# RDLC Enforcement

## Before Every Research Task
1. Read `RDLC.md` first when it exists; it is the repo-local research contract.
2. State the claim, audience, source plan, and confidence before drafting.
3. Use `gpt-5.5` with `xhigh` reasoning for this wizard repo.
4. Keep this repo on `maximum` (`gpt-5.5` `xhigh` throughout); do not switch codex-rdlc-wizard work to `mixed`, mini-only, or lower-reasoning profiles because this wizard work is meta and high-blast-radius.
5. If `GOALS.md` exists, treat it as the active-scope contract and keep `ROADMAP.md` as backlog/history.

## Research Workflow
1. Define the claim or hypothesis before gathering evidence.
2. Prefer primary sources and authoritative databases over secondary summaries.
3. Label every substantive claim: `VERIFIED`, `SUPPORTED`, `INFERRED`, or `UNVERIFIED`/`GAP` as configured in `RDLC.md`.
4. Cite the source at first mention in each file.
5. Run the slop scan and fact regression suite before commit.
6. For high-stakes deliverables, run self-review and Codex-native review with `xhigh` reasoning.

## Codex Shape
- `skills = explicit workflow layer`
- `hooks = silent event enforcement`
- `repo docs = source of local truth`

Use `$rdlc` for research/fact/deliverable work. Do not pretend Codex has a native `/rdlc` slash command.

## Commit Gate
- No commit without fresh verification evidence.
- No push without reviewed proof.
- Every defect found in research or review becomes a permanent regression assertion.