# RDLC Enforcement

## Project Context

- **Language/tooling:** {{LANGUAGE}}
- **Research/source path:** {{SOURCE_DIR}}
- **Regression path:** {{TEST_DIR}}
- **Domain:** {{DOMAIN}}
- **Repo shape:** {{REPO_SHAPE}}
- **Setup confidence:** {{SETUP_CONFIDENCE}}

Read `RDLC.md` before research, fact-checking, drafting, or generated deliverable work. Read `TESTING.md` for regression and proof guidance. Read `ARCHITECTURE.md` for generator, source layout, and infrastructure context. If `GOALS.md` exists, read it before `ROADMAP.md` and treat it as the active-scope contract.

## Honest Codex Shape

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

Use `$rdlc` for the visible workflow contract. Do not pretend Codex has native slash commands when it does not.

## Before Every Research Task

1. State the claim, audience, source plan, and confidence.
2. Prefer the highest available source tier.
3. Cite sources at first mention in each file.
4. Label claims as `VERIFIED`, `SUPPORTED`, `INFERRED`, or `UNVERIFIED`/`GAP`.
5. Preserve audience boundaries between private, internal, and public deliverables.
6. Run slop scan, fact regressions, and review before commit.

## Commands

| Action | Command |
|--------|---------|
| Slop scan | `bash scripts/slop_scan.sh` |
| Fact regressions | `bash scripts/regression_test.sh` |
| Generate deliverables | `python3 scripts/generate_deliverable.py` |
| Code tests, if applicable | {{TEST_COMMAND}} |
| Lint, if applicable | {{LINT_COMMAND}} |
| Build, if applicable | {{BUILD_COMMAND}} |

If any command above shows N/A, figure out the right command for this project and update this table.

## Setup Status

- Known: {{SETUP_KNOWN_SUMMARY}}
- Unresolved: {{SETUP_UNRESOLVED_SUMMARY}}

## Model Profile

- Selected profile: {{MODEL_PROFILE}}
- `mixed`: `gpt-5.4-mini` for the main pass plus `gpt-5.5` at `xhigh` for review.
- `maximum`: `gpt-5.5` at `xhigh` for the whole slice.
- If confidence is below `95%`, research more first. If it still stays below `95%`, escalate review to `xhigh`.

## Feedback and Repo Focus

- If you uncover a proven reusable wizard lesson, prefer filing a direct GitHub issue instead of leaving a loose note.
- Keep the active session focused on the product repo.
- Only switch to live wizard work when the product repo is actually blocked.

## Self-Improving

If you discover something about this project that these docs do not cover, update the relevant doc. These files are living documents, not static templates.