[FEATURE] Make companion/buddy feedback actionable inline

Preflight Checklist

• [x] I have searched existing requests and this feature hasn't been requested yet
• [x] This is a single feature request (not multiple features)

Problem Statement

Problem Statement:
The companion (buddy) acts as a third-party observer that catches half-baked logic, errors, and weak spots while Claude Code is
working. For someone learning development, it's one of the best feedback loops available.

The problem is there's no way to act on the buddy's feedback directly. My current workflow is: read the buddy's speech bubble,
mentally note or copy the observation, then manually paste it into the Claude Code input to get Claude to address it. That friction
adds up fast, especially when the buddy fires off multiple good catches in a single session. The feedback is right there — I just
can't do anything with it without leaving my flow.

Proposed Solution

Proposed Solution:
A way to send the buddy's comment directly to Claude Code as input. Some options:

• Clicking the buddy's speech bubble injects its text as a new prompt to Claude Code
• A keyboard shortcut that means "address what the buddy just said"
• A small action button on the bubble (e.g., "Send to Claude") that forwards the observation

The buddy already has the right context and is making sharp observations — it just needs a shorter path from observation to action.

Alternative Solutions

Alternative Solutions:
Currently I copy the buddy's comment text, paste it into the Claude Code input, and manually ask Claude to address it. It works but
it's clunky and breaks flow, especially when the buddy catches something mid-task and I want to act on it immediately.

Priority

High - Significant impact on productivity

Feature Category

Interactive mode (TUI)

Use Case Example

Use Case Example:

1. I'm building a new feature in a Next.js project
2. Claude Code is writing a component and the buddy comments "that effect cleanup is missing a dependency — this will cause stale

state on re-render"

1. I read the bubble and think "yes, good catch" — but now I have to copy that text, paste it into the input, and type something like

"hey, address what the buddy said about the effect cleanup"

1. With this feature, I'd click the buddy's comment (or hit a shortcut) and it would go straight to Claude as a prompt
2. This keeps me in flow and makes the buddy's observations immediately useful instead of just informational

Additional Context

Additional Context:
The buddy is particularly valuable for developers who are still building their mental model. It catches things I wouldn't notice on my
own. The gap isn't in the quality of its feedback — it's in the friction between seeing the feedback and acting on it.