> For the complete documentation index, see [llms.txt](https://docs.pepeunchained.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.pepeunchained.com/mcp/resources-and-prompts.md). # Resources and Prompts The five `pepu://` resources the server exposes and its six prompts: what each is for, its arguments, and how an agent should chain them with the tools. Read this if you attach context by hand in a client UI, or if you want to know what the prompt starters tell the agent to do. Resources and prompts are static text and bundled data; nothing here sends a transaction or needs a key. ## Prerequisites * The server is installed in your client. See [Install per Client](/mcp/install.md). * A client that supports MCP resources and prompts. Not every client does: Claude Code lists prompts as slash commands and resources as `@` mentions, other clients vary. Tools work everywhere. ## Resources Resources are read-only context that you or the agent can attach without a tool call. | URI | Content | MIME type | | ---------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ------------------ | | `pepu://chain-facts` | Mainnet network facts (chain `97741`), a note that the testnet `97740` is out of scope, and the known gotchas list | `application/json` | | `pepu://addresses` | Every tracked mainnet contract from the SDK registry (`getPepuAddresses`): deployed addresses with notes (including the v2 paymaster and the retired v1 paymaster, flagged `retired`), and contracts not yet deployed, with the reason | `application/json` | | `pepu://docs/{slug}` | A page of these docs, from a snapshot bundled at build time. `{slug}` is a page path such as `onboarding/session-keys` | `text/markdown` | | `pepu://skills/{name}` | A skill's `SKILL.md`, bundled at build time. `{name}` is a skill name such as `pepu-session-keys` | `text/markdown` | | `pepu://sdk/exports` | Every `@pepu-chain/sdk` import path with its exports and one-line summaries, generated from the SDK build | `application/json` | Notes: * `pepu://docs/{slug}` and `pepu://skills/{name}` are resource templates. A client lists the available slugs and names, and the server completes partial values. * The docs and skills snapshots are bundled when the package is built. They are as current as the installed server version. For a newer answer, update the server (`npx -y @pepu-chain/mcp` fetches the latest unless a version is pinned) or read the docs directly. * An unknown slug or skill name returns an error that lists what is available. * `pepu://addresses` shows `paymasterV2` (and its alias `paymaster`) at the deployed paymaster `0xdb28528C9399C33c9A0297E8A62F39145acbC30D`, and `paymasterV1` as retired. Do not deposit to the retired v1 paymaster. See [Address Map](/contracts-and-addresses/address-map.md). The tools that mirror resources: `pepu_chain_info` and `pepu_known_gotchas` (chain facts), `pepu_get_docs` (docs), `pepu_sdk_exports` (SDK exports). Use the resource when you want to attach the whole thing; use the tool when the agent should decide. ## Prompts A prompt returns one user message that tells the agent which tools and SDK functions to use, in order. Each ends with a short reminder: sponsorship uses the hosted bundler and paymaster service with a project token, Privy session mode is off-chain and the user pays gas, Kernel session mode is on-chain, deposits to the v2 paymaster are one-way, and a transaction is never called gasless unless a paymaster paid it. | Prompt | Arguments (all optional) | What it tells the agent to do | | ----------------------- | ---------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | `new-pepu-app` | `idea`, `directory` (absolute path) | Read `pepu://chain-facts`, call `pepu_list_templates` and choose a template, call `pepu_scaffold_app`, run `pepu_check_env`, then `pepu_check_privy_config` with the dev origin and port, and use `pepu_sdk_snippet` and `pepu_sdk_exports` for extra wiring. | | `add-google-login` | `projectDir` | Apply the `privy-provider` and `ensure-wallet-route` snippets, run `pepu_check_env` (template `consumer-app`) then `pepu_check_privy_config`, and pin Coinbase to `eoaOnly` and the embedded wallet when wagmi is used. | | `add-session` | `mode` (`privy` or `kernel`), `contract`, `functions` (comma-separated signatures) | Explain the two trust models and confirm one, call `pepu_session_policy`, apply the session snippets (Privy mode also runs `pepu_generate_authorization_key` and asks you to register the public key), and show the consent text and trust notice. | | `add-sponsorship` | `projectDir` | Run `pepu_sponsorship_status`, `pepu_bundler_health` and `pepu_paymaster_health`; apply `sponsored-send` (and `sponsored-send-server` for scripts); use `pepu_explain_error` on failures; do not tell you to deposit into v2. | | `add-sponsored-session` | `contract`, `functions` | Build the Kernel scope with `pepu_session_policy`, apply `sponsored-session` and `sponsored-session-client` (the scope on server and client must match), confirm with `pepu_sponsorship_status`, size spend caps above the gas reservation (about 20.5 PEPU for a session's first operation, 6.9 later), and warn that `createMemoryStore` is dev only. | | `debug-tx` | `hash`, `error` | With a hash: `pepu_explain_tx`. With an error: `pepu_explain_error` using the text exactly as given. Then `pepu_sponsorship_status` if sponsorship is involved, `pepu_decode_error` for custom errors, and `pepu_known_gotchas`. | The three pillar prompts map one to one onto the three pillars: `add-google-login` (Google to wallet), `add-session` (session keys), `add-sponsorship` (sponsored transactions), and `add-sponsored-session` is the combination. See [The Three Pillars and How They Combine](/get-started/pillars.md). ## How agents should chain them 1. Start from a prompt when one fits. It encodes the right order. In Claude Code the [plugin commands](/agent-tooling/plugin.md) wrap the same flows (`/pepu-new-app`, `/pepu-add-login`, `/pepu-add-session`, `/pepu-add-sponsorship`, `/pepu-add-sponsored-session`, `/pepu-doctor`). 2. Ground the facts first: `pepu_chain_info` or the `pepu://chain-facts` resource. Never answer chain questions from memory. 3. Pick code from `pepu_sdk_snippet` and confirm names with `pepu_sdk_exports` before writing anything by hand. 4. Verify with `pepu_check_env` and `pepu_check_privy_config`, then the project's own typecheck and build. 5. When something fails, `pepu_explain_error` with the exact text, then `pepu_known_gotchas`. ## What to expect * Invoking a prompt produces a message, not a result. The agent then makes the tool calls itself, and you approve them according to your client's permission rules. * The read-only tools run without prompting in most clients. The two write tools (`pepu_scaffold_app`, `pepu_generate_authorization_key`) ask first in clients that honour the annotation hints. ## Common failures | Symptom | Cause | Fix | | ---------------------------------------- | --------------------------------------- | ------------------------------------------------------------------------------- | | The client shows no prompts or resources | The client only supports MCP tools | Use the tools directly, or the plugin commands in Claude Code. | | `pepu://docs/...` returns "No docs page" | The slug is not in the bundled snapshot | Read the slug list the error prints, or call `pepu_get_docs` with no arguments. | | Resource content looks stale | The snapshot is bundled at build time | Update the server to the latest version, or pin a newer one. | ## Related * [MCP Server Overview](/mcp/index.md) * [Tool Reference](/mcp/tools.md) * [Skills Catalog](/agent-tooling/skills.md) * [Claude Code Plugin](/agent-tooling/plugin.md) --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://docs.pepeunchained.com/mcp/resources-and-prompts.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.