Now liveThe Skillselion MCP - thousands of ranked skills, loaded into your agent mid-task. No install.Get it →
franalgaba avatar

Grimoire

  • 1.1k installs
  • 6 repo stars
  • Updated May 8, 2026
  • franalgaba/grimoire

grimoire provides documented workflows for Install and operate Grimoire, author .spell files with full syntax coverage (including advisory decision logic), and run compile/validate/simulate/cast safely.

About

The grimoire skill install and operate Grimoire, author .spell files with full syntax coverage (including advisory decision logic), and run compile/validate/simulate/cast safely. Use when users ask to create, edit, debug, validate, simulate, execute, or explain Grimoire strategies. # Grimoire CLI Skill This skill is the base operating playbook for Grimoire. ## When To Use Use this skill when the task includes: - install/setup of Grimoire tooling - creating or editing `.spell` files - syntax questions about DSL capability - advisory (`advise`) authoring, debugging, and replay workflows - setup/compile/validate/simulate/cast/resume workflows - debugging spell compile/runtime failures ## Mandatory Loading Rules All `references/` and `docs/` paths below are relative to this skill directory (`skills/grimoire/`). These rules are **required** - they solve syntax coverage gaps that cause authoring errors. **STOP - read this first:** Do NOT search the codebase, grep for syntax patterns, or rely on memory for DSL syntax. The reference files bundled with this skill are the single source of truth.

  • install/setup of Grimoire tooling
  • creating or editing `.spell` files
  • syntax questions about DSL capability
  • advisory (`advise`) authoring, debugging, and replay workflows
  • setup/compile/validate/simulate/cast/resume workflows

Grimoire by the numbers

  • 1,058 all-time installs (skills.sh)
  • +3 installs in the week ending Jul 28, 2026 (Skillselion tracking)
  • Ranked #110 of 1,382 Code Review & Quality skills by installs in the Skillselion catalog
  • Security screen: MEDIUM risk (skills.sh audit)
  • Data as of Jul 28, 2026 (Skillselion catalog sync)
At a glance

grimoire capabilities & compatibility

Capabilities
install/setup of grimoire tooling · creating or editing `.spell` files · syntax questions about dsl capability · advisory (`advise`) authoring, debugging, and re · setup/compile/validate/simulate/cast/resume work
Use cases
documentation
From the docs

What grimoire says it does

# Grimoire CLI Skill This skill is the base operating playbook for Grimoire.
SKILL.md
npx skills add https://github.com/franalgaba/grimoire --skill grimoire

Add your badge

Show developers this skill is listed on Skillselion. Paste this into your README.

Listed on Skillselion
Installs1.1k
repo stars6
Security audit1 / 3 scanners passed
Last updatedMay 8, 2026
Repositoryfranalgaba/grimoire

How do I use grimoire for the task described in its SKILL.md triggers?

Install and operate Grimoire, author .spell files with full syntax coverage (including advisory decision logic), and run compile/validate/simulate/cast safely. Use when users ask to create, edit, deb.

Who is it for?

Teams invoking grimoire when the user request matches documented triggers and prerequisites.

Skip if: Skip when cached docs are missing, the request is a negative trigger, or another sibling skill owns the workflow.

When should I use this skill?

Install and operate Grimoire, author .spell files with full syntax coverage (including advisory decision logic), and run compile/validate/simulate/cast safely. Use when users ask to create, edit, debug, validate, simulat

What you get

Step-by-step guidance grounded in grimoire documentation and reference files.

  • Forked local Anvil node
  • Spell simulation preview output

Files

SKILL.mdMarkdownGitHub ↗

Grimoire CLI Skill

This skill is the base operating playbook for Grimoire.

When To Use

Use this skill when the task includes:

  • install/setup of Grimoire tooling
  • creating or editing .spell files
  • syntax questions about DSL capability
  • advisory (advise) authoring, debugging, and replay workflows
  • setup/compile/validate/simulate/cast/resume workflows
  • debugging spell compile/runtime failures

Mandatory Loading Rules

All references/ and docs/ paths below are relative to this skill directory (skills/grimoire/). These rules are required — they solve syntax coverage gaps that cause authoring errors.

STOP — read this first: Do NOT search the codebase, grep for syntax patterns, or rely on memory for DSL syntax. The reference files bundled with this skill are the single source of truth. IMMEDIATELY use the Read tool on the files listed below before writing or editing any spell content.

1. For any .spell authoring/editing task — IMMEDIATELY read these files using the Read tool before doing anything else:

  • references/syntax-capabilities.md
  • references/authoring-workflow.md

2. For CLI flag details — IMMEDIATELY read using the Read tool:

  • references/cli-quick-reference.md

3. For any advisory task (advise, advisors, replay) — IMMEDIATELY read using the Read tool:

  • docs/how-to/use-advisory-decisions.md
  • docs/explanation/advisory-decision-flow.md

4. For local fork preview workflows — IMMEDIATELY read using the Read tool:

  • references/anvil-cheatsheet.md
  • docs/how-to/simulate-on-anvil-fork.md

5. For wallet setup and execution key flows — IMMEDIATELY read using the Read tool:

  • docs/how-to/use-wallet-commands-end-to-end.md

6. For RPC/signer/transaction diagnostics — IMMEDIATELY read using the Read tool:

  • references/cast-cheatsheet.md

Installation Resolution

Select the first working invocation and reuse it for the session.

1. Global:

  • npm i -g @grimoirelabs/cli
  • command prefix: grimoire

2. No-install:

  • command prefix: npx -y @grimoirelabs/cli

3. Repo-local:

  • command prefix: bun run packages/cli/src/index.ts --

If one path fails, move to the next path automatically.

If grimoire venue doctor ... fails with Unknown venue adapter "doctor", prefer repo-local invocation (bun run packages/cli/src/index.ts) or upgrade global CLI.

When using repo-local Bun execution, always keep the trailing -- so Bun forwards flags to Grimoire instead of consuming them.

Fast Start (Immediate Success Path)

Use this sequence before writing custom spells:

1. <grimoire-cmd> --help 2. <grimoire-cmd> setup (guided interactive execute onboarding) 3. <grimoire-cmd> validate spells/compute-only.spell 4. <grimoire-cmd> simulate spells/compute-only.spell --chain 1

If all three pass, proceed to spell authoring.

Setup security/runtime expectations:

  • setup prompts for hidden passwords and never echoes input
  • blank RPC input falls back to chain default public RPC
  • setup may write .grimoire/setup.env unless --no-save-password-env is used
  • CLI auto-loads nearest .grimoire/setup.env at startup without overriding existing env vars

Authoring and Execution Policy

1. Read syntax references first (mandatory rule above). 2. Author/update spell. 3. Run format to canonicalize layout before validation. 4. Run validate (use --strict for advisory-heavy spells). 5. Fix errors/warnings and re-run until validation passes. 6. Run simulate. 7. Before venue metadata queries or value-moving runs, execute venue doctor for the target adapter/chain.

  • Example: <grimoire-cmd> venue doctor --adapter uniswap --chain 1 --rpc-url <rpc> --json

8. Before value-moving runs on EVM venues, verify endpoint and signer state with Foundry Cast quickchecks (chain-id, block-number, balance, nonce).

  • Do not apply Anvil/Cast checks to offchain venues such as hyperliquid.

9. For advisory steps intended for deterministic execution, record and then use --advisory-replay <runId> in dry-run/live cast. 10. If spell includes irreversible actions, require cast --dry-run before any live cast. 11. Ask for explicit user confirmation before live value-moving cast. 12. For cross-chain mode, require explicit per-chain RPC mappings:

  • --rpc-url <sourceChainId>=<url>
  • --rpc-url <destinationChainId>=<url>

13. For cross-chain Morpho actions, require explicit market mapping via:

  • --morpho-market-id <actionRef>=<marketId> (repeatable), or
  • --morpho-market-map <path>

14. For Morpho supply-only strategies, prefer vault_deposit / vault_withdraw (explicit vault address).

  • If vault address is missing, list candidate vaults and require user to pick one; never auto-select.

15. For Morpho market strategies (borrow/collateral/lend), require explicit market_id and use explicit actions:

  • morpho_blue.supply_collateral(asset, amount, market_id)
  • morpho_blue.withdraw_collateral(asset, amount, market_id)

16. Use bare 0x... address literals in action token fields; quoted address-like strings trigger QUOTED_ADDRESS_LITERAL. 17. For Morpho doctor readiness checks, set wallet env explicitly (GRIMOIRE_WALLET_ADDRESS preferred, fallback WALLET_ADDRESS). 18. If a cross-chain run is left waiting, continue with resume <runId> (use --watch to poll settlement). 19. Never place passwords/private keys in agent prompts or inline command assignments. 20. Prefer keystore + --password-env over --private-key for dry-run/live casts. 21. Treat .grimoire/setup.env as plaintext secret material: keep local-only and rotate/remove when no longer needed. 22. For commands run outside the project tree, set GRIMOIRE_SETUP_ENV_FILE=/abs/path/to/.grimoire/setup.env when needed.

Command Surface (Core)

  • init
  • setup
  • format
  • compile
  • compile-all
  • validate
  • triggers
  • simulate
  • cast
  • venues
  • venue
  • venue doctor
  • history
  • log
  • resume
  • wallet (generate, address, balance, import, wrap, unwrap)

Use references/cli-quick-reference.md for concise command signatures and safety-critical flags.

Runtime Behavior Model

  • One runtime semantics: preview first, commit only for execute paths.
  • simulate and cast --dry-run are preview-only flows.
  • Live cast can commit irreversible actions when policy and runtime checks pass.
  • simulate supports explicit --rpc-url, with precedence: --rpc-url -> RPC_URL_<chainId> -> RPC_URL.
  • Phase 1 cross-chain execution uses two-spell orchestration (--destination-spell) with one logical run id and resume support.
  • Use triggers <spell> to discover stable handler ids natively before targeted execution.
  • simulate and cast support selected-trigger execution via --trigger-id, --trigger-index, and legacy --trigger.
  • For multi-handler spells, prefer --trigger-id; --trigger is label-based and can be ambiguous.
  • Cross-chain simulate / cast must forward the same selected trigger into per-chain execution.
  • In --json mode, parse stdout only; progress and spinner output can still appear on stderr.

Query Functions (price / balance / apy / metric)

Always prefer query functions over advisory for structured data fetching. These are deterministic, fast, and don't require LLM calls.

  • price(BASE, QUOTE[, SOURCE]) — live token price via query provider (requires Alchemy RPC URL)
  • balance(ASSET[, ADDRESS]) — on-chain token balance via RPC (any RPC URL)
  • apy(VENUE, ASSET[, SELECTOR]) — venue-backed APY surface (for example Aave and Morpho)
  • metric(SURFACE, VENUE[, ASSET[, SELECTOR]]) — generic protocol metric surface
  • Selector guidance:
  • market/vault id selector: apy(morpho, USDC, "wbtc-usdc-86")
  • vault selector (Morpho): metric("vault_net_apy", morpho, USDC, "vault=0xVaultAddress")
  • Morpho vault_apy / vault_net_apy require explicit selector; do not rely on implicit defaults.
  • key/value selector: metric("quote_out", uni_v3, USDC, "asset_out=WETH,amount=1000000,fee_tier=3000")
  • Never use an advisory (advise) just to fetch prices, balances, APYs, or other structured metrics

Use advisory only when the task requires LLM judgment, reasoning, or interpretation.

Advisory Operating Rules

  • Advisory must be explicit statement form: x = advise advisor: "prompt" { ... }.
  • Treat advisory outputs as typed contracts; enforce schema with output.
  • Require timeout and fallback in every advisory block.
  • Prefer validate --strict when advisory logic gates value-moving actions.
  • Use replay for determinism when moving from preview/dry-run to live execution.
  • For runtime debugging, use --advisory-trace-verbose (non-JSON mode) to stream detailed advisory traces.

Venue Metadata and Snapshots

Use venue skills for snapshot parameters and market metadata:

  • grimoire-aave
  • grimoire-uniswap
  • grimoire-morpho-blue
  • grimoire-hyperliquid
  • grimoire-pendle
  • grimoire-polymarket

Formatting policy for venue CLI output:

  • prefer --format json for automation or nested payloads
  • use --format table for human-readable summaries

References

  • references/syntax-capabilities.md
  • references/authoring-workflow.md
  • references/anvil-cheatsheet.md
  • references/cast-cheatsheet.md
  • references/cli-quick-reference.md
  • docs/how-to/simulate-on-anvil-fork.md
  • docs/how-to/use-wallet-commands-end-to-end.md
  • docs/how-to/use-advisory-decisions.md
  • docs/how-to/compare-protocol-metrics.md
  • docs/explanation/advisory-decision-flow.md
  • docs/reference/cli.md
  • docs/reference/spell-syntax.md
  • docs/reference/grimoire-dsl-spec.md
  • docs/reference/compiler-runtime.md

Related skills

How it compares

Use grimoire for fork-based EVM spell previews; use grimoire-aave when the task is reading Aave V3 market data rather than simulating transactions.

FAQ

What does grimoire do?

Install and operate Grimoire, author .spell files with full syntax coverage (including advisory decision logic), and run compile/validate/simulate/cast safely. Use when users ask to create, edit, debug, validate, simulat

When should I use grimoire?

Install and operate Grimoire, author .spell files with full syntax coverage (including advisory decision logic), and run compile/validate/simulate/cast safely. Use when users ask to create, edit, debug, validate, simulat

What are common prerequisites?

--- name: grimoire description: Install and operate Grimoire, author .spell files with full syntax coverage (including advisory decision logic), and run compile/validate/simulate/cast safely.

Is Grimoire safe to install?

skills.sh reports 1 of 3 security scanners passed. Review the Security Audits panel on this page before installing in production.

This week in AI coding

Five minutes, every Monday - the tools, releases and tactics for developers.

unsubscribe anytime.