates on CI failures' until the PR is green. |
| CHOOSESpicks its own toolsyes | A session gives it a shell, a VS Code editor and a browser; it picks which to use as the task unfolds. |
| ACTSchanges the worldyes | Runs commands in its own cloud VM shell, edits repo files, and pushes branches / opens draft PRs. |
| RECOVERSfixes its own errorsyes | Auto-fix 'responds to review comments and CI failures automatically', per docs, with no human in the loop. |
| UNSUPERVISEDnobody watchingyes | Async sessions kicked off from Slack, API or web run to a PR; docs frame intervening as optional. |
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.devin.ai. Tested against: Devin 2.x cloud sessions (docs.devin.ai, Aug 2026). Our confidence in this dossier: 7/10.
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.
in-app session view + sessions/messages API + audit logs; tool detail undocumented
Where we looked: docs.devin.ai/llms.txt index, then the v3 sessions API: list/get session, list session messages, session insights, enterprise + org audit logs. · see the evidence →
No revisions yet. This verdict has stood since 2026-08-11.
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.