docs: two pages that state the opposite of what the code does

[rainmanjam]

Batch B, docs half. The encoder-profile defects (hevc_vaapi, NVENC capped VBR) are being fixed separately.

Both of these are verified against source, and both survived because they are stated confidently.

PLATFORMS.md said a scope is not requested. It is.

"moderation:ban is deliberately not requested: nothing in polyemesis bans or times out a viewer."

internal/oauth/kick.go:102 requests it. internal/automod/matrix.go:54 carries ActionTimeout and ActionBan. The page was wrong twice — it omitted the scope from its list and explicitly denied it.

This isn't drift, it's a decision the page never caught up with. kick.go:88-94 records the reversal in full:

"moderation:ban was previously omitted, on the grounds that nothing here banned a viewer and that asking a restreamer's audience for the power to do so read as overreach. That was a product decision and the maintainer has reversed it: banning and timing out are implemented, so the scope is requested. The old reasoning is kept rather than deleted because it is the argument to re-read if the decision is ever revisited."

The page kept only the old half — the half the code explicitly labels as superseded.

A false statement about what a consent screen will ask for is the worst class of error on this page: an operator reads it, connects an account, and is shown a permission the documentation promised would not be requested.

Also carried across the fact an operator actually needs: adding a scope later does not upgrade an existing token, it forces everyone to disconnect and reconnect. That's why the list is settled in one go, and it explains the "reconnect once" notice they may already have hit.

HARDWARE.md had the default encoder exactly inverted

"libx264 stays the default even on a machine with a working GPU. … Hardware is an opt-in you make deliberately."

Tools.DefaultVideoEncoder (internal/ffmpeg/detect.go:415) walks encoderPreference — videotoolbox, nvenc, qsv, vaapi, amf, x264 — and returns the first that passed the probe. Hardware wins, and the function's own comment says why:

"a machine with a usable GPU that silently software-encodes cannot serve the feature it was bought for"

RENDITIONS.md:295 has always had this right, so the two pages contradicted each other and a reader had no way to tell which to believe.

Rewritten to match the code, including the every-probe-failed fallback (still libx264, because an empty -c:v is neither usable nor legible). Kept the true half of the old text — libx264 is identical everywhere and is a legitimate choice — reframed as the override it is rather than the default it isn't.

Verification

• grep -c '"moderation:ban"' internal/oauth/kick.go → 1
• grep -rn "opt-in you make deliberately" docs/ → 0 remaining
• HARDWARE.md and RENDITIONS.md now agree

Docs-only, so this also exercises the #357 path-filter fix again.

https://claude.ai/code/session_01HeLrWaDmsNeeNSbHQfEofX

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[Copilot]

Pull request overview

Updates two documentation pages to match the current behavior of the codebase: (1) the Kick OAuth scopes actually requested (including moderation:ban) and the operational impact of adding scopes later, and (2) the actual default video encoder selection behavior (hardware-first when probes succeed, with libx264 as fallback).

Changes:

• Corrected PLATFORMS.md to list moderation:ban as requested on Kick and documented the historical rationale + reversal, plus the reconnect requirement when scopes change.
• Corrected HARDWARE.md to reflect the code’s hardware-first default encoder selection order and the “all probes failed” fallback behavior.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
docs/PLATFORMS.md	Fixes incorrect claim about moderation:ban not being requested; documents scope-change reconnect implications.
docs/HARDWARE.md	Fixes inverted default encoder behavior; documents hardware preference order and fallback to libx264.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.