Developer documentation
Build with the governance gate
The ONLY Lang verifier is open source. The WASM target runs in any browser. Here is how to get started, what every command does, and how to embed the gate in your application. For the full PrimeSwarm runtime (HITL, memory, SIEM), see the offer sheet.
Quickstart
Run the verifier locally in 5 minutes. Clone, build, run a test card.
ONLY Lang reference
Every command, its syntax, and what the gate does when it runs.
WASM embedding
Embed the verifier in a browser or Node.js app. 162KB, no server.
SDK guides
Rust, Python, and TypeScript SDKs for integrating the governance gate.
Free developer account
Get a free API key to run live probes against your own systems.
Quickstart
1. Clone and build the verifier
git clone https://github.com/vdmo/only-dgv-verifier.git cd only-dgv-verifier pip install -r requirements.txt
2. Run all 89 test cards
python3 dgv_runner.py --all
3. Run a single card and verify the receipt
python3 dgv_runner.py --card DGV-TC-001 python3 verify_receipt.py --receipt evidence/DGV-TC-001.receipt.json
4. Try it in the browser (no install)
Go to /try and run ONLY Lang scripts directly in your browser using the WASM verifier.
ONLY Lang command reference
| Command | Arguments | Description |
|---|---|---|
| harmony(tolerance) | tolerance: f64 | Set the equilibrium tolerance. The gate passes if the residual is below this value. Smaller = stricter. Default: 1e-12. |
| evolve(index) | index: usize | Evolve a field index toward equilibrium. The solver adjusts the value to minimize residual. |
| corrupt(index) | index: usize | Corrupt a field index to test detection. The gate should detect the corruption and heal it. |
| residual() | none | Check the current equilibrium residual. Returns the sum of squared deviations from equilibrium. |
| report() | none | Generate a text receipt of the evaluation. Includes gate state, residual, healed indices, and field values. |
| assert_bounds(min, max, idx) | min: f64, max: f64, idx: usize | Assert that field[idx] is within [min, max]. The gate denies if the assertion fails. |
| assert_bounds_named(name, min, max) | name: string, min: f64, max: f64 | Named bounds assertion for provenance tracking. |
| budget_limit(value) | value: f64 | Set a budget limit. The gate denies if the sum of field values exceeds this limit. |
| time_window(seconds) | seconds: f64 | Set a time window for rate limiting. Denies if too many operations occur within the window. |
| volatility_limit(value) | value: f64 | Set a volatility limit. Denies if field changes exceed this rate. |
| max_exposure(value) | value: f64 | Set a maximum exposure limit. Denies if cumulative exposure exceeds this value. |
| escalate(reason) | reason: string | Force escalation to human review. The gate returns ESCALATE and waits for HITL approval. |
| if_broken{ ... } | block: commands | Execute inner commands if equilibrium is broken. Used for conditional healing. |
| if_greater_than(idx, value, { ... }) | idx: usize, value: f64, block: commands | Execute inner commands if field[idx] > value. |
| bind(name, index) | name: string, index: usize | Bind a name to a field index for provenance tracking. |
| alias(index, name) | index: usize, name: string | Create an alias for a field index. |
| link_identity(id) | id: string | Link the evaluation to an identity for audit trail. |
| track_lineage(label) | label: string | Track lineage of the evaluation for provenance receipts. |
| generate_zk_proof(label) | label: string | Generate a zero-knowledge proof of correct evaluation (experimental). |
WASM embedding
The WASM build is 162KB and runs in any modern browser or Node.js. No server required. The API is a single function:
evaluate_script_wasm(script: string, field: Float64Array) → {
pass: boolean,
residual: number,
indices_healed: number[],
residual_history: number[],
report?: string,
field: Float64Array
}See the /try page for a live example. The source is at only-lang-wasm.
SDK guides
rust
use only_lang::evaluate_script;
let script = "harmony(1e-12)\nevaluate(0)\nresidual()\nreport()";
let mut field = vec![100.0, 200.0, 300.0];
let result = evaluate_script(&signs, &mut field, script)?;
println!("pass: {}, residual: {}", result.pass, result.residual);python
from only_lang import evaluate_script
script = "harmony(1e-12)\nevaluate(0)\nresidual()\nreport()"
field = [100.0, 200.0, 300.0]
result = evaluate_script(script, field)
print(f"pass: {result['pass']}, residual: {result['residual']}")typescript
import init, { evaluate_script_wasm } from "/wasm/only_lang_wasm.js";
await init("/wasm/only_lang_wasm_bg.wasm");
const script = "harmony(1e-12)\nevaluate(0)\nresidual()\nreport()";
const field = new Float64Array([100.0, 200.0, 300.0]);
const result = evaluate_script_wasm(script, field);
console.log("pass:", result.pass, "residual:", result.residual);Free developer account
Get a free API key to run live probes against your own AI systems. No credit card. No sales call. Just a key and the documentation.
What you get
- API key for the governance gate
- 30 live probes against your target system (real HTTP requests, real responses)
- PSR-001 receipts for every probe — download, verify, share
- Access to the ONLY Lang SDK (Rust, Python, TypeScript)
- DGV Auditor Toolkit desktop app (Windows, offline)
How to get started
# 1. Download the desktop app (Windows)
# https://only.institute/auditors
# 2. Connect to your target
# Open the app → Connect to target → Enter your URL + API key
# 3. Run 30 live probes
# The app sends real HTTP requests to your system and records responses
# 4. Download receipts
# Each probe produces a PSR-001 receipt — verify at /verify
# 5. Integrate via API
curl -X POST https://api.only.institute/v1/probes \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"target": "https://your-ai-system.example.com", "probe": "DGV-PROBE-001"}'Limits
- 30 probes per day on the free tier
- Receipts are stored locally — not on our servers
- No SLA — this is for development and testing, not production
- For production use, see the offer sheet