Domain knowledge and workflow automation for Pact 5 / KDA-CE smart contract development — packaged for Claude Code, Codex, and Gemini CLI.
- Claude Code, Codex, or Gemini CLI — see Tool Support for per-host details
- git and bash — only needed for the curl-pipe installer
- Pact 5.4ce — the KDA-CE fork of kadena-io/pact-5
Claude Code
claude plugins add Pact-Community-Organization/pact-kitCodex
codex plugins add Pact-Community-Organization/pact-kitGemini CLI
gemini extension install https://github.com/Pact-Community-Organization/pact-kitAny host — installs directly into ~/.claude/
bash <(curl -fsSL https://raw.githubusercontent.com/Pact-Community-Organization/pact-kit/main/scripts/install.sh)The installer merges into ~/.claude/ and never removes existing files. Safe to run again after updates.
After installing, open a Pact project in Claude Code:
/new-pact-module scaffold a new module with REPL tests and a deploy script
/validate-pact-module run the 4-phase validation: REPL analysis → devnet deploy → verify
/gas-analysis measure gas per function against the 150k ceiling
security review invoke the independent auditor before shipping any module
Copy the starter configuration and fill in the identity placeholder:
cp ~/.claude/CLAUDE.md.template ~/.claude/CLAUDE.mdThen add the hooks snippet from ~/.claude/CLAUDE.md.template to ~/.claude/settings.json
to enable automatic static analysis on every .pact and .repl edit.
25 domain skills covering the full Pact 5 / KDA-CE surface. Loaded on demand — your AI assistant draws on precise, current knowledge for the task without carrying it in every session.
| Area | What it covers |
|---|---|
| Core language | Capabilities, guards, schema design, module architecture, interfaces, defpacts, events, invariants |
| Testing & validation | REPL test patterns, devnet workflows, 4-phase module validation, CLI tooling, debugging |
| Security & correctness | Security review, capability audits, fungible-v2 / xchain-v1 compliance, formal verification, offensive red-team testing |
| Gas & cross-chain | Gas analysis against the 150k ceiling, gas station design, cross-chain transfer patterns |
| Platform | KDA-CE network compliance, devnet lifecycle management |
21 commands covering the full development lifecycle. Type /command-name in Claude Code.
| Command | What it does |
|---|---|
/new-pact-module |
Scaffold module + REPL tests + deploy script |
/design-defpact |
defpact steps, yield/resume, SPV, rollback, and test plan |
/generate-repl-tests |
REPL test suite from an ADR or acceptance criteria |
/generate-integration-stubs |
TypeScript @kadena/client stubs for a module |
/validate-pact-module |
4-phase validation: REPL analysis → REPL exec → devnet deploy → verify |
/verify-pact-module |
Formal verification and typecheck |
/capability-audit |
Capability hierarchy map with bypass path analysis |
/pact-security-audit |
Security-focused code review checklist |
/full-security-audit |
5-phase audit (in-session; use pact-auditor for pre-ship reviews) |
/red-team |
Executable offensive pass: runs real .repl attacks and proves HELD/BROKEN |
/security-assessment |
STRIDE per public function with attack simulation |
/threat-model |
STRIDE threat model with attack trees and mitigations |
/analyze-feature |
7-step impact analysis before implementing a feature |
/architecture-lockdown |
Architecture freeze: options → decision → ADR outputs |
/architecture-review |
Review an ADR or proposal for gas, DAG acyclicity, KDA-CE compliance |
/design-test-suite |
Test suite design from ADRs and acceptance criteria |
/deploy-to-devnet |
Ordered deploy with gas tracking and a verification report |
/gas-analysis |
Gas per function vs. 150k ceiling with baseline comparison |
/migrate-pact-schema |
New table, admin migration functions, bless list, row-count test |
/developer-handoff |
Implementation spec: schemas, capabilities, gas estimate, deploy notes |
/review-pr |
Dual-scope review: isolated diff + full regression |
16 instruction files that load automatically based on the files you're editing. Opening a
.pact or .repl file activates Pact language rules, trap avoidance, and security checklists.
Deployment, testing, gas, and refactoring rules activate for their respective contexts.
Two fresh-context agents with no implementation history, one defensive and one offensive.
pact-auditor reviews. Invoke it by saying security review or ready to ship in
Claude Code. It runs a 5-step protocol — static analysis, capability audit, Pact 5.4ce trap
check, security checklist, and STRIDE per public function — and returns a structured finding
table.
red-team-attacker attacks. Given one module and one front, it writes and executes real
.repl attacks (20 mutations per front) with the pact CLI, applies the REPL-artifact
filter, and reports HELD or BROKEN with reproducible proof — the offensive counterpart driven
by the /red-team command.
examples/example-token.pact + examples/example-token.repl — a runnable, CI-verified
reference module demonstrating the conventions the kit enforces (capability layering,
managed caps, principal discipline, trust-boundary validation, node-safe enforce patterns).
docs/reference-repos.md — the vetted primary sources (language repo, official docs, pact-util-lib, production contracts) and the idioms they agree on. When in doubt, primary sources beat web search results and model memory.
pact-static-check.sh is a two-tier static analysis gate (Pact CLI + semantic greps).
Configure it as a PostToolUse hook to run automatically on every .pact and .repl edit.
session-end-secrets-scan.sh scans modified files for credential patterns at session end.
The hooks configuration snippet is in ~/.claude/CLAUDE.md.template.
Open an issue or discussion before submitting a PR. See CONTRIBUTING.md.
Apache-2.0 License · Changelog · Security Policy · Code of Conduct