---
title: "behavior-implementer"
description: "Implements features scenario-first: Given/When/Then scenarios seen failing, the three laws of TDD, and a measured quality gate before every commit."
canonical: https://agentpluginsdirectory.com/plugins/behavior-implementer
last-updated: 2026-09-21
---

# behavior-implementer
Implements features scenario-first: Given/When/Then scenarios seen failing, the three laws of TDD, and a measured quality gate before every commit.
- Slug: behavior-implementer
- Publisher: Tien Nguyen
- Repository: https://github.com/tiennguyentt/agent-plugins
- Manifest: behavior-implementer/plugin.json
- Version: 1.3.0
- License: MIT
- Category (editorial): testing
- Skills: 5 (gate-commit, implement-behavior, review-changes, strengthen-tests, write-scenarios)
- MCP servers: 0
- Stars: 0
- Repository created: 2026-07-31
- Repository last pushed: 2026-08-14
- Publisher type: User
- Listing: https://agentpluginsdirectory.com/plugins/behavior-implementer
- Schema: https://agent-plugins.org/schemas/1.0.0/plugin.schema.json

## What behavior-implementer does, in the publisher's words

A dual-host agent plugin (Claude Code + Codex) that implements features behavior-first: natural-language Given/When/Then scenarios written before any code, every scenario seen failing before it may pass, the three laws of TDD during implementation, and a measured coverage/lint/complexity gate before every commit. It never pushes without asking.

Standalone is measured, not asserted: no file in agents/ or skills/ names a repo-owned plane, so the package carries every rule it enforces.

This is one of several agent plugins in this repository, where agent teams do the work end to end. It is published on its own because a part that only runs inside the repo that grew it is not a part, it is a dependency.

- strengthen-tests takes a module and its green suite, runs mutation testing in a fresh sandbox (never a prior run's cache: that is evidence, not a scratchpad), triages every survivor into accepted-with-a-written-reason or a real hole, and drafts exact-match killing tests for the real holes.

From the project README, punctuation lightly normalized. Full text: https://raw.githubusercontent.com/tiennguyentt/agent-plugins/HEAD/behavior-implementer/README.md

## Skills

- gate-commit: The last step before a commit: measures suite, red-run evidence, coverage, linters, and complexity against the current tree, and reports each as pass, fail, or NOT RUN. Use when an implementation is green and about to be committed, "check before I commit", "gate this", "chạy gate trước khi commit"…
- implement-behavior: The entry skill. Implements features scenario-first: every scenario seen failing, then the three laws of TDD until green. Use when the user asks to "implement this feature", "build X test-first", "làm tính năng này", "code cái này theo TDD", "make these scenarios pass", or hands over any feature re…
- review-changes: The review seat: given a diff and the scenario set it claims to satisfy, reviews every changed hunk for red-first evidence, whether the code does only what a scenario licenses, and anything untested slipping in, reporting every finding, including low- confidence and low-severity ones, as what is w…
- strengthen-tests: The mutation seat: given a module and its test suite, runs a mutation-testing instrument (mutmut or equivalent) in a fresh sandbox, never inside a prior run's cache, which is evidence, not a scratchpad, triages every survivor into accepted-with-a-written- reason or a real hole, and writes exact-m…
- write-scenarios: Turns a requested behavior into natural-language Given/When/Then scenarios before any code exists. Use before implementing any new behavior, when the user asks to "write scenarios for X", "spec this feature as Given/When/Then", "viết scenario cho tính năng này", "define the behavior before we code…

Descriptions come from the frontmatter of each SKILL.md, punctuation lightly normalized.