Skip to content

gryszzz/Kaspa-Ai-Agent-Skill

Folders and files

NameName
Last commit message
Last commit date

Latest commit

Β 

History

79 Commits
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 
Β 

Repository files navigation

ChatGPT Image May 30, 2026, 10_28_23 PM (1)

🧠 Kaspa Ai Agent Skill

Kaspa researcher + engineering skill with cross-platform adapters for Codex, OpenAI-style agents, Claude, Cursor, OpenClaw, Gemini CLI, and generic LLM workflows.

πŸš€ What This Is

This repository publishes a reusable AI skill package for serious Kaspa research and engineering.

  • Skill ID: $kaspa-sovereign-architect-engine
  • Main skill file: SKILL.md
  • Release downloads: GitHub Releases
  • Published release: v1.6.2 - GitHub Releases and Packages
  • Repository skill version: v1.8.0 - deterministic sync and behavioral evaluation update, not yet published
  • Positioning: package-first distribution (not a website product)
  • Freshness model: current-aware, source-grounded, KIP-aware, and research-radar aware
  • Build scope: Kaspa protocol, wallet, indexer, WASM, dApp, Kdapp, and infrastructure work

πŸ”₯ What This Agent Is Good At

$kaspa-sovereign-architect-engine focuses on:

  • Kaspa research radar: economics, L1/L2, consensus, mining, KIPs, papers, and ecosystem engineering
  • Toccata R&D intelligence: branch-aware tracking for covenants, ZK opcodes, sequencing commitments, SilverScript, Based Apps, and vProgs
  • Toccata integration contracts: scheduled-versus-active DAA gating, storageMass/storage_mass, compute_commit, fee-policy separation, miner template preservation, and one-way DB migration
  • Protocol engineering: BlockDAG, GHOSTDAG, DAGKNIGHT status checks, mempool, UTXO semantics, and future-feature handling
  • KIP-aware analysis: separates research ideas, KIP proposals, merged code, activation, wallet/indexer support, and app usability
  • Indexer architecture: DAG-aware ordering, idempotent ingestion, dedupe, checkpoints, retries, sync lag, and query-layer separation
  • WASM + TypeScript apps: browser/Node separation, watch-only prototypes, wallet-provider signing, RPC trust assumptions
  • Wallet + UX safety: signing boundaries, custody warnings, phishing, UI spoofing, RPC hijacking, replay, and supply-chain risk
  • Backend and infrastructure reliability: APIs, workers, retries, caching, validation, observability, deployment, and recovery plans

⚑ Quick Start (Codex)

Use the skill name in your prompt:

$kaspa-sovereign-architect-engine
Design a production-ready Kaspa indexer + API with retry-safe workers and monitoring.

Or embed it directly:

Use $kaspa-sovereign-architect-engine to audit my wallet backend for replay, nonce, and signing risks.

πŸ›° Example Prompts

Use $kaspa-sovereign-architect-engine in Research Radar mode. Check what is new across Kaspa Research categories and identify builder-relevant changes.
Use $kaspa-sovereign-architect-engine in Engineering Build Plan mode. Design a testnet-first Kaspa WASM wallet dashboard with watch-only mode and safe signing boundaries.
Use $kaspa-sovereign-architect-engine in Deep Protocol Audit mode. Audit a Kaspa indexer design for DAG-aware ordering, idempotency, replay handling, and sync lag.
Use $kaspa-sovereign-architect-engine in KIP Status mode. Build a table of relevant KIPs, their activation status, builder impact, and what must be verified.
Use $kaspa-sovereign-architect-engine in Toccata R&D Intelligence mode. Re-check Rusty Kaspa v2.0.0, current mainnet DAA versus activation DAA 474165565, recent branch-delta impact lanes, KIP-16/17/20/21, and produce the next covenant/indexer/wallet compatibility plan.

🧩 Compatibility Matrix

Platform Status Adapter
Codex βœ… Native SKILL.md
OpenAI-style agents βœ… Adapter agents/openai.yaml
Claude / Anthropic βœ… Adapter agents/anthropic.md
Cursor βœ… Adapter agents/cursor.mdc
OpenClaw βœ… Adapter agents/openclaw.md
Gemini CLI βœ… Adapter agents/gemini.md
Any LLM platform βœ… Adapter agents/generic-system-prompt.md

Compatibility metadata:

Automated verification:

Toccata source monitoring:

πŸ›  Install

Option A: Install latest release (recommended)

mkdir -p "$CODEX_HOME/skills"
curl -L -o /tmp/kaspa-sovereign-architect-engine.zip \
  https://github.com/gryszzz/Kaspa-Ai-Agent-Skill/releases/latest/download/kaspa-sovereign-architect-engine.zip
unzip -o /tmp/kaspa-sovereign-architect-engine.zip -d "$CODEX_HOME/skills"

Verify artifact integrity:

curl -L -o /tmp/SHA256SUMS.txt \
  https://github.com/gryszzz/Kaspa-Ai-Agent-Skill/releases/latest/download/SHA256SUMS.txt
(cd /tmp && grep "kaspa-sovereign-architect-engine.zip$" SHA256SUMS.txt | shasum -a 256 -c -)

Alternative tarball install:

curl -L -o /tmp/kaspa-sovereign-architect-engine.tar.gz \
  https://github.com/gryszzz/Kaspa-Ai-Agent-Skill/releases/latest/download/kaspa-sovereign-architect-engine.tar.gz
tar -xzf /tmp/kaspa-sovereign-architect-engine.tar.gz -C "$CODEX_HOME/skills"

Option B: Pull from GitHub Packages

The currently published skill is also available as an OCI artifact in GitHub Packages:

oras pull ghcr.io/gryszzz/kaspa-sovereign-architect-engine:v1.6.2

The package contains the versioned ZIP, versioned tarball, and SHA256SUMS.txt. Use the GitHub Release ZIP install path when you want the simplest Codex installation flow.

Option C: Install from source

mkdir -p "$CODEX_HOME/skills/public"
cp -R skills/public/kaspa-sovereign-architect-engine "$CODEX_HOME/skills/public/"

Option D: Install scripts

macOS/Linux:

./skills/public/kaspa-sovereign-architect-engine/scripts/install-codex.sh

Windows PowerShell:

.\skills\public\kaspa-sovereign-architect-engine\scripts\install-codex.ps1

Export adapter bundle:

./skills/public/kaspa-sovereign-architect-engine/scripts/export-adapters.sh

Safe repository/local synchronization

The repository skill directory is canonical. Preview deterministic drift before copying in either direction:

node skills/public/kaspa-sovereign-architect-engine/scripts/sync-local-skill.mjs \
  --check --report /tmp/kaspa-skill-drift.json

Dry-run is the default. Use --to-local or --from-local to select a direction, and add --apply only after reviewing the report. Conflicts block writes unless --force is explicit; use --backup when resolving them. See local-skill-sync.md for the full workflow.

Option E: Install for OpenClaw

OpenClaw loads AgentSkills-compatible folders from ~/.openclaw/skills or <workspace>/skills.

Install globally for OpenClaw (macOS/Linux):

./skills/public/kaspa-sovereign-architect-engine/scripts/install-openclaw.sh

Install manually into an OpenClaw workspace:

mkdir -p ./skills
cp -R skills/public/kaspa-sovereign-architect-engine ./skills/

Option F: Install for Gemini CLI

Gemini CLI loads context from ~/.gemini/GEMINI.md and supports @ imports.

Install adapter globally (macOS/Linux):

./skills/public/kaspa-sovereign-architect-engine/scripts/install-gemini.sh

Manual install:

mkdir -p ~/.gemini
cp skills/public/kaspa-sovereign-architect-engine/agents/gemini.md ~/.gemini/kaspa-sovereign-architect-engine.gemini.md
printf "\n@%s\n" "$HOME/.gemini/kaspa-sovereign-architect-engine.gemini.md" >> ~/.gemini/GEMINI.md

🌍 Use Outside Codex

  1. Open an adapter file from agents/.
  2. Copy its contents into your platform's system/developer instructions.
  3. Add your actual task prompt.

Example task prompt:

Design a Kaspa DAG-aware indexer for 100k users with failure recovery and replay-safe event handling.

πŸ“¦ Releases

🌐 Deploy + Marketing

πŸ“ Repository Map