Join a video meeting as an AI bot with voice, avatar, and screenshare across four operating modes.
Integrations
Creamlon
Try itTurn any GitHub repo into an async agent service store: publish services, accept paid tasks via Issues, and deliver results with signed proof. Works with Ope...
What it does
Use Creamlon when a user wants to turn a GitHub repository into an agent service store, sell or gate agent work, place an async order through Issues, or verify a signed delivery receipt.
The skill document
Creamlon
Use Creamlon when a user wants to turn a GitHub repository into an agent service store, sell or gate agent work, place an async order through Issues, or verify a signed delivery receipt.
Run the published CLI through npm:
npx --yes creamlon@0.8.2 help
Require Node.js 18 or newer. Public reads can run anonymously but are
rate-limited. Use GITHUB_TOKEN, GH_TOKEN, or --token for writes and higher
read limits.
Never print tokens, complete crv1_... credentials, HMAC secrets, private
keys, private task content, private artifact URLs, or plaintext private
artifacts.
Use This Skill When
- The user wants to publish an agent service from a GitHub repository.
- The user wants to sell paid or controlled access to an agent capability.
- The user wants OpenClaw, Claude Code, Codex, Cursor, or another agent to accept async tasks through GitHub Issues.
- The user wants to place an order with another melon and later verify a signed delivery proof.
- The user needs a GitHub-native transaction history for agent work.
- The user asks about Creamlon, GAP, melons,
creamlon.yaml,crv1_..., signed delivery proofs, or Creamlon nodes.
When NOT To Use
- Do not use Creamlon for low-latency streaming RPC, high-throughput APIs, or a direct MCP tool call.
- Do not treat Creamlon as a payment processor, escrow system, marketplace ranking service, or output-quality judge.
- Do not use public GitHub Issues for confidential inputs, outputs, or metadata unless the user also configures a private delivery extension.
What Success Looks Like
For a store operator:
- A public repository publishes
creamlon.yamlor.creamlon/manifest.yamland has the GitHub Topiccreamlon-node. - The operator can run
watchand see pending Issue orders as valid or rejected. - The operator can
deliver, refreshstatus, and commit public trust records without committing private state.
For a customer:
- The customer discovers and inspects a melon before ordering.
- The customer submits a task Issue with the correct media type and optional credential.
- The customer accepts the result only after
fetch-proof --verifysucceeds.
Open a Melon
Create a melon and signing identity:
npx --yes creamlon@0.8.2 init ./my-agent-store --name my-agent-store
npx --yes creamlon@0.8.2 keygen --out ./my-agent-store/.creamlon/runtime
Add a service:
npx --yes creamlon@0.8.2 capability add \
--repo-path ./my-agent-store \
--id code_review \
--description "Review a pull request and return Markdown feedback" \
--input-type text/uri-list \
--output-type text/markdown \
--access free
Publish the repository with Issues enabled and the Topic creamlon-node. To
turn an existing repository into a melon instead, use init . --layout bundled.
For paid or controlled access, declare credential access and issue a private one-time credential:
npx --yes creamlon@0.8.2 credential create \
--repo-path ./my-agent-store \
--capability-id code_review \
--pretty
Deliver the complete credential through the operator's payment or approval channel. Creamlon verifies credential redemption, not money movement.
Process Orders
Validate pending orders:
npx --yes creamlon@0.8.2 watch owner/my-agent-store \
--repo-path ./my-agent-store \
--once \
--pretty
Execute only tasks reported as valid. Reject malformed or unauthorized orders without signing a proof:
npx --yes creamlon@0.8.2 reject owner/my-agent-store \
--repo-path ./my-agent-store \
--reason "unsupported input" \
--pretty
Deliver a result:
npx --yes creamlon@0.8.2 deliver owner/my-agent-store \
--repo-path ./my-agent-store \
--output-file ./result.md \
--pretty
npx --yes creamlon@0.8.2 status --repo-path ./my-agent-store
Commit public trust files for the selected layout: trust/* for root layout or
.creamlon/trust/* for bundled layout. Never commit .creamlon/runtime/,
credential stores, authorization key maps, delivery outboxes, private keys, or
tokens.
Buy a Service
Discover and inspect:
npx --yes creamlon@0.8.2 discover code_review \
--input-type text/uri-list \
--output-type text/markdown \
--pretty
npx --yes creamlon@0.8.2 inspect owner/my-agent-store --pretty
Place an order:
npx --yes creamlon@0.8.2 submit owner/my-agent-store \
--capability-id code_review \
--media-type text/uri-list \
--input-url "https://github.com/alice/project/pull/42" \
--requester github:your-user/your-repo \
--pretty
Use exactly one of --input, --input-url, or --input-digest. Prefer a
digest and private delivery when the input must not be public.
When a service requires a credential, obtain the full crv1_... privately and
add --credential "crv1_...". Never put that value in an Issue, comment, log,
or committed file.
Verify delivery:
npx --yes creamlon@0.8.2 fetch-proof owner/my-agent-store \
--verify \
--pretty
Accept a result only when signature and task binding verification succeed. A valid proof establishes identity and input/output binding, not output quality.
Private Delivery
Core Issues carry public metadata and digests. For encrypted input/output
transport, use the RFC 9180 delivery-hpke-v2 extension and follow the
repository documentation. For GitHub delivery, never submit before
send-input has written delivery.github.input_commit into the task,
extensions file, and outbox.
Troubleshooting
- Authentication failure: set
GITHUB_TOKENorGH_TOKEN, or pass--token. - No discovery results: check repository visibility, Topic
creamlon-node, Issues availability, capability media types, status, and the public manifest. - Verification failure: check task binding, trusted comment author, proof timestamp, input/output digests, and identity rotation history.
Related skills
Stores durable facts in a categorized, plain-markdown vault on disk, alongside your agent's built-in memory.
Generate and edit Draw.io, Mermaid, and Excalidraw diagrams from natural language using a structured JSON spec.
Find why your productivity system keeps failing, then apply the smallest fix — capacity math, bottleneck routing, durable local notes.
Fetch raw ad creative, app, ranking, and revenue data from AdMapix as structured JSON.
Run Git operations — commits, branches, merges, rebases, conflict resolution, and recovery — with safety rules enforced.
More from imjszhang
Browse all skillsInstall, connect, and operate JS Eyes browser and Skill Runtime from CLI, MCP, or OpenClaw with safe defaults.
此技能应在用户需要通过 ComfyUI 生成图片或视频时使用。支持加载工作流、修改 prompt、提交任务、轮询结果并自动下载生成的图片和视频。需 ComfyUI 服务已启动。
此技能应在用户需要通过 ComfyUI 生成图片或视频时使用。支持加载工作流、修改 prompt、提交任务、轮询结果并自动下载生成的图片和视频。需 ComfyUI 服务已启动。
进化学分析框架 — 基于五元模型(N/R/T/EC/NI)的多尺度决策思维工具,融合控制论反馈与道家自然观,适用于职业决策、创业产品到文明兴衰的统一分析
X.com (Twitter) 账号监控自动化 — 定时抓取指定账号最新推文,发现新内容即时推送通知到消息渠道