CLI that orchestrates Parley Deck multi-agent cooperation workflows.
parley orchestrates multi-agent Parley Deck cooperation across local CLI agents: risk-tiered tracks (fast, standard, deliberation), deliberation rounds, consensus and finalize, implementation with living execution plans, a live TUI, and a human-braked outer loop. v1.36.0 adds self-documenting agent display names (family_model_effort, e.g. codex_gpt-5.6-sol_xHigh), fixes the roster-identity vs adapter split with a fail-closed participant resolver so a deck whose roster is claude-1 runs end-to-end, adds parley roster show and parley roster init to view and (re)build the roster and its family mapping, makes autonomous (yolo) write mode first-class per agent, and makes fast the standard speed on a separate axis from reasoning effort (same model and effort, faster output).
