hermes - 💡(How to fix) Fix [feature] Interpret Telegram emoji reactions as agent feedback

Problem or Use Case

When I use Hermes via Telegram, I frequently react to the agent's messages with emojis like 👍 (thumbs up) to silently confirm "looks good, move on", 👎 (thumbs down) for "that's not right, try again", or ❤️ (heart) for "nice work, keep this pattern".

Currently these reactions are invisible to the agent — Telegram handles them as message metadata and they never reach the conversation loop. I have to type a full text message for every acknowledgment, which breaks flow especially on mobile.

Proposed Solution

The Telegram gateway adapter should capture emoji reactions on agent-sent messages and inject them into the conversation as a system-level event, e.g.:

[User reacted with 👍 to message ID <id>]

The agent can then infer intent from the reaction — acknowledgment, rejection, approval, etc. — without the user needing to type.

Considerations to account for:

• Which reactions to capture? All of them, or a curated set?
• Delivery model: Should the reaction be appended to the conversation automatically, or notified via a tool call?
• Rate limiting: A flurry of reactions shouldn't flood the context — debouncing or dedup may be needed
• Configurability: The user should be able to opt out of reaction capture if they prefer silent operation

Alternatives Considered

• A webhook or callback system for reactions — overengineered for this use case
• A custom bot command like /react — adds friction, defeats the purpose of one-tap reactions
• Ignoring reactions entirely and requiring text-only communication (current behavior, which is the status quo I'd like to improve)

Feature Type

Gateway / messaging improvement

Scope

Medium (few files, < 300 lines)

Contribution

• I'd like to implement this myself and submit a PR

Debug Report (optional)