Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →kennyzir/roblox-site-architect
Orchestrate an evidence-backed seven-stage Roblox site growth pipeline from opportunity assessment through keyword research, source collection, site planning, SEO QA, freshness, and growth.
Passed 18 of 19 test cases with the skill, and 11 of 19 without it.
and 8 more in the full report
7 fixed11 passed either way1 still failing0 broke3 errored
Tokens, turns and time describe resource use, not a dollar price.
One of 7 test cases the skill fixed. Same model, same prompt, with and without the skill.
We analyzed the Roblox game 'Doors' in our opportunity phase at `/sites/doors/pipeline/01-opportunity-report.md`. What are the three valid decision values allowed by the pipeline schema for Stage 01, and what happens if the decision is `reject`?
| Checked | Without skill | With skill |
|---|---|---|
| Lists pursue, hold, and reject as the valid decision values | Missed | Passed |
| Specifies that a reject decision ends the pipeline workflow | Passed | Passed |
The instructions your model reads when the skill loads.
Coordinate research, planning, implementation, validation, and growth work for a Roblox-focused site. This Skill is a workflow orchestrator: it defines stage gates and deliverables, but it does not claim to generate or deploy a complete production site by itself.
The pipeline has seven ordered stages. Each stage writes one durable Markdown artifact under <project-root>/pipeline/ using the schema in references/pipeline-contracts.md. Read that contract before starting or resuming work.
Use a named supporting Skill only when it is available in the current environment and relevant to the request. If it is unavailable, either:
partial or blocked, record the missing capability and evidence gap, and stop at the affected gate.Never state that a Skill, tool, build, deployment, submission, or external action ran unless its actual result is available in the current work record.
complete: all exit criteria are met with cited evidence.partial: useful work exists, but named gaps remain. Proceed only when those gaps cannot invalidate the next decision.blocked: a required input, authorization, or capability is absent. Do not advance.When resuming, inspect existing artifacts in numeric order. Preserve prior evidence, note superseded conclusions, and update timestamps instead of silently rewriting history.
Establish these before Stage 01:
<project-root> if a target project already exists;Unknown inputs stay explicitly unknown. A game name alone does not authorize building, purchasing, publishing, or outreach.
Use scripts/init_pipeline.py to prepare the seven contract artifacts from the maintained pipeline templates. The initializer creates only <project-root>/pipeline/ and its seven Markdown files; it does not build a website or access the network.
Run a dry-run first from any working directory. Dry-run is the default and prints a JSON plan without creating files:
test cases passed, vs 11 / 19 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.