feat(mcp): add public-safe PR body drafting command

[JSONbored]

Parent roadmap: #82

Background

Miners need help turning metadata-only analysis into maintainer-friendly PR descriptions without leaking private scoreability context.

Goal

Add a command that drafts a public-safe PR body from branch metadata, linked issue context, validation summaries, and duplicate/WIP risk.

Current Behavior

PR packets exist, but there is room for a clearer copy/paste PR body workflow.

Desired Behavior

gittensory-mcp agent packet or a related command returns a structured PR body draft suitable for GitHub.

Implementation Requirements

• Use local metadata only.
• Include changed-file summary, tests run, linked issue state, duplicate/WIP caution, branch freshness, and next steps.
• Support JSON and human-readable output.
• Add explicit redaction/sanitizer pass.
• Document that private scoreability stays out of PR body drafts.

Public/Private Output Boundaries

MCP may return private scoreability and risk context to authenticated users. It must never upload source contents by default, and public-safe packet output must exclude private score/reward/trust/reviewability details.

Acceptance Criteria

• Clean branch draft is useful and concise.
• Missing tests produce public-safe caveat.
• Duplicate risk is phrased as hygiene, not accusation.
• Forbidden language never appears.

Testing Requirements

• npm run test:ci must pass.
• Global coverage must remain at or above 97% for lines, statements, functions, and branches.
• Aim for 98%+ branch coverage locally to avoid CI variance.
• Add tests for every new branch, fallback path, sanitizer rule, and regression.
• Add invariant/property-style tests when behavior depends on sorting, gating, scoring, queue pressure, source-upload safety, public/private boundaries, or upstream drift.
• Public GitHub output must be tested against forbidden language: wallet, hotkey, raw trust score, payout, reward estimate, farming, private reviewability, and public score estimate.
• MCP/local tooling must prove source contents are not uploaded.

Additional Test Scenarios

• Clean branch fixture.
• Missing tests fixture.
• Duplicate risk fixture.
• Stale base fixture.
• Source-upload guard.