Versions:
Parley Deck CLI, published by Feci, is a command-line tool in the multi-agent orchestration and AI-assisted software development category. Its purpose is to coordinate Parley Deck multi-agent cooperation workflows across locally installed AI agent CLIs, running structured design, implementation, and review rounds while preserving the cooperation trail in a repository-local directory. Typical use cases include teams or individuals who drive several local CLI agents through iterative design proposals, implementation passes, and review cycles, and who need a durable, inspectable record of how agent collaboration produced each artifact. The software is currently at version 1.47.0, with 25 versions published in total. This release changes the cooperation protocol so that, before anything is designed, agents are required to determine what already exists. Every round-1 artifact must now carry an "Existing alternatives" section enumerating the mechanisms a proposal builds by hand and, for each, naming what the toolchain, standard library, dependencies, or platform already provides. A scoped null result is permitted but must identify the sources consulted, on the principle that a recorded search can be challenged while an unexamined absence cannot. This requirement is deliberate rather than stylistic: an open-ended instruction to consider alternatives is measured not to work, whereas naming the specific items to check does. The rule is embedded in the round-1 prompt and enforced by a runtime validator, so a bare heading or passing prose mention does not satisfy it. Design rounds previously had no artifact validator at all, unlike review rounds, allowing an agent process to exit successfully having written nothing; they now have one. The conditional correlated-agreement section it replaces excluded the mechanically decidable work it was most needed for, whereas the new requirement is unconditional, binds every track, and is smaller than the text it replaced.
Tags: