Zhanlin Liu EPOCH
Run evidence-based multi-round loops to tune prompts, agent rules, hyperparameters, and code instead of one-shot tweaks.
Overview
zhanlin-liu-EPOCH is a plugin marketplace for the Operate phase that distributes the EPOCH evidence-based multi-round optimization plugin for prompts, rules, hyperparameters, and code.
What is this marketplace?
- Single epoch plugin for multi-round optimization across prompts, rules, hyperparameters, and code
- Evidence-based iteration rather than guess-and-check tuning
- Fits agent harnesses and codegen workflows where small config changes compound
- Version 1.0.0 from ProRata-affiliated authors with GitHub home at zhanlin-liu/EPOCH
- Marketplace lists 1 plugin: epoch (v1.0.0).
- Plugin category in manifest: development.
Community signal: 3 GitHub stars.
What problem does it solve?
Agent prompts, rules, and code drift get tuned in one-off edits with no repeatable way to know what actually improved behavior.
Who is it for?
Solo builders with an existing agent or codegen setup who want a formal iterate loop on prompts, rules, and code.
Skip if: Greenfield projects with nothing to measure yet, or teams that only need a single static prompt with no iteration plan.
What do I get? / Deliverables
After you add the marketplace and run EPOCH, you get structured optimization rounds with evidence you can compare before committing to the next change.
- Installed epoch plugin from a 1-plugin marketplace
- Documented optimization rounds with evidence tied to each change
Plugins in this marketplace
1 plugin — install individually after you add the marketplace.
Recommended Marketplaces
Journey fit
Canonical shelf is Operate because EPOCH is built around sustained iteration after you already have prompts, rules, or code in play. Iterate subphase matches the framework’s core loop: measure, adjust, and re-run until improvements are justified by evidence.
How it compares
Optimization methodology plugin, not an MCP server or a curated multi-skill review board.
Common Questions / FAQ
Who is Zhanlin Liu EPOCH for?
Indie developers and small teams using Claude Code who already ship agent-assisted work and need repeatable tuning of prompts, rules, hyperparameters, and code.
When should I use Zhanlin Liu EPOCH?
Use it when quality plateaus after initial build or ship and you want evidence-backed rounds instead of random prompt edits.
How do I add Zhanlin Liu EPOCH to my agent?
Register the zhanlin-liu/EPOCH marketplace in Claude Code, install the epoch plugin from the bundle, then invoke it against the artifacts you are optimizing per the plugin’s SKILL flow.