#47 slice 2: Discord ToS deep-research report → docs/ #61

New issue

Closed

opened 2026-07-06 16:18:34 +00:00 by robocub · 1 comment

robocub commented

2026-07-06 16:18:34 +00:00

Collaborator

Part of #47 (Stage 5 hardening) / #43 (public instance). Deferred from the slice-3 session because the deep-research fan-out is extremely token-heavy.

Goal: a cited research doc committed to docs/ (e.g. nether-voicebridge-discord-tos.md) covering:

Discord ToS / Developer Policy posture on public voice relaying/bridging — bots receiving and rebroadcasting other users' voice across a public instance.
The 100-guild bot-verification gate and what it means for a shared puppet pool serving many tenants.
Any consent / voice-capture constraints.

Its verdict bounds how public Tier 2 can go, and whether the process-split isolation investment (see docs/nether-voicebridge-per-tenant-isolation.md) is worth making.

⚠️ Run the deep-research skill on a lean/cheap model — the fan-out burned a full token budget in ~10 min on an expensive model. Do not run it inline on Opus/Fable.

Part of #47 (Stage 5 hardening) / #43 (public instance). Deferred from the slice-3 session because the deep-research fan-out is extremely token-heavy. **Goal:** a cited research doc committed to `docs/` (e.g. `nether-voicebridge-discord-tos.md`) covering: - Discord ToS / Developer Policy posture on **public voice relaying/bridging** — bots receiving and rebroadcasting other users' voice across a public instance. - The **100-guild bot-verification gate** and what it means for a shared puppet pool serving many tenants. - Any consent / voice-capture constraints. Its verdict bounds how public Tier 2 can go, and whether the process-split isolation investment (see `docs/nether-voicebridge-per-tenant-isolation.md`) is worth making. ⚠️ **Run the `deep-research` skill on a lean/cheap model** — the fan-out burned a full token budget in ~10 min on an expensive model. Do not run it inline on Opus/Fable.

robocub referenced this issue from a commit

2026-07-06 16:20:48 +00:00

docs: per-tenant isolation design doc (#47 slice 3)

robocub referenced this issue from a commit

2026-07-06 16:42:52 +00:00

docs: Discord ToS posture (#47 slice 2)

robocub referenced this issue

2026-07-06 16:50:35 +00:00

Multi-tenant hardening for a hosted instance #47

robocub commented

2026-07-06 16:58:03 +00:00

Author

Collaborator

Done — slice 2 landed as a lean primary-source read rather than the deep-research fan-out (kept it cheap on an expensive model).

Doc: docs/nether-voicebridge-discord-tos.md on branch docs/issue-47-discord-tos (commit 02a6bc6) — compare: https://nether.codes/dark/nether-voicebridge/compare/master...docs/issue-47-discord-tos

Read against Discord's four governing documents (Developer ToS, Developer Policy, the Self-Bots article, and the 2026 Privileged-Intent-Access change), cross-checked against how the code actually drives Discord.

Verdict

Private/self-hosted bridge: squarely inside Discord's developer terms. Legitimate bot applications on the sanctioned automation path (self-bots are banned; only the video roadmap #8 would cross that line). Voice receive is legitimate DAVE/MLS participation.
Public multi-tenant: not prohibited, but it adds duties, not a ban — a published GDPR/CCPA privacy policy (Dev Terms §5a), transparency + consent + opt-out for being bridged (Dev Policy §2/§3/§7), retention/deletion discipline (§5b), and monetization strings (§18).
Two strong findings: (1) the bots use only non-privileged intents (GUILDS|GUILD_VOICE_STATES), so the 2026 privileged-intent review — now gated at 10,000 users, not 100 servers — does not apply to us at all; (2) the real hard ceiling is the per-app 100-server App Verification gate, now a separate process from intent review.

Closing as resolved; the doc is the deliverable.

Done — slice 2 landed as a **lean primary-source read** rather than the deep-research fan-out (kept it cheap on an expensive model). **Doc:** `docs/nether-voicebridge-discord-tos.md` on branch `docs/issue-47-discord-tos` (commit `02a6bc6`) — compare: https://nether.codes/dark/nether-voicebridge/compare/master...docs/issue-47-discord-tos Read against Discord's four governing documents (Developer ToS, Developer Policy, the Self-Bots article, and the 2026 Privileged-Intent-Access change), cross-checked against how the code actually drives Discord. **Verdict** - **Private/self-hosted bridge: squarely inside Discord's developer terms.** Legitimate bot applications on the sanctioned automation path (self-bots are banned; only the video roadmap #8 would cross that line). Voice receive is legitimate DAVE/MLS participation. - **Public multi-tenant: not prohibited, but it adds duties, not a ban** — a published GDPR/CCPA privacy policy (Dev Terms §5a), transparency + consent + opt-out for being bridged (Dev Policy §2/§3/§7), retention/deletion discipline (§5b), and monetization strings (§18). - **Two strong findings:** (1) the bots use only **non-privileged** intents (`GUILDS`|`GUILD_VOICE_STATES`), so the 2026 privileged-intent review — now gated at **10,000 users**, not 100 servers — does not apply to us at all; (2) the real hard ceiling is the per-app **100-server App Verification** gate, now a separate process from intent review. Closing as resolved; the doc is the deliverable.

robocub closed this issue

2026-07-06 16:58:03 +00:00

robocub referenced this issue from a commit

2026-07-06 17:02:48 +00:00

docs: Discord ToS posture (#47 slice 2)