This practice completed bilingual source review and public content publication, but did not install AgentLimb, complete a real product task, or demonstrate new user impact. The three conclusions are recorded separately so “article published” never becomes “product validated.”
Read the boundary first
This case proves the AgentLimb profile was publicly released. It does not prove BitShovel tested AgentLimb hands-on or that the release created new real users.
01 / LAYERED VERDICT
Keeping all three states together prevents “live” from silently becoming “the product works” or “growth happened.”
The bilingual AgentLimb profile is published on a public BitShovel page and appears in the public content feed.
Next evidence
Future content updates should retain sources, verification dates, and change summaries.
This was a read-only review of public material. We did not install the Chrome extension, start the local bridge, grant browser permissions, or run a task in a real account.
Next evidence
Complete one fixed task in an isolated browser profile and low-risk test account, preserving permissions, steps, page outcomes, and failures.
There is currently no evidence attributing new visits, outbound clicks, feedback, or retention to the AgentLimb profile.
Next evidence
Exclude internal QA traffic first, then build an attributable same-period funnel from the problem page to the profile, outbound link, and feedback.
02 / VERIFICATION PATH
Every step points to inspectable evidence; local, shadow, production, and public states do not substitute for one another.
Read the AgentLimb site, official GitHub repository, public store listing, and privacy material, separating maker-presented capabilities from facts BitShovel personally verified.
The public page retains maker sources, image attribution, and the verification date, and explicitly states what was not installed, authorized, or reproduced.
Treat content publication, hands-on product validation, and user benefit as separate states; only the first is marked complete today.
03 / PUBLIC EVIDENCE
BitShovel release records are evidence of our own execution, not independent third-party validation.
The public page records sources, trial boundaries, incomplete checks, evidence images, and the verification date.
Open public sourcePublic content-distribution record used to confirm publication, not to infer readership, conversion, or user impact.
Open public sourceMaker README, architecture description, and demo material used for content review, not a BitShovel install or real-task test.
Open public sourceContent publication: completed
The bilingual AgentLimb profile is published on BitShovel's public website.
Hands-on product validation: not completed
BitShovel's current AgentLimb profile is not the result of an install or real-task product test.
User impact: not completed
Any new real-user impact caused by the AgentLimb profile is currently unknown.
04 / CHANGE HISTORY
The public update log records boundary changes across publication, hands-on testing, and user impact; it does not yet provide full prior versions or automatic field diffs.
The work is split into content publication, hands-on product validation, and user impact; only publication is marked complete.
05 / LIMITS & NEXT
Return this practice to the full problem journey
The AI says it is “done.” How do I know it is actually deliverable?
