fix(slack): tighten strict mention routing

[Claude-Kimn]

What does this PR do?

Improves Slack channel/thread routing for strict mention mode and nudges non-tool empty model responses to produce a short visible answer before giving up.

This carries the AI-5142 production hotfixes onto the current main branch so deployments can be reproduced from source instead of a server-only runtime branch.

Related Issue

Fixes #8019

Type of Change

• 🐛 Bug fix (non-breaking change that fixes an issue)
• ✨ New feature (non-breaking change that adds functionality)
• 🔒 Security fix
• 📝 Documentation update
• ✅ Tests (adding or improving test coverage)
• ♻️ Refactor (no behavior change)
• 🎯 New skill (bundled or hub)

Changes Made

• gateway/platforms/slack.py
  • allow configured plain-text mention_patterns to count as explicit Slack calls, including in strict_mention mode
  • keep strict mode from auto-engaging on prior thread memory/session presence alone
  • allow only narrow same-user direct follow-up after this bot's immediately preceding reply
  • ignore strict follow-up messages that mention a different Slack user
• gateway/config.py
  • bridge top-level Slack mention_patterns into platform extra config/env
• run_agent.py
  • nudge non-tool empty responses to produce a short visible answer before giving up
• hermes_cli/commands.py
  • route Slack native slash aliases such as /hermes-status to Hermes commands
• tests/run_agent/test_run_agent.py
  • cover non-tool empty response retry behavior

How to Test

1. Configure Slack with require_mention: true, strict_mention: true, and mention_patterns for a bot name.
2. In a Slack channel thread, verify that @bot and configured name calls route to the bot.
3. Verify that unrelated thread replies, replies mentioning another user, and old thread/session presence do not auto-engage the bot.
4. Verify that same-user direct follow-up after the bot's immediately previous reply is still routed.
5. Verify that a non-tool empty response receives a nudge asking the model to produce a short visible reply.

Tested:

./venv/bin/python -m pytest tests/gateway/test_slack_mention.py tests/run_agent/test_run_agent.py -k "slack_mention or truly_empty_response"
# 44 passed

Local macOS clone did not have pytest installed; the test command above was run in the Hermes runtime venv on Linux.

Checklist

Code

• I've read the Contributing Guide
• My commit messages follow Conventional Commits (fix(scope):, feat(scope):, etc.)
• I searched for existing PRs to make sure this isn't a duplicate
• My PR contains only changes related to this fix/feature (no unrelated commits)
• I've run pytest tests/ -q and all tests pass
• I've added tests for my changes (required for bug fixes, strongly encouraged for features)
• I've tested on my platform: Linux runtime venv

Documentation & Housekeeping

• I've updated relevant documentation (README, docs/, docstrings) — or N/A
• I've updated cli-config.yaml.example if I added/changed config keys — or N/A
• I've updated CONTRIBUTING.md or AGENTS.md if I changed architecture or workflows — or N/A
• I've considered cross-platform impact (Windows, macOS) per the compatibility guide — or N/A
• I've updated tool descriptions/schemas if I changed tool behavior — or N/A

For New Skills

N/A

Screenshots / Logs

Production symptoms addressed:

• Slack strict-mode bot replied to thread messages due to prior thread/session participation.
• Non-tool empty model responses repeated empty outputs without an explicit recovery instruction.