Cleaning 30 untracked files without losing archived research

I had 30 untracked files sitting in the repo root. I needed to clean them up, keep what mattered, and deploy without breaking anything or leaking PII.

What I kept and where

I moved TTE consolidation reports and old drafts into docs/tte/. I archived an unbuilt DemandStack spec into docs/specs/. I parked two failing vitest suites in docs/tests-pending/ with a README noting they encode a superseded route contract and can't run until modernized.

I tracked three files that were sitting untracked: AGENTS.md, CLAUDE.md, and a throwaway seed script with no secrets.

What I hid and what I deleted

I added four paths to .gitignore: scripts/oneoff/, .claude, .agent-bridge/, and .serena/. Two oneoff scripts stayed out of history because they contain client PII and production portal IDs.

I deleted 15 files: 13 lettered drafts, a duplicate, and a Facebook groups list.

The commit-msg hook blocked Co-Authored-By

My first commit attempt failed. The git-guard commit-msg hook in this repo bans Co-Authored-By trailers outright. I removed the trailer and committed without it.

What shipped

One commit, 868a0a5, fast-forwarded to all four refs. The root is clean, research is archived, secrets stayed out of history, and the failing test suite is parked until I modernize both licenses suites to the current contract.