By iankodj
Plan, build, and test agentic dev loop. A planner skill, dev-executor and tester subagents, context7 docs lookup, and auto-formatting on save.
Runs the test suite and validates against a plan's acceptance criteria. Use to verify implementation is complete. Reports pass/fail per criterion with the underlying test output. Does not modify code.
Implements code from a structured plan. Use after a plan exists and you need someone to do the actual edits. Reads the plan from disk, makes changes, and returns a summary.
End-to-end plan-build-test loop. Drafts a plan, asks for confirmation, delegates to the dev-executor subagent to implement, then to the tester subagent to verify. Loops on test failure up to a configurable cap. Use whenever the user describes a feature or fix they want done end-to-end.
Drafts a structured plan for a development task — frontmatter with goal, affected files, dependencies, and acceptance criteria, plus a prose body of steps and risks. Writes the plan to .triad/plans/ and echoes it in chat. Use whenever the user asks "what's the plan to..." or describes a feature/bug/refactor they want help scoping.
Runs the project's test suite and reports pass/fail with a failure summary. Use when the user asks to run tests, after implementing a change, or to validate the current working tree.
Modifies files
Hook triggers on file write and edit operations
Uses power tools
Uses Bash, Write, or Edit tools
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
This folder is a Claude Code plugin marketplace. Claude Code discovers it from .claude-plugin/marketplace.json, which currently publishes the triad plugin from ./triad.
From Claude Code, add this folder as a marketplace and install triad:
/plugin marketplace add /Users/idjemere/repos/my-plugins/plugins
/plugin install triad@my-plugins
After changing marketplace metadata or plugin versions, refresh Claude Code's listing:
/plugin marketplace update my-plugins
A Claude Code plugin that wraps a real plan → build → test development loop. Three skills, two subagents, one MCP server, one hook.
/triad:build add a healthcheck endpoint that returns 200
triad drafts a plan, asks you to confirm it, hands the plan to a dev-executor subagent that implements the changes, then to a tester subagent that verifies the result against the plan's acceptance criteria. Files saved by the dev-executor are auto-formatted on the way out. Library docs are pulled live from context7 during planning and implementation.
| Entry point | What it does |
|---|---|
/triad:plan <task> | Drafts a structured plan with goal, affected files, dependencies, and acceptance criteria. Writes to .triad/plans/, echoes to chat. |
/triad:build <task> | End-to-end loop: plan → confirm → dev-executor → tester. Bounded retry on test failure (cap configurable). |
/triad:test [scope] | Runs the project's test suite. Auto-detects the runner from project files. |
@triad:dev-executor | Implements code from a plan path. Restricted to non-destructive operations. Self-validates via /triad:test. |
@triad:tester | Read-only validator. Maps acceptance criteria → PASS/FAIL with the underlying test output. |
triad showcases all four canonical Claude Code plugin building blocks composing in one workflow:
/triad:plan, /triad:build, /triad:test@triad:dev-executor, @triad:tester.mcp.json)hooks/auto-format.sh, which dispatches by file extension to prettier / ruff / gofmt / rustfmtgit clone <repo-url> triad
cd <your-project>
claude --plugin-dir /path/to/triad
/plugin marketplace add /Users/idjemere/repos/my-plugins/plugins
/plugin install triad@my-plugins
| Env var | Effect | Default |
|---|---|---|
TRIAD_NO_AUTOFORMAT | Set to 1 to bypass the auto-format hook (e.g. if you have your own format-on-save tooling). | 0 |
TRIAD_MAX_ITERATIONS | Override the /triad:build retry cap. Triad re-invokes @triad:dev-executor up to this many times when tests fail. | 3 |
TRIAD_HOOK_TRACE | Set to 1 to append a diagnostic line per hook invocation. Useful when debugging whether the hook is firing, which formatter ran, or what payload the host sent (especially under VS Code Copilot, which ignores the matcher field). | 0 |
TRIAD_HOOK_LOG | Destination for the trace lines when TRIAD_HOOK_TRACE=1. If unset, the hook writes to <cwd>/.triad/hook.log (project-level, next to plan files), falling back to /tmp/triad-hook.log if .triad/ can't be created. | (see Effect) |
TRIAD_HOOK_NOTIFY | Set to 0 to suppress the system-reminder the hook injects back into the agent's context after a formatter runs (the message that lets the agent tell you "auto-formatted X with prettier"). | 1 |
Plans are markdown files with structured frontmatter, written to <cwd>/.triad/plans/<slug>-<timestamp>.md. Add .triad/ to your .gitignore.
---
goal: Add dark mode toggle to settings page
affected_files:
- src/components/SettingsPage.tsx
- src/styles/themes.ts
dependencies:
- none
acceptance_criteria:
- Toggle persists across page reloads
- Theme applies to all child components
- Existing theme tests still pass
---
# Plan: Add dark mode toggle
## Steps
1. Extend `themes.ts` with a `dark` palette …
2. …
## Risks
- …
The frontmatter is the contract between the planner, dev-executor, and tester. The acceptance_criteria list is what the tester checks.
The hook formats files on every Edit / Write, dispatching by extension:
| Extension | Formatter |
|---|---|
.js .jsx .ts .tsx .mjs .cjs .json .css .html .md .yaml .yml | prettier --write |
.py | ruff format |
.go | gofmt -w |
.rs | rustfmt |
npx claudepluginhub iankodj/my-marketplace --plugin triadHarness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
Comprehensive feature development workflow with specialized agents for codebase exploration, architecture design, and quality review
Reliable automation, in-depth debugging, and performance analysis in Chrome using Chrome DevTools and Puppeteer
Access thousands of AI prompts and skills directly in your AI coding assistant. Search prompts, discover skills, save your own, and improve prompts with AI.
Upstash Context7 MCP server for up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your LLM context.
Comprehensive skill pack with 66 specialized skills for full-stack developers: 12 language experts (Python, TypeScript, Go, Rust, C++, Swift, Kotlin, C#, PHP, Java, SQL, JavaScript), 10 backend frameworks, 6 frontend/mobile, plus infrastructure, DevOps, security, and testing. Features progressive disclosure architecture for 50% faster loading.