wallkop

15 mods across 1 repository, 11 stars between them.

opc-develop

02

wallkop/opc-develop

Plugin Claude Code

Fable-5-era product delivery: 产品定义 product-truth assets, harness-adjudicated acceptance, and a slimmed brainstorm/design/build skill set.

11 1mo ago A tokens not measured original MIT

opc-implementer

03

wallkop/opc-develop

Agent

Follow ${CLAUDEPLUGINROOT}/shared/prompts/implementer.md. Own exactly the bounded task and allowed paths named by the dispatcher. You receive no creator conversation and do not broaden scope.

11 1mo ago A 41 tokens original MIT

opc-reviewer

04

wallkop/opc-develop

Agent

Independent opc-develop reviewer for standard-increment reality/final gates and explicit requirement/demo/PRD/technical gates. Use with a fresh empty context and the exact rubric, diff, artifacts, and evidence only. Read-only by design.

11 1mo ago A 52 tokens original MIT

brainstorm

05

wallkop/opc-develop

Skill Claude CodeCodex

A requirements review step that questions whether a proposed feature is worth building before development starts.

11 1mo ago A 89 tokens original MIT

demo

06

wallkop/opc-develop

Skill Claude CodeCodex

A realistic product preview that uses fake data so people can try an unfinished idea in the real application interface.

11 1mo ago A 86 tokens original MIT

deploy

07

wallkop/opc-develop

Skill Claude CodeCodex

Use when one or more human-accepted Lite or Build increments are merged to trunk and should release to production safely. Build uses fresh acceptance receipts; Lite uses test acceptance, focused evidence, matching risk checks, and the project's complete release gate. Performs fail-closed preflight, backups/rollback…

11 1mo ago A 82 tokens original MIT

design

08

wallkop/opc-develop

Skill Claude CodeCodex

A product-definition process that combines product requirements, technical decisions, and acceptance tests into one reviewed document.

11 1mo ago A 115 tokens original MIT

harness-init

09

wallkop/opc-develop

Skill Claude CodeCodex

A guided setup workflow for new software projects using Harness, a development environment for running, observing, and judging test cases. It helps make decisions about the technology stack, infrastructure, architecture, documentation, and safety rules.

11 1mo ago A 86 tokens original MIT

harness-retrofit

10

wallkop/opc-develop

Skill Claude CodeCodex

A guide for improving the development and testing setup of an existing project without rebuilding it from scratch. A harness is the set of tools, environments, logs, tests, and rules used to operate and verify the project.

11 1mo ago A 81 tokens original MIT

lite

11

wallkop/opc-develop

Skill Claude CodeCodex

A workflow for making a small code change without changing the product's agreed behavior or requirements.

11 1mo ago A 79 tokens original MIT

oncall

12

wallkop/opc-develop

Skill Claude CodeCodex

Use when something is wrong in production (or on the test environment): triages severity, investigates through correlation-ID logs, read-only DB queries, traces, and recent releases, produces a root-cause/blast-radius report, executes rollback, a Lite-or-Build hotfix through ship/deploy, or mitigation, and leaves an…

11 1mo ago A 80 tokens original MIT

retro

13

wallkop/opc-develop

Skill Claude CodeCodex

Run the measured loop-engineering pass: audit ledger/usage quality, evaluate executable incident cases and crystallized rules, mine recurrence and rework, then propose verified process changes for human approval.

11 1mo ago A 40 tokens original MIT

ship

14

wallkop/opc-develop

Skill Claude CodeCodex

Use when a verified Lite or Build increment should deploy to test, be accepted, and merge. Build keeps its approved demo/PRD/testcase, receipt, and review gates. Lite instead uses focused regression evidence, matching risk checks, and the project's complete predeploy gate without manufacturing feature artifacts.…

11 1mo ago A 70 tokens original MIT

vibe

15

wallkop/opc-develop

Skill Claude CodeCodex

A coding workflow for quickly producing code for a person to inspect and accept manually. It intentionally leaves testing and automated verification to the human reviewer.

11 1mo ago A 53 tokens original MIT