Proof first: every claim here traces to real client work at candavarci.com.tr.
Web Production Pipeline Kit
Research to launch in six gated stages, with the agents and skills for each.
Four agents and seven skills wired into one six-stage line: research, concept, design, code, quality gate, launch and measurement. Each stage has an owner, an input, an output and a gate, written down in WEB-PIPELINE.md. The Web Engineering Kit gives you the team; this kit gives you the team plus the order it works in and the gates between steps.
Not a promise. A record.
Run the playbook. Then hold the records up to the light.
how to run it
Install the four agents and seven skills
Read WEB-PIPELINE.md once
Run the six stages page by page
Ship only what passes the sim-matrix gate
The six-stage line we use to build and ship sites, written as WEB-PIPELINE.md in English and Turkish and shipped inside the kit next to every agent and skill it names.
brands this system has worked for · 200+ clients
Inside the run · no black box
How the six stages hand work to each other
This kit is not a team list. It is a line: each stage has one owner, takes a written input, leaves a written output, and has a gate the work must pass before the next stage starts.
- Research: ux-visionary maps buyer moments and the funnel, seo-alchemist maps real search demand, Design Intelligence brings the sector data.
- Concept: five ideas from five forced archetypes compete, the winner is named, and a grey skeleton must pass the squint test before any colour.
- Design: a style guide and token set are locked first, then variants are made and scored on a 10-axis rubric; under 80 never leaves the room.
- Code and gate: code-architect builds on the locked tokens with tests, then sim-matrix returns PASS or FAIL at 375, 768 and 1440 px.
- Launch: Deployment Pipeline Design sets the release and rollback, then live checks and Core Web Vitals readings at +3, +14 and +28 days close the loop.
RESEARCH: understanding note and query map
CONCEPT: tournament record, skeleton, blueprint
DESIGN: style guide and a scorecard of 80 or more
CODE: build, tests and a stack note
GATE: PASS/FAIL report and CRO scorecard
LAUNCH: release note with rollback and a measurement log
One power source. 6 lines out.
web-production-pipeline-kit · core
core active · 6 lines
- ✓ building a new site where
Building a new site where research, design and code happen in a fixed order instead of whoever starts first
- ✓ running a five-concept i…
Running a five-concept idea tournament before any code, so the first idea does not win by default
- ✓ locking a style guide and
Locking a style guide and token set so colours, spacing and type cannot drift page to page
- ✓ putting every page through
Putting every page through an independent PASS/FAIL gate at 375, 768 and 1440 px before it goes live
- ✓ scoring a design on a 10…
Scoring a design on a 10-axis rubric and holding back anything under 80
- ✓ launching with a written
Launching with a written rollback and proving on the live page that analytics events actually fire
Yours to keep.
Drag time forward. Watch what stays.
Forever
That's what owning means.
The rented stack
ai writing tool: subscription
expired · access lostanalytics suite: subscription
expired · access lostdesign platform: subscription
expired · access lost(nothing left)
Your forge
-
An order of work, not a pile of tools: every stage says who owns it and what must be true before the next
license: perpetual -
Four agents and seven skills that already cover every stage, so no stage is left without an owner
license: perpetual -
The maker never approves their own work: sim-matrix gives the verdict, not the builder
license: perpetual -
Design taste gets a number before anyone sees it, which cuts rejection rounds
license: perpetual -
“Deployed” stops meaning “done”: the line ends with live checks and a measurement rhythm at +3, +14 and +28 days
license: perpetual
subscriptions expire · deeds don't
11 parts. One line.
Watch the work travel, every station is a product you could buy alone. Together, they run as one.
ux-visionary (agent)
view part → code-architect (agent)
view part → sim-matrix (agent)
view part → seo-alchemist (agent)
view part → Design Intelligence
view part →Design Craft
view part →Next.js App Router Patterns
view part →Web Performance
view part →Core Web Vitals
view part →Landing Page CRO Audit
view part →Deployment Pipeline Design
view part →every part also sold alone, the line is the discount.
Everything in the box.
Pick a piece up. Watch it work.
ux-visionary (agent)
11 parts · one working system · ships instantly by email
This wasn't forged for everyone.
- Not for you if you'd rather rent a tool than own one.
- Not for you if you want someone else to run your stack.
- Not for you if you're happy guessing.
Freelancers and small studios who want every site to go through the same proven line · Developers who can code but want the research, concept and quality steps they usually skip · Teams already using AI agents who need the sequence and the gates, not more agents
then this was forged for you.Catch what's on your mind.
-
How is this different from the Web Engineering Kit?
The Web Engineering Kit is four agents: the team. This kit is a line: four agents and seven skills plus WEB-PIPELINE.md, which sets the order of six stages, who owns each one and the gate between them.
-
What is in WEB-PIPELINE.md?
Six stages: research, concept, design, code, quality gate, launch and measurement. For each one it lists the owning agent or skill, the input, the written output and the gate. It is in English and Turkish.
-
Do I have to use Next.js?
No. The code stage starts with a stack decision by purpose: static generation for content and local sites, a CMS when the owner edits, Next.js for real commerce or apps. The Next.js skill is there for the cases that need it.
-
Does it design the site for me automatically?
No. It gives the order, the agents and the gates. You still make the calls, pick the variant and approve the release; the line makes sure nothing ships without passing its gate.
-
Any refunds?
No refunds, every claim is proven before you buy. The proof is your safety net.
-
How is it delivered?
A file + install guide, instantly by email. No account needed.
-
Does it only work with Claude?
No. The format is open and any AI can set it up; it just needs an agent harness: Claude Code or a compatible setup (the guide explains it).