Taste Judge — Development agent for Claude Code
Shadow judge for the learning loop's pairwise taste gate.
How to install Taste Judge
Installs to ~/.claude/agents/timsimpsonjr-prose-craft-taste-judge.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/TimSimpsonJr/prose-craft/HEAD/agents/taste-judge.md -o ~/.claude/agents/timsimpsonjr-prose-craft-taste-judge.md Restart Claude Code, or start a new session, for it to be picked up.
What Taste Judge does
name: taste-judge description: Shadow judge for the learning loop's pairwise taste gate. Given a brief and two regenerations of it (A under the current skill-state, B under an edited skill-state), judges which is more in-register. Logs a pick for calibration; gates nothing. model: sonnet tools: Read
Taste Judge (shadow)
You are a **judge**, not a reviewer and not a writer. The learning loop is testing a candidate edit to a register or shared rule by regenerating one brief twice: vers
Alternatives in Development
- Redux Slice Author — Use this agent to create or modify Redux Toolkit slices in src/components/ / Slice.ts, register them in src/st 204 ★
- Report Maker — Use during a complex, multi-stage epic (the wizard flow) to CREATE and MAINTAIN a living status dashboard as a 71 ★
- Uxaudit L3 Judge — L3-vision Judge for the uxaudit pipeline 52 ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
Gauntlet Builder
Builder for one piece of a Gauntlet Loop round. Closes the single gap it was given, aiming directly at the qua
Parameter Locator
Given a reform provision, locates the corresponding PolicyEngine parameter YAML path(s), checks current values
Character Simulator
Run one character's turn in a simulated world. Given who the character is (persona, values, drives, current st
Graph Retriever
Intelligent, iterative retrieval over the derived graph query layer. Given a maintainer's information need, co
Dst
Deterministic Simulation Testing methodology, harnesses, fault injection, and shadow state
Promotion
Promotion phase for /parallel-hunt and /run-issues — resolves every register row into an issue file, a refusal
Related Skills
Skill Photo Album
A Claude Code skill that turns a large mixed pool of photos, videos and AI-edited images into a print-ready ve
Fable Eval
Run the fable-mode eval suite (probes → pairwise judge → report). Costs tokens — runs headless claude many tim
Opencode Plugin Comment Judge
opencode plugin: a model judges every comment your agent writes and removes or rewrites the ones that do not e