What is workbench-builder?

A skill for building temporary browser workbenches on your computer with Bun, SQLite, React, and Astryx. A workbench is a live interface where a person works in the browser while an agent works from the terminal.

Use it to create evaluation viewers, pull-request review rooms, document redlining tools, spreadsheets, triage queues, and trace replays. It provides server and front-end templates plus recipes for these workbench types.

It gives coding, evaluation, review, data, and triage sessions a shared interface with updates flowing between the person and the agent. The setup is designed to run locally without a build configuration or deployment.

workbench-builder — Skill by theagenticguy

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods

npx agentmods add skills/theagenticguy/agentic-plugins/workbench-builder

Any agent

npx skills add theagenticguy/agentic-plugins --skill workbench-builder

Clone the repo

git clone --depth 1 https://github.com/theagenticguy/agentic-plugins

Made for: Claude Code, Codex.

Per session 237 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 3,459 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 1 finding. Scan, not verified.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5	$0.00237	$0.03459
Opus 5	$0.00118	$0.01729
Sonnet 5	$0.00047	$0.00692
Haiku 4.5	$0.00024	$0.00346

Measured yesterday against content hash 4cb4f3f3425c, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

workbench-builder scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.

The scan reads SKILL.md. This mod also ships 24 executable files (doc-review/scripts/review.ts, doc-review/scripts/wait-for-work.ts, doc-review/server.ts, …), listed below but not scanned — reading those needs a real analyzer, not pattern matching.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Makes network callslowCapability

Not a fault in itself. Listed so you know the mod talks to something, and to what.

Phase 0 is inline orchestrator work — name the workbench type and write the data model. Phases 1–3 each own one artifact and can run as a general-purpose `Agent` against `references/orchestrator.md`; for a single-surface

plugins/agentic-skills/skills/workbench-builder/SKILL.md · 144 lines

How it starts

The opening of the file, as written. The whole thing — 144 lines — stays where its author put it; the contents beside it link to each section on GitHub.

File	When to load
`references/orchestrator.md`	Running the pipeline — the five phases, the subagent prompt shape, check-in cadence, stuck detection
`references/write-protocol.md`	Canonical write-protocol block, copied verbatim into every phase worklog and subagent prompt
`references/architecture.md`	The stack contract — SSE-as-invalidation, the fan-out, `useRegion`, the two-way loop, gotchas
`references/dependencies.md`	The pinned dependency set, Astryx grounding commands (`bunx astryx component`), the Graphite theme
`references/rendering.md`	Markdown/mermaid/ECharts patterns, the palette pairing rule, disclosure, overflow guards
`references/recipes.md`	The recipe catalog — one data model + layout per workbench type, with the unique value of each
`templates/server.ts`	Backend scaffold — schema, SSE fan-out + `publish()`, region queries, mutation + loop endpoints
`templates/src/`	Frontend scaffold — `main.tsx`, `App.tsx`, `useRegion.ts`, `charts.ts`, region components
`templates/index.html`	Fullstack entry Bun.serve imports — the whole "no bundler" trick is this one file
`templates/package.json`	The exact-pinned dependency manifest copied into every workbench
`templates/graphite-theme/`	Canonical Graphite theme — `defineTheme` source, built CSS/JS, the `@font-face` fallback
`templates/terminal-helper.ts`	Zero-dependency bun script the terminal/agent runs to act on shared state and close the loop
`templates/wait-for-work.ts`	Wake-on-work watcher — blocks on the SSE stream, exits with a digest when the human hands off work
`templates/worklog-skeleton.md`	Per-phase worklog with the write-protocol block embedded and a `Status: IN PROGRESS` line

Read the full file on GitHub · 144 lines

Files

What ships with it

60 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

yesterday First seen · 144 lines · 237 tokens per session scan A 4cb4f3f3425c

Subscribe to this mod's changes

workbench-builder is a skill published in the GitHub repository theagenticguy/agentic-plugins (5 stars, last pushed 20d ago), licensed MIT. It adds 237 tokens to every session and 3,459 once invoked, about $0.0012 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Code review Databases Frontend agent-skills agentic-ai ai-agents ai-coding-agents anthropic claude

genesis

Use this skill BEFORE drafting any agentic primitive module (skill, persona scoping file, scope-attached rule file, orchestrator workflow) or when refactoring an existing one. Activate whenever the task asks to design, restructure, or critique an agentic module across any agent harness (Claude Code, Copilot, Cursor…

danielmeppiel/genesis · 169 tokens

genesis-evals

Use this skill to run the genesis maintainer-side eval suite against a target model (default: claude-opus-4.7). Activate when validating a genesis PR, when changing the genesis catalogue (architectural-patterns, primitives, design-patterns, refactor-patterns, composition-substrate, pattern-tradeoffs, SKILL.md), or…

danielmeppiel/genesis · 195 tokens

agents

Guidance for AI coding agents working on the adeptability (adept) codebase. Humans: see README.md for usage and CONTRIBUTING.md for the contributor workflow.

itaywol/adeptability · 5 tokens

using-adept

Use the adept CLI to author AI skills once and render them into Claude Code, Cursor, Codex, Copilot, and OpenCode. Apply when installing/syncing skills, editing skill.yaml/SKILL.md, or touching .adeptability/.

itaywol/adeptability · 54 tokens

authoring-adept-agents

Write a good, portable adept agent (subagent): trigger-shaped description, one job per agent, generator/evaluator separation, explicit boundaries, restricted tools. Apply when creating or editing an agent file or running adept agent add.

itaywol/adeptability · 47 tokens

authoring-adept-loops

Compose a loop — a scheduled system that discovers work, hands it to agents, verifies with an independent evaluator, persists state, and reschedules itself. Apply when the user wants automation that runs on a timer, a triage/babysitter routine, or asks about adept loop add.

itaywol/adeptability · 60 tokens

workbench-builder