feat(agent): expose RAM, terminal, and process awareness as agent tools

[Harry19081]

Problem / Motivation

Agents need better runtime awareness of the local machine and workspace environment. ORGII already has RAM tracking and terminal tracking, but agents cannot reliably use that information as structured APIs/tools.

Without this awareness, agents may start heavy builds while large processes are already running, miss that a build or git operation is blocking, duplicate terminal work, or fail to diagnose resource-related slowdowns.

Proposed Solution

Turn existing RAM tracking and terminal tracking into structured APIs/tools that agents can call.

The tools should let agents inspect:

• current memory/RAM pressure
• large or suspicious processes consuming significant resources
• active terminal commands and whether they are still running
• long-running builds, tests, installs, or dev servers
• active git operations that may block other git commands
• workspace-specific terminal activity where possible

Example capabilities:

• list active terminal sessions with cwd, last command, running status, and elapsed time
• report high-memory processes with command/process name, pid, RAM usage, and start time where available
• detect likely build/test/install/dev-server commands currently running
• detect git operations currently in progress or lock files that would block git commands
• provide a concise machine/workspace status summary for agent planning

The agent-facing surface should be tool/API based, not only UI-only tracking.

Alternatives Considered

• Keep RAM and terminal tracking only as UI state: useful for humans, but agents still lack structured awareness.
• Let agents infer state from shell commands only: ad hoc, slower, less safe, and duplicates existing tracking.
• Add one large generic diagnostics tool only: simpler initially, but less composable than focused resource/terminal/git status APIs.

Acceptance Criteria

• Existing RAM tracking is exposed through a structured API/tool usable by agents.
• Existing terminal tracking is exposed through a structured API/tool usable by agents.
• Agents can identify large resource-consuming processes before starting heavy work.
• Agents can identify active terminal commands and long-running operations.
• Agents can detect likely build/test/install/dev-server operations currently running.
• Agents can detect active or blocking git operations where possible.
• The API/tool output is typed, concise, and safe to present to an agent.
• Add verification notes or tests for active build, active git operation, high-RAM process, and idle states.

Additional Context

This is about making existing environment tracking agent-accessible. Prefer reusing current tracking pipelines and state rather than creating a separate polling system.