# FND007-LAB1 — learner guide

Purpose: recover a bounded local repository state and produce **Project State Handoff Record v1** without relying on the README or narrative memory.

Requirements: Python 3 and Git available locally. No network, credentials, destructive reset, or production repository is required.

1. Run `python setup_fixture.py`. It creates a fresh synthetic repository under `workspace/` and prior evidence under `evidence/`.
2. Before cleaning, staging, committing, or deleting anything, run `python inspect_state.py` and capture the initial observation.
3. Read `issue_brief.json`, inspect the visible files/deltas, and compare current state with `evidence/prior_verification.json`.
4. Fill `handoff_record.csv`. State class and intent class are separate fields; do not infer intent from staged/untracked status.
5. Record one bounded human decision: ACCEPT, MODIFY, REJECT, or BOUNDED_DEFER. Include at least one non-claim and any material AI/support disclosure.
6. Give the record to another learner/reviewer for recovery. Exact prose or command sequence is not required; reconstructed state and decision boundaries are.

`python run_lab.py` only checks fixture integrity. It does not grade or reveal the canonical handoff classification.