bench-slice
Teach Bench one of YOUR domains so it generates that proprietary code as cleanly as core Laravel — builds a full skill→agent→pattern slice for a domain you…
Skill by PDX-Apps
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/pdx-apps-bench-slice
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: documents, agent-building
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/bench-slice
What's inside
- Skill name
bench-slice- Description
- Teach Bench one of YOUR domains so it generates that proprietary code as cleanly as core Laravel — builds a full skill→agent→pattern slice for a domain you point at. Use when the user says "make a slice for app/Reports", "teach Bench my Reports domain", "I want a /report command that scaffolds reports like core does controllers", "build a skill+agent+pattern for our Sagas", or "scaffold Bench support for app/{Domain}". Writes the triple under ./.bench/ (skill + paired agent + the domain pattern).
- License
- MIT
- Instructions
- 70 lines in SKILL.md · view on GitHub
Agent Skills spec issues
- name: missing
Files (1)
- SKILL.md
SKILL.md (first lines)
You're the **/bench-slice** skill. Build a project-local skill→agent→pattern slice for a domain the user names, by orchestrating the three authoring agents. You don't author files yourself — you sequence the authors and relay their output.
The user's request: **$ARGUMENTS**
## Step 1: Identify the domain
- Resolve the domain name + path (e.g. `Reports` → `app/Reports/`). Confirm it exists with a quick `find`/`ls`. If it doesn't exist yet, say so — the slice will be more speculative and the first run needs close review.
- Pick the slug for the slice: `/{domain}` skill, `{domain}` agent, `{Domain}` pattern (e.g. `report` / `report` / `Report`).
## Step 2: CAPTURE the domain pattern (pattern-author)
Delegate first — the pattern is what the agent will read, so it comes first.
```
Task(subagent_type: "pattern-author", description: "Capture {domain} pattern", prompt: """
intent: capture
domain: {domain} (code lives at {path})
This is the pattern half of a /bench-slice — author a project pattern describing
how THIS project writes {domain} classes (structure, naming, base classes, the
full file set per unit, conventions). defer_rebuild: true.
project_root: {cwd} bench_install_root: <PLUGIN_ROOT>
""")
```
Keep the returned **findings + pattern path + generation surface** — you pass them forward so the next authors don't re-scan.
## Step 3: Design the skill + agent (skill-author → agent-author)
```
Task(subagent_type: "skill-author", description: "Design /{domain} skill+agent", prompt: """
intent: new
name: {domain-slug}
change_request: a /{domain} command that scaffolds a {Domain} unit + its full
generation surface, matching this project.
Generation surface (from pattern-author): {file list}
Domain pattern to read: {.bench pattern path}
Hand off to agent-author for the paired {domain-slug} agent; its Pattern Lookup
points at the domain pattern above. defer_rebuild: true.
project_root: {cwd} bench_install_root: <PLUGIN_ROOT>
""")
```
`skill-author` cascades to `agent-author` internally, so this yields both the skill and its agent.
## Step 4: Approve, rebuild once, report
After the user approves all three drafts and they're written, rebuild a single time:
```bash
<PLUGIN_ROOT>/bin/bench rebuild
```
Report the slice:
```
Slice created for {Domain}:
- skill: .bench/skills/{slug}/SKILL.md
- agent: .bench/agents/{slug}.md
- pattern: .bench/patterns/{group}/{Domain}-...md
Generation surface: {files one invocation produces}
Try it: /{slug} create {Example}
```
## Notes
- **Sequence matters**: pattern first (the agent reads it), then skill+agent.
- **One rebuild** at the end — pass `defer_rebuild: true` to each author so they don't each rebuild.
- Surface every author's **divergence callouts** to the user; don't bake in a guessed convention silently.
- A slice is project-local under `.bench/`. If it'd serve many projects, suggest promoting it to a standalone addon later.From PDX-Apps/bench, MIT.
More from PDX-Apps/bench
- 0.470score
ai
PDX-Apps
Build AI features with the laravel/ai SDK — routes to the right worker for agents, tools (function calling), and RAG (embeddings + vector search).
SkillCFreshness: CClaude Code +9 - 0.470score
ai-tool
PDX-Apps
Generate a Laravel AI Tool class (laravel/ai) for function-calling — description, typed parameter schema, handle().
SkillCFreshness: CClaude Code +9 - 0.470score
e2e-run
PDX-Apps
Verify a ticket's acceptance criteria in a LIVE browser (Chrome MCP or Playwright MCP) and report pass/fail per criterion with evidence — no test file written.
SkillCFreshness: CClaude Code +9 - 0.470score
bench-addon-add
PDX-Apps
Add a single Bench addon to this project from inside Claude and run only its onboarding — install the addon, interview the concerns it ships that aren't…
SkillCFreshness: CClaude Code +9 - 0.470score
bench-configure
PDX-Apps
Run Bench's guided concern setup — interview the user about essential project concerns (auth, test framework, permissions, response shape, layout, and any…
SkillCFreshness: CClaude Code +9 - 0.470score
bench-init
PDX-Apps
First-run setup that tailors Bench to THIS project.
SkillCFreshness: CClaude Code +9 - 0.470score
bench-list
PDX-Apps
Discover what's available in Bench — patterns, skills, agents — both bundled defaults and project-local overrides under ./.bench/.
SkillCFreshness: CClaude Code +9 - 0.470score
bench-override
PDX-Apps
Change a Bench default for THIS project — override a pattern (how code is generated), a skill (how a slash command behaves), or an agent (how a worker runs).
SkillCFreshness: CClaude Code +9 - 0.470score
bench-revert
PDX-Apps
Remove a project-local Bench override and restore the bundled default.
SkillCFreshness: CClaude Code +9 - 0.470score
ai-agent
PDX-Apps
Generate a Laravel AI Agent class (laravel/ai) — instructions, provider/model attributes, optional structured output, conversation memory, tools.
SkillCFreshness: CClaude Code +9 - 0.470score
bench-status
PDX-Apps
Synthesized status of the Bench install — versions, addons, project-local overrides, version drift, suggested next steps.
SkillCFreshness: CClaude Code +9 - 0.470score
filament-resource
PDX-Apps
Generate a Filament 4 admin-panel Resource (form schema + table) for a model.
SkillCFreshness: CClaude Code +9
Signals
- Stars
- 0
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Jun 8, 2026 (4 months ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/PDX-Apps/bench
- Website
- github.com/PDX-Apps/bench
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from PDX-Apps
- 0.470score
bench-list
PDX-Apps
Discover what's available in Bench — patterns, skills, agents — both bundled defaults and project-local overrides under ./.bench/.
SkillCFreshness: CClaude Code +9 - 0.470score
bench-override
PDX-Apps
Change a Bench default for THIS project — override a pattern (how code is generated), a skill (how a slash command behaves), or an agent (how a worker runs).
SkillCFreshness: CClaude Code +9 - 0.470score
bench-addon-add
PDX-Apps
Add a single Bench addon to this project from inside Claude and run only its onboarding — install the addon, interview the concerns it ships that aren't…
SkillCFreshness: CClaude Code +9 - 0.470score
bench-configure
PDX-Apps
Run Bench's guided concern setup — interview the user about essential project concerns (auth, test framework, permissions, response shape, layout, and any…
SkillCFreshness: CClaude Code +9 - 0.470score
bench-init
PDX-Apps
First-run setup that tailors Bench to THIS project.
SkillCFreshness: CClaude Code +9 - 0.470score
bench-revert
PDX-Apps
Remove a project-local Bench override and restore the bundled default.
SkillCFreshness: CClaude Code +9
Cite this page
bench-slice skill. Agent Mart. https://agentmart.dev/skills/pdx-apps-bench-slice (last updated 2026-10-24)