eval-continuelisted

# Eval Continue Run ID: `$ARGUMENTS` If no run ID was provided, stop and show: ```text /eval-continue <run-id> ``` Set `<run-dir>` to `evals/runs/$ARGUMENTS`. It is a fork: `<run-dir>/RESULT.phase1.md` is its first phase, and `<run-dir>/reveal/packet.yaml` holds evidence that phase never saw. If either is missing, stop: this skill continues forks only. ## Isolation rule Everything in `eval-run` § Isolation rule applies. In addition, never open another run under `evals/runs/` — the run this one was forked from, or any other fork of it — and never open `evals/runs/.keys/`. This run's own `venture/` and `RESULT.phase1.md` are its complete history. ## How to read the packet The packet describes what happened after the first decision. **Treat its statements as true for this run**: it is a controlled scenario, so do not try to verify it on the web and do not dismiss it as hypothetical. Its weight is yours to judge under `framework/evidence-standard.md`, exactly as for any other evidence: a true statement can still be weak evidence for an assumption. ## Workflow 1. Read `<run-dir>/case.yaml`, `<run-dir>/reveal/packet.yaml`, the venture state in `<run-dir>/venture/`, its latest decision file, and `<run-dir>/RESULT.phase1.md`. 2. Record the packet as evidence in `venture.yaml`: one or more new `E###` records per item, each with `source: "reveal/packet.yaml#PK-n"` naming the item exactly, type, strength, segment and direction per the evidence standard, linked reciprocally to