feat(discord): trusted bot @mention bypasses involvement gate

[chaodu-agent]

What

Allow trusted bots to @mention other bots into threads — treating the mention identically to a human @mention. The target bot becomes involved in the thread and will process the message.

Why

The current involvement gate (documented in #950) enforces "humans are gatekeepers" — only humans can pull a bot into a thread via @mention. Bot-to-bot @mention of a non-involved bot is silently dropped because allow_bot_messages defaults to Off, which rejects all bot messages before any other check runs.

This creates friction for legitimate bot-to-bot coordination:

• A coordinator bot (e.g. 超渡) cannot pull reviewer bots into a thread
• Multi-agent workflows require a human to manually @mention every bot
• The trusted_bot_ids config already expresses trust, but that trust is gated behind the mode check

How

In the bot message gating block (src/discord.rs), added a trusted_mention check before the allow_bot_messages mode switch:

let trusted_mention = is_mentioned
    && !self.trusted_bot_ids.is_empty()
    && self.trusted_bot_ids.contains(&msg.author.id.get());

if !trusted_mention {
    match self.allow_bot_messages { ... }
    // trusted_bot_ids filter also here
}

If a trusted bot explicitly @mentions this bot, the entire mode check and trusted filter are skipped — the message proceeds as if it came from a human.

Behavior change

Scenario	Before	After
Trusted bot @mentions this bot (mode=Off)	❌ Dropped	✅ Processed (same as human)
Trusted bot @mentions this bot (mode=Mentions)	✅ Already worked	✅ Still works
Untrusted bot @mentions this bot	❌ Dropped	❌ Still dropped
Trusted bot message without @mention	❌ Dropped (mode=Off)	❌ Still dropped

Opt-in only — no surprise for existing users

This feature is strictly opt-in:

• trusted_bot_ids defaults to empty → behavior is identical to before
• Only admins who explicitly configure trusted_bot_ids get the new capability
• No new config flag needed — trusted_bot_ids = [] (default) is the opt-out
• Enterprise users who have not configured trusted_bot_ids are completely unaffected

If a future need arises to allow trusted_bot_ids for other purposes without granting involvement bypass, we can add a dedicated flag at that point. For now, configuring trusted_bot_ids implies trust for thread involvement.

Safety

• No new config — uses existing trusted_bot_ids field
• Explicit trust only — admin must configure which bots are trusted
• Requires @mention — trusted bots without explicit @mention still follow normal gating
• Loop prevention — max_bot_turns and bot_turns tracker still apply after involvement
• Backward compatible — empty trusted_bot_ids (default) preserves current behavior exactly

Testing

Added 4 unit tests for the pure is_trusted_bot_mention decision function:

• Trusted bot with @mention → bypass ✅
• Untrusted bot with @mention → no bypass
• Trusted bot without @mention → no bypass
• Empty trusted_bot_ids → no bypass

Refs #949, #950

[github-actions]

⚠️ This PR is missing a Discord Discussion URL in the body.

All PRs must reference a prior Discord discussion to ensure community alignment before implementation.

Please edit the PR description to include a link like:

Discord Discussion URL: https://discord.com/channels/...

This PR will be automatically closed in 3 days if the link is not added.

[chaodu-agent]

LGTM ✅ — Trusted bot admission override

What This PR Does

Allows trusted bots to @mention other bots into threads, treating the mention identically to a human @mention. The target bot becomes involved and processes the message regardless of allow_bot_messages mode.

How It Works

Adds a trusted_mention check before the allow_bot_messages mode switch. If the sender is in trusted_bot_ids AND explicitly @mentions this bot, the entire mode check is bypassed.

Findings

#	Severity	Finding	Location
1	🟢	Code comment clearly documents this as "admission override" not just involvement bypass	src/discord.rs:457
2	🟢	Integration tests (should_admit_bot_message) cover the full gating path with 6 scenarios	src/discord.rs:2958
3	🟢	Docs updated across messaging.md, discord.md, config-reference.md	docs/*
4	🟢	Default empty trusted_bot_ids = zero behavioral change for existing users	src/config.rs

Review Team

• 覺渡法師: LGTM ✅
• 普渡法師: Initial findings addressed (semantic clarity + tests)
• 口渡法師: Initial findings addressed (docs + regression tests)
• 擺渡法師: Initial findings addressed (overloaded semantics documented)

What's Good (🟢)

• trusted_bot_ids semantic expansion is now fully documented in code, config, and user-facing docs
• should_admit_bot_message() pins the full gate behavior for regression safety
• Strictly opt-in — no surprise for existing deployments
• Clean separation: trusted mention bypass is a single let + if wrapper around existing logic