/SKILL.md
SUTRA — A minimal, deterministic agent-to-agent communication language with 8 Sanskrit-inspired keywords (INTENT, FACT, QUERY, OFFER, ACCEPT, REJECT, COMMIT, ACT). Use when agents need to negotiate deals, query knowledge, commit to obligations with Ed25519 cryptographic signatures, or coordinate multi-agent workflows. Includes HTTP transport, multi-agent runtime with auto-response, and conversation tracking. Alternative to unstructured English or rigid JSON APIs for agent communication.
npx skillsauth add subhajitlucky/sutra sutraInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
SUTRA (Structured Universal Transaction & Reasoning Architecture) is a minimal, deterministic, formally-specified language that enables AI agents to communicate, negotiate, and commit to agreements. Sanskrit-inspired ontology, English-readable syntax.
Use SUTRA when you need agents to:
Use instead of unstructured English (ambiguous) or raw JSON APIs (no semantic meaning).
| Keyword | Sanskrit | What It Does |
|-----------|------------|---------------------------------------|
| INTENT | Sankalpa | Declare what the agent wants |
| FACT | Pramana | Assert verified knowledge |
| QUERY | Prashna | Ask another agent for information |
| OFFER | Samvida | Propose a deal with terms |
| ACCEPT | Svikrti | Accept an offer |
| REJECT | Nirakrti | Reject an offer (with reason) |
| COMMIT | Dharma | Bind to an obligation (signed) |
| ACT | Kriya | Execute an action |
// Buyer agent declares intent
INTENT buy(item="SmartTV", max_price=50000);
// Buyer queries seller
QUERY available(item="SmartTV") FROM "seller@store";
// Buyer makes an offer
OFFER id="deal-001" TO "seller@store" {
give: {money: 48000},
want: {item: "SmartTV", brand: "LG"}
};
// After acceptance — buyer commits (cryptographically signed)
COMMIT pay(amount=48000, to="seller@store") BY "2026-03-02";
// Execute the payment
ACT transfer(amount=48000, to="seller@store");
When your agent wants to communicate, produce valid SUTRA:
OFFER id="task-001" TO "worker@gpu-cluster" {
task: "train_model",
dataset: "imagenet-1k",
budget_usd: 50
};
SUTRA has a formal grammar (8 keywords, under 30 rules). Any compliant parser can read it.
When Agent A sends QUERY available(item="GPU") FROM "provider@cloud";, the runtime auto-checks Agent B's knowledge base and responds with matching FACT statements.
Register evaluation logic and the runtime auto-accepts/rejects:
def my_evaluator(agent, offer_id, from_agent, fields):
if fields["budget_usd"] >= 40:
return "accept"
return "reject:budget_too_low"
COMMITs and OFFERs are Ed25519-signed — tamper-proof, verifiable:
[COMMIT] pay(amount=48000) BY 2026-03-02 🔏 ed25519:a1b2c3d4...
┌─────────────────────────────────────────┐
│ SUTRA v0.4 Stack │
├─────────────────────────────────────────┤
│ Runtime — Multi-agent process │ ← v0.4: agents talk directly
│ Transport — HTTP server/client │ ← v0.2: agents talk over network
│ Crypto — Ed25519 sign/verify │ ← v0.3: tamper-proof commitments
│ Interpreter — AST executor │ ← v0.1: deterministic execution
│ Parser — Recursive descent │ ← v0.1: formal grammar
│ Lexer — Tokenizer │ ← v0.1: 8 keywords
│ Agent — State model (BDI-lite) │ ← v0.1: beliefs, goals, offers
└─────────────────────────────────────────┘
pip install sutra-lang
# or clone:
git clone https://github.com/subhajitlucky/sutra.git
cd sutra && python -m sutra demo
python -m sutra demo # Local negotiation demo
python -m sutra network-demo # HTTP-based multi-agent demo
python -m sutra signed-demo # Cryptographically signed demo
python -m sutra runtime-demo # Multi-agent runtime demo (v0.4)
python -m sutra run file.sutra # Execute a .sutra file
python -m sutra keygen "agent-id" # Generate signing keys
spec/SUTRA_SPEC.md — full language specificationspec/GRAMMAR.ebnf — formal EBNF grammarexamples/ — buyer, seller, knowledge sharing scripts| Feature | English | JSON API | SUTRA | |----------------------|---------|----------|-------| | Human-readable | ✅ | ❌ | ✅ | | Machine-executable | ❌ | ✅ | ✅ | | Deterministic | ❌ | ✅ | ✅ | | Expresses intent | ✅ | ❌ | ✅ | | Negotiation-native | ❌ | Partial | ✅ | | Cryptographically signed | ❌ | ❌ | ✅ | | Auditable transcript | ❌ | Partial | ✅ |
development
Maintainer-only workflow for handling GitHub Secret Scanning alerts on OpenClaw. Use when Codex needs to triage, redact, clean up, and resolve secret leakage found in issue comments, issue bodies, PR comments, or other GitHub content.
development
Maintainer workflow for OpenClaw releases, prereleases, changelog release notes, and publish validation. Use when Codex needs to prepare or verify stable or beta release steps, align version naming, assemble release notes, check release auth requirements, or validate publish-time commands and artifacts.
development
Run, watch, debug, and extend OpenClaw QA testing with qa-lab and qa-channel. Use when Codex needs to execute the repo-backed QA suite, inspect live QA artifacts, debug failing scenarios, add new QA scenarios, or explain the OpenClaw QA workflow. Prefer the live OpenAI lane with regular openai/gpt-5.4 in fast mode; do not use gpt-5.4-pro or gpt-5.4-mini unless the user explicitly overrides that policy.
development
End-to-end Parallels smoke, upgrade, and rerun workflow for OpenClaw across macOS, Windows, and Linux guests. Use when Codex needs to run, rerun, debug, or interpret VM-based install, onboarding, gateway smoke tests, latest-release-to-main upgrade checks, fresh snapshot retests, or optional Discord roundtrip verification under Parallels.