[Refactor] Rename the --slot flag to --model

[Yunaik]

Why

slot had two meanings: the position, the model argument of an openJiuwen agent ("the model slot"), and the occupant (--slot jev, "the llm slot", "both slots"). A reader meets the second reading first, in every command. In English a slot is the position. The README had both in one paragraph: "run one agent on both slots" and, seven lines later, "the chat model in the same slot".

How

The occupant is called the model. --model jev|laya|cua|llm|random|rule is the flag on every agent, on decide and on probe. The help text says that llm is the chat model in MODEL_NAME. "Model slot" stays for the position, where the architecture docs define it. "Slot model" stays for the Model subclass that fills it (ToolDecisionModel, BrowserDecisionModel).

What

• CLI: --slot becomes --model on run, decide, probe and the rails. The MCP server's list_agents flags and the skills follow.
• Code: SLOTS and its siblings become *_MODEL_NAMES, the string parameter is model_name, Trial.slot is Trial.model, slot_label is model_label.
• Records: the results table's column, the replay page's data and CSS, the profiler JSON and a browser run's answer.json use model. The replay reads the slot key of records written by 0.1.0, with a test. The two tracked Allrecipes records use the new key.
• Kept: the random baseline's seed string random-slot-{seed}. A new string would change every seeded run's draws. The 0.1.0 changelog text. "Decision-policy slot" as the name of the upstream seam.
• Not added: a --slot alias. argparse rejects the old flag with unrecognized arguments: --slot. A skill installed before this change needs npx skills add again.

Verification

• uv run ruff format --check ., uv run ruff check ., uv run ty check: pass.
• uv run pytest -q: 487 passed, 36 skipped. The new case is tests/test_replay.py::TestBrowserRun::test_a_run_written_by_0_1_0_names_the_model_under_slot.
• scripts/smoke.sh: ok.
• To try, without a key: uv run s1a run ticket_router --model random --rethink off --episodes 1 prints the summary.
• uv run python -m evals.table evals/results shows the model column.
• uv run python -m evals.replay docs/results/allrecipes/jev docs/results/allrecipes/llm --out /tmp/replay renders the migrated records.

🤖 Generated with Claude Code