Where Public Prompts End and Private Source Material Begins

Public pages contain intentionally published or editorially selected material; private workspace prompts and uploaded sources do not become Plaza entries merely because they were processed.

Separate public content from private work

Lingmo has pages intended for search and browsing as well as an authenticated private workspace. Public templates, tutorials, and the Prompt Plaza support discovery. The workspace holds personal prompts, uploads, generations, and versions. Shared visual components do not make them one content pool. The product boundary presented at https://lingmofang.com/ is clear: private workspace prompts are not automatically published.

“Not automatically public” does not mean every sensitive source should be uploaded. Users should still minimize data, supply only what the task requires, and remove unnecessary identities, credentials, and internal information before processing.

What belongs on a public page

Public material should result from an explicit act, such as an author deliberately publishing to the Plaza or editors creating a template or tutorial. Saving, generating, comparing, versioning, and restoring are not publication actions. A private adaptation of a public template does not inherit public visibility. Visitors should see the published copy, not the private attachments or draft chain behind it.

Before sharing, create a separate redacted copy. Replace customer names with variables; remove real email addresses, tokens, internal URLs, unreleased metrics, and source excerpts that cannot be distributed. Check example output for information copied from private materials. Explain the intended scenario and the need for human review. A prompt whose value depends entirely on secret context may not be suitable for publication.

Uploaded sources require additional judgment

Images and videos can contain faces, locations, notification banners, contract text, or unreleased products. Crop and redact them even when the immediate goal is only visual reverse engineering. Text documents may contain hidden columns, comments, or credentials overlooked during copying. Users need authority to process the material and should review third-party model processing, storage, deletion, and legal-retention boundaries in the current privacy policy.

Lingmo extracts prompts and structured observations from images or keyframes; it does not provide downstream image or video generation. Moving the result to another tool creates a new privacy and authorization boundary. Settings in Lingmo should not be assumed to extend automatically to an external platform.

Minimize exposure in team workflows

Teams can use variables, summaries, and approved examples instead of raw customer records. A programming investigation should include a minimal reproduction rather than a production database. Marketing analysis can use an essential aggregate range rather than a list of people. Editorial work can quote only authorized excerpts. If the model lacks information, asking a question is safer than uploading an entire archive.

Generated code and recommendations do not automatically modify Git or connect to production. That non-execution boundary reduces accidental operational change, but copied commands and patches still need security review, especially for dependencies, destructive operations, and credential placeholders.

Publication creates lasting responsibility

Other people may copy, alter, and run a public prompt with different models. Publishers should not claim a permanent model list, unlimited free service, or guaranteed outcome. Current models belong to https://lingmofang.com/models, pricing and quota belong to live product pages, membership does not auto-renew, and policy refusals carry no charge. Hard-coding dynamic rules in a template makes it age poorly and can mislead adopters.

If a publication exposes sensitive information, use the product's current management and support routes promptly and assess downstream copies. Removing the source page may not retrieve material that others already saved, which makes pre-publication review essential.

Does generating or saving a private prompt put it in the Plaza?

No. Generation, saving, version history, and multi-model comparison are not publication actions. Only content the user explicitly chooses to publish should enter the Plaza. Even then, make a redacted copy and verify that attachments, customer information, and internal rules are absent from every visible field.