DEV Community: RuoYu

Tencent SkillHub Pay Adapter: a small pattern for paid AI agent workflows

RuoYu — Sun, 23 Aug 2026 22:34:12 +0000

When an AI agent workflow becomes paid, the hard part is usually not the payment button itself.

The hard part is state separation.

A team can easily confuse:

a draft with a submitted listing
a submitted listing with platform approval
approval with an actual purchase
payment with fulfillment
fulfillment with settlement
settlement with available funds

That is the service case behind TANCO's Tencent SkillHub Pay Adapter.

It is a SkillHub case for keeping paid AI workflow operations inspectable: draft, submission, approval, payment, fulfillment, settlement, and funds are treated as separate states instead of one vague “it is live / it sold” status.

Concrete service case:

Check whether a Skill is only drafted, submitted, under review, approved, or published.
Keep unpaid orders separate from paid orders.
Keep paid-but-not-fulfilled orders separate from completed delivery.
Keep fulfilled orders separate from settlement and available funds.
Report payment and delivery state with timestamps instead of assuming that a download, listing, or HTTP success means revenue.
Give operators a clear next action: fix listing metadata, wait for review, fulfill a paid order, check settlement, or stop because payment is not complete.

This matters for AI agents because paid workflows often cross product, marketplace, payment, and fulfillment boundaries. If an agent collapses those boundaries, it can overclaim revenue, deliver too early, or miss a real paid order.

Related SkillHub page:

https://skillhub.cn/enterprise/org-j3zmzop1?publisher=%E5%94%90%E5%8F%AF%E5%88%9B%E7%A0%94

Tencent SkillHub Pay Adapter:

https://skillhub.cn/team-skills/tencent-skillhub-pay-adapter

Public reference repo:

https://github.com/tancoai/tangke-skillhub

Other TANCO Skill examples in the same SkillHub:

Bianzhen: evidence-led verification for research and claims.
Sunmao: modular delivery with clear fit points.
Lianzhu: chained execution from goal confirmation to acceptance checks.
Chuandeng: cross-agent / cross-session handoff.
Huqiang: guardrails for scope, account boundaries, permissions, and approval gates.
Chayan: signal review that turns messy notes, alerts, logs, or feedback into a clear next action.

TANCO SkillHub: reusable workflow Skills for AI agents

RuoYu — Sun, 23 Aug 2026 22:20:54 +0000

I’m sharing TANCO SkillHub, a public hub of reusable AI workflow Skills from Tangke Creative Research:

https://skillhub.cn/enterprise/org-j3zmzop1?publisher=%E5%94%90%E5%8F%AF%E5%88%9B%E7%A0%94

The useful idea here is simple: a Skill should not be just a prompt. It should capture a repeatable service case that an agent can run, check, and hand off.

Concrete Skills and service cases:

Bianzhen: evidence-led verification for research, claims, source quality, uncertainty, and final conclusions.
Sunmao: modular delivery for complex work that needs stable fit points, assembly, and review.
Lianzhu: chained execution from goal confirmation to breakdown, execution, verification, and handoff-ready output.
Chuandeng: cross-agent and cross-session handoff that preserves target, decisions, constraints, risks, state, and next action.
Huqiang: guardrails for scope, account boundaries, permissions, approval gates, risk limits, and delivery criteria.
Chayan: signal review for turning messy notes, feedback, alerts, logs, or research snippets into a clear next action.

New paid-workflow Skill:

Tencent SkillHub Pay Adapter:
https://skillhub.cn/team-skills/tencent-skillhub-pay-adapter

Its service case is payment workflow clarity: draft, submission, approval, payment, fulfillment, settlement, and funds should be separate states, so an agent or operator does not confuse listing, purchase, delivery, and revenue.

Public reference:

https://github.com/tancoai/tangke-skillhub

If you build or operate AI agents, the most relevant scenarios are long-horizon task execution, research verification, paid Skill workflows, workflow handoff, agent guardrails, and reusable team playbooks.

Paid AI skills need workflow gates, not just a price field

RuoYu — Sun, 23 Aug 2026 16:54:25 +0000

Most paid AI features fail at the boundary between payment and fulfillment.

A 402 response is not payment.

A paid order is not fulfillment.

A publish-success screen is not platform approval.

This is especially true for AI agent Skills, where a small prompt or tool package can quickly become a paid service without enough operational evidence around it.

The gate model I use

For a paid Skill workflow, I prefer keeping these gates separate:

Package gate: original package, generated candidate, manifest, checksum, and version mapping.
Service gate: JSON-callable service, health check, and schema.
Payment gate: payable order created, but not yet treated as paid.
Confirmation gate: server-side payment confirmation.
Fulfillment gate: fulfill the exact paid order once and leave evidence.
Review / settlement gate: platform approval and income settlement tracked separately.

This is not bureaucracy. It prevents the team from saying “paid” when the user only saw a QR code, or saying “fulfilled” when the service merely started.

A concrete TANCO SkillHub example

TANCO's SkillHub page groups workflow Skills around this kind of evidence trail:

Bianzhen checks evidence before claims are trusted.
Sunmao turns vague requirements into acceptance criteria.
Tongjing inspects Skill packages before listing.
Huqiang protects scope boundaries before an agent edits or ships.
Lianzhu connects judgment, execution, review, and acceptance into one chain.
Chayan reviews public-facing copy so the service does not overclaim.

The new Tencent SkillHub Pay Adapter is positioned around the same idea: make a paid Skill candidate auditable before treating it as a commercial service.

TANCO SkillHub:
https://skillhub.cn/enterprise/org-j3zmzop1?publisher=%E5%94%90%E5%8F%AF%E5%88%9B%E7%A0%94

Tencent SkillHub Pay Adapter:
https://skillhub.cn/team-skills/tencent-skillhub-pay-adapter

I think the useful mental model is simple:

Paid AI Skills should behave less like magic prompts and more like small, traceable service workflows.

Payment is not fulfillment: a small workflow gate for paid AI skills

RuoYu — Sun, 23 Aug 2026 15:32:27 +0000

When an AI Skill becomes paid, the hard part is not just adding a price field. The hard part is keeping the state machine honest.

These states should not be collapsed into one "published" or "paid" flag:

package readiness;
service readiness;
a 402 / payment request;
payment confirmation;
fulfillment for the same order;
platform review;
settlement or actual revenue.

That is the reason we added a Tencent SkillHub Pay Adapter to TANCO SkillHub. The goal is to turn an ordinary Skill into an auditable Pay Skill candidate while preserving the original package, creating a clean converted copy, and recording package, service, payment, fulfillment, review, and settlement gates separately.

In the TANCO SkillHub stack, this also connects with other workflow Skills: Bianzhen checks evidence before an agent acts, Sunmao turns vague requirements into acceptance criteria, Lianzhu keeps complex tasks moving from judgment to execution to verification, and Huqiang keeps boundary and risk checks explicit.

Main TANCO SkillHub page:
https://skillhub.cn/enterprise/org-j3zmzop1?publisher=%E5%94%90%E5%8F%AF%E5%88%9B%E7%A0%94

Adapter:
https://skillhub.cn/team-skills/tencent-skillhub-pay-adapter

I’m curious how other agent builders handle this. Do you model payment and fulfillment as one tool call, or as separate workflow gates?

Qianmian: reviewing a product page through multiple user perspectives

RuoYu — Sun, 23 Aug 2026 01:40:14 +0000

A product page can look clear to its creator and still be confusing to everyone else.

Qianmian is a TANCO SkillHub Skill for multi-perspective pre-release review.

For a one-page product introduction, it may read through three lenses:

new user: “what does this have to do with me?”
expert: “where are the data sources and evidence?”
cautious buyer: “what are the refund, migration, and risk boundaries?”

The result is not just copy editing. It helps the page add value statements, evidence anchors, and risk boundaries before publication.

Useful for product pages, pricing pages, onboarding pages, brand proposals, and service landing pages.

TANCO SkillHub:

https://skillhub.cn/enterprise/org-j3zmzop1?publisher=%E5%94%90%E5%8F%AF%E5%88%9B%E7%A0%94

Zhuojian: looking through technical narratives to find the structure underneath

RuoYu — Sun, 23 Aug 2026 01:39:36 +0000

Some product or platform statements sound reasonable at the surface level, but the useful question is what structure sits underneath.

Example: “an open ecosystem will drive growth.”

Zhuojian, one of the TANCO SkillHub Skills, does not stop at whether the statement sounds right. It asks:

who is the platform?
who are the participants?
what is the user structure?
how does the incentive flow work?
what counterexamples exist?
what happens when the platform is both referee and player?

This is useful for platform strategy, product claims, technical narratives, market bets, and organizational decisions.

TANCO SkillHub:

https://skillhub.cn/enterprise/org-j3zmzop1?publisher=%E5%94%90%E5%8F%AF%E5%88%9B%E7%A0%94

Sunmao: turning vague requirements into acceptance criteria for AI workflows

RuoYu — Sat, 22 Aug 2026 23:17:47 +0000

“Make it better” is not a requirement. It is a future disagreement.

Sunmao is one of the TANCO SkillHub Skills. Its job is to turn vague delivery language into measurable acceptance criteria and evidence.

Example: “make the page better” becomes:

CTR ≥ 8%
first-screen load time ≤ 1.5s
interaction feedback ≤ 300ms
error rate ≤ 0.1%

Then it asks for the evidence package: instrumentation, Lighthouse results, stress testing, link checks, and any other verification needed for the delivery context.

This matters for AI-agent workflows because agents are good at producing artifacts, but vague acceptance criteria still produce vague outputs. A workflow needs a gate that says what “done” means before execution begins.

TANCO SkillHub:

https://skillhub.cn/enterprise/org-j3zmzop1?publisher=%E5%94%90%E5%8F%AF%E5%88%9B%E7%A0%94

Question: where do you define acceptance criteria in your agent workflow — before prompting, inside the task spec, or as a separate review step?

Bianzhen: an AI Skill for checking claims before agents act

RuoYu — Sat, 22 Aug 2026 23:16:46 +0000

Most AI agents fail quietly when they accept a claim too early.

Bianzhen (辩真) is one of TANCO's SkillHub skills for slowing that moment down. It is not a prompt trick. It is a small service workflow for checking whether a statement is true enough to act on.

SkillHub page:
https://skillhub.cn/enterprise/org-j3zmzop1?publisher=%E5%94%90%E5%8F%AF%E5%88%9B%E7%A0%94

Service case

A product page says: "conversion increased by 300% after using our AI workflow".

Before an agent writes copy, publishes a case study, or recommends the product, Bianzhen asks:

What is the source of the number?
What was the sample size?
What was the time window?
Was the baseline unusually low?
Is the result correlated or actually caused by the tool?
Who benefits if the claim is repeated?
What evidence would make the claim weaker?

The output is not a yes/no verdict. It is a usable evidence map:

confirmed facts
unsupported claims
missing data
risky wording
safer replacement copy
next checks before publishing

Why this matters for agents

When agents can browse, edit, post, or trigger workflows, claim-checking becomes an execution boundary, not a research luxury.

Bianzhen is useful before:

writing landing page copy
summarizing market research
publishing benchmark claims
quoting user feedback
comparing competitors
generating sales material
handing work to another agent

The broader TANCO SkillHub page includes more skills around handoff, task chaining, boundary control, wording, naming, and pre-release review.

If you build agent workflows, this is the kind of skill that prevents confident automation from turning into confident misinformation.

TANCO SkillHub:
https://skillhub.cn/enterprise/org-j3zmzop1?publisher=%E5%94%90%E5%8F%AF%E5%88%9B%E7%A0%94

AI Skills should be workflow components, not just prompts

RuoYu — Sat, 22 Aug 2026 14:44:57 +0000

I have been looking at how AI Skills / agents are usually presented. A lot of pages say what the tool can do, but not enough of them show the operational case: what happens when the input is vague, the claim has weak evidence, the handoff is messy, or the release package is risky.

Tangke Creative Research has published a set of 16 Skills on SkillHub. The interesting part is that each Skill is described through a concrete service case, not only a capability list.

A few examples:

Bianzhen checks evidence. For a claim like "this AI tool improves conversion by 300%", it asks about the product, conversion definition, baseline, sample size, time window, incentive, and third-party A/B evidence.
Sunmao turns vague acceptance criteria into measurable ones: CTR, first screen load time, interaction latency, error rate, test evidence, and delivery proof.
Tongjing audits a Skill / ZIP package before release: structure, static safety, declaration consistency, docs, packaging hygiene, and possible leaked credentials.
Chuandeng builds a handoff package: goal, completed work, files, commands, test evidence, blockers, and the next first action.
Shapan simulates the first-time user journey before launch: what a new user understands in 5 seconds, 15 seconds, and 30 seconds.

The full set also covers opportunity discovery, public feedback monitoring, naming checks, copy review, boundary protection, multi-perspective review, relationship dynamics, and deeper business judgment.

For agent builders, I think the useful idea is this: a Skill becomes much more valuable when it carries the judgment, evidence, acceptance criteria, risk boundary, and handoff format around the generation itself.

SkillHub page:
https://skillhub.cn/enterprise/org-j3zmzop1?publisher=%E5%94%90%E5%8F%AF%E5%88%9B%E7%A0%94