Fixed

The documentation website's profile pages described the wrong format. The profile-schema reference and the writing-a-profile guide documented and demonstrated the profile as TOML, a format the tool stopped accepting when the profile format moved to JSON; both pages are now JSON, and their examples carry the schema's required top-level keys (schema, kind, fidelity) and validate against the published schema. Command examples name a .json profile path rather than .toml.

The verbatim game slug eso no longer appears throughout the docs as if it were a shipped profile. One concrete slug is kept, as the example value under the CLI --profile reference; every other occurrence is a typed placeholder (<game-id>, <client>.exe, <profile>.json), and the sentence apologizing that the example was illustrative and not a shipped profile is gone.

The two wide Mermaid diagrams on the Architecture page (the dependency model and the runtime data flow) are laid out top to bottom instead of left to right, so they fit the content column and are legible at normal widths instead of running off the side.

The site footer no longer detaches on documentation pages. It rendered once at the body level, a sibling of the fumadocs docs layout whose grid forces a full viewport of height, which parked the footer a full viewport below the content on every docs page. The footer now renders in flow: the home group renders it after its layout, and docs pages render it inside the docs content column after the body, so it sits directly under the content, exactly one footer per page, with the home page unchanged.

The fragcap.exe binary now carries its real version. Its Windows PE version resource was never stamped, so Get-Command fragcap (and Explorer file properties, and inventory tools) reported a FileVersion of 0.0.0.0 even though fragcap --version printed the true version (issue #104). The build now embeds a VERSIONINFO resource stamped from the crate version, so the FileVersion and ProductVersion match fragcap --version and track releases automatically. The version is single-sourced from the workspace version, so the two can never disagree. This is a Windows build-time change only; no runtime behavior changes.

fragcap doctor now lists the real capture-capable network interfaces, naming each adapter beside its address, instead of always reporting that none were found. The empty-set warning appears only when enumeration genuinely finds no interfaces, and an enumeration that fails is reported as a failure rather than presented as a successfully observed empty machine.

fragcap doctor now reports loopback capture support from the actual loopback adapter rather than an unrelated driver file, so a machine that has loopback support installed is no longer told it is missing; when the state cannot be determined the report says so rather than claiming loopback is absent.