LAWDE Shield · Documentation

HTTP / Direct API

Language-agnostic Sentinel ingest. Any runtime that can POST JSON over HTTPS can send crashes. Use cURL for a smoke test, then call the same endpoint from your main app’s global error path for live monitoring.

What is the Sentinel HTTP ingest API?

A single authenticated POST /api/v1/logs/ingest endpoint that accepts scrubbed crash JSON, runs built-in privacy scrubbing (<50ms design target), and queues async clustering/diagnosis. It never accepts a repository upload — 0 bytes durable source.

Stage	Timing
Built-in privacy scrub	<50ms design target (hot path)
Atlas / Athena / Aegis	Async after ACK (≥0.70 to surface fixes)

1. Endpoint & authentication

URL: POST https://lawde.net/api/v1/logs/ingest
Required header: x-lawde-api-key: $LAWDE_API_KEY
Also accepted: Authorization: Bearer $LAWDE_API_KEY
Content-Type: application/json

2. JSON payload contract

Field	Required	Notes
`error_class`	Yes	Exception name, e.g. `TypeError`
`error_message`	Yes	Human-readable message
`stack_trace`	Yes	Full stack string (scrubbed server-side)
`file_path`	Yes	Primary file for clustering signature
`line_number`	Yes	Positive integer
`environment`	No	Defaults to `production`
`app_id`	No	UUID of a workspace app
`column_number`	No	Non-negative integer
`metadata`	No	Object — may include `lawdeignore_content`

Repeats of the same crash join one issue. Look-alikes join when similarity is at least 0.85.

3. Phase A — cURL smoke test

Run once from your laptop or CI to confirm the key works. This is the same idea as test_ingest.sh in the Apps hub.

cURL

curl -X POST https://lawde.net/api/v1/logs/ingest \
  -H "Content-Type: application/json" \
  -H "x-lawde-api-key: $LAWDE_API_KEY" \
  -d '{
    "error_class": "TypeError",
    "error_message": "Cannot read property of null",
    "stack_trace": "TypeError: Cannot read property of null\n    at handler (src/api/route.ts:42:11)",
    "file_path": "src/api/route.ts",
    "line_number": 42,
    "environment": "production",
    "app_id": "'"$LAWDE_APP_ID"'"
  }'

Bearer alternative

cURL · Bearer

curl -X POST https://lawde.net/api/v1/logs/ingest \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $LAWDE_API_KEY" \
  -d '{
    "error_class": "RuntimeError",
    "error_message": "System memory allocation exceeded",
    "stack_trace": "RuntimeError: Memory limit\n  at runner.ts:88:12",
    "file_path": "src/core/runner.ts",
    "line_number": 88,
    "environment": "production"
  }'

4. Phase B — Call the same endpoint from your MAIN app

Identify where your runtime surfaces unhandled errors.
On each error, build the JSON fields above from the exception object / stack parser.
POST asynchronously (or with a short timeout) and swallow network failures so reporting never crashes the host.
Register that hook at process boot so it survives for the life of the service.

Live wire pattern

// Pseudocode — put this inside your MAIN app's global error path
// (Express error middleware, FastAPI exception_handler, Go recover(), etc.)

function onUnhandledError(err) {
  // 1) Extract class, message, stack, file, line from your runtime
  // 2) POST to Sentinel (never throw if this fails)
  fetch("https://lawde.net/api/v1/logs/ingest", {
    method: "POST",
    headers: {
      "Content-Type": "application/json",
      "x-lawde-api-key": process.env.LAWDE_API_KEY,
    },
    body: JSON.stringify({
      error_class: err.name || "Error",
      error_message: err.message || String(err),
      stack_trace: err.stack || String(err),
      file_path: err.file || "unknown",
      line_number: err.line || 1,
      environment: process.env.NODE_ENV || "production",
      app_id: process.env.LAWDE_APP_ID, // optional
    }),
  }).catch(() => {});
}

// Register ONCE at boot — examples:
// process.on("uncaughtException", onUnhandledError)
// app.use(errorMiddleware)
// @app.exception_handler(Exception)

Ready-made helpers: JavaScript agent (auto hooks + scrubbing) · Python FastAPI/Flask/Django.

5. Success response

Successful ingest returns HTTP 200. Example body:

JSON

{
  "success": true,
  "status": "ingested",
  "signature_hash": "…",
  "cluster_id": "…",
  "log_id": "…",
  "deduplicated": false,
  "is_restricted_zone": false,
  "ai_queued": false,
  "execution_ms": 42,
  "fail_open": false
}

deduplicated: true — this crash already had an issue; the occurrence count increased
ai_queued: false — Athena never auto-runs on ingest; click Diagnose Issue in the dashboard to spend one AI credit
is_restricted_zone: true — Nuclear Abort matched .lawdeignore

6. Rate limits & tips

Hobby 50 req/min · Pro 250 · Business 500 · Enterprise 1,000 per workspace
Responses include X-RateLimit-Remaining when available
Prefer the JS agent when you want automatic scrubbing + global hooks; use this HTTP contract for every other language

LAWDE Shield · Documentation

HTTP / Direct API

What is the Sentinel HTTP ingest API?

Stage	Timing
Built-in privacy scrub	<50ms design target (hot path)
Atlas / Athena / Aegis	Async after ACK (≥0.70 to surface fixes)

1. Endpoint & authentication

URL: POST https://lawde.net/api/v1/logs/ingest
Required header: x-lawde-api-key: $LAWDE_API_KEY
Also accepted: Authorization: Bearer $LAWDE_API_KEY
Content-Type: application/json

2. JSON payload contract

Field	Required	Notes
`error_class`	Yes	Exception name, e.g. `TypeError`
`error_message`	Yes	Human-readable message
`stack_trace`	Yes	Full stack string (scrubbed server-side)
`file_path`	Yes	Primary file for clustering signature
`line_number`	Yes	Positive integer
`environment`	No	Defaults to `production`
`app_id`	No	UUID of a workspace app
`column_number`	No	Non-negative integer
`metadata`	No	Object — may include `lawdeignore_content`

Repeats of the same crash join one issue. Look-alikes join when similarity is at least 0.85.

3. Phase A — cURL smoke test

Run once from your laptop or CI to confirm the key works. This is the same idea as test_ingest.sh in the Apps hub.

cURL

curl -X POST https://lawde.net/api/v1/logs/ingest \
  -H "Content-Type: application/json" \
  -H "x-lawde-api-key: $LAWDE_API_KEY" \
  -d '{
    "error_class": "TypeError",
    "error_message": "Cannot read property of null",
    "stack_trace": "TypeError: Cannot read property of null\n    at handler (src/api/route.ts:42:11)",
    "file_path": "src/api/route.ts",
    "line_number": 42,
    "environment": "production",
    "app_id": "'"$LAWDE_APP_ID"'"
  }'

Bearer alternative

cURL · Bearer

curl -X POST https://lawde.net/api/v1/logs/ingest \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $LAWDE_API_KEY" \
  -d '{
    "error_class": "RuntimeError",
    "error_message": "System memory allocation exceeded",
    "stack_trace": "RuntimeError: Memory limit\n  at runner.ts:88:12",
    "file_path": "src/core/runner.ts",
    "line_number": 88,
    "environment": "production"
  }'

4. Phase B — Call the same endpoint from your MAIN app

Identify where your runtime surfaces unhandled errors.
On each error, build the JSON fields above from the exception object / stack parser.
POST asynchronously (or with a short timeout) and swallow network failures so reporting never crashes the host.
Register that hook at process boot so it survives for the life of the service.

Live wire pattern

// Pseudocode — put this inside your MAIN app's global error path
// (Express error middleware, FastAPI exception_handler, Go recover(), etc.)

function onUnhandledError(err) {
  // 1) Extract class, message, stack, file, line from your runtime
  // 2) POST to Sentinel (never throw if this fails)
  fetch("https://lawde.net/api/v1/logs/ingest", {
    method: "POST",
    headers: {
      "Content-Type": "application/json",
      "x-lawde-api-key": process.env.LAWDE_API_KEY,
    },
    body: JSON.stringify({
      error_class: err.name || "Error",
      error_message: err.message || String(err),
      stack_trace: err.stack || String(err),
      file_path: err.file || "unknown",
      line_number: err.line || 1,
      environment: process.env.NODE_ENV || "production",
      app_id: process.env.LAWDE_APP_ID, // optional
    }),
  }).catch(() => {});
}

// Register ONCE at boot — examples:
// process.on("uncaughtException", onUnhandledError)
// app.use(errorMiddleware)
// @app.exception_handler(Exception)

Ready-made helpers: JavaScript agent (auto hooks + scrubbing) · Python FastAPI/Flask/Django.

5. Success response

Successful ingest returns HTTP 200. Example body:

JSON

{
  "success": true,
  "status": "ingested",
  "signature_hash": "…",
  "cluster_id": "…",
  "log_id": "…",
  "deduplicated": false,
  "is_restricted_zone": false,
  "ai_queued": false,
  "execution_ms": 42,
  "fail_open": false
}

deduplicated: true — this crash already had an issue; the occurrence count increased
ai_queued: false — Athena never auto-runs on ingest; click Diagnose Issue in the dashboard to spend one AI credit
is_restricted_zone: true — Nuclear Abort matched .lawdeignore

6. Rate limits & tips

Hobby 50 req/min · Pro 250 · Business 500 · Enterprise 1,000 per workspace
Responses include X-RateLimit-Remaining when available
Prefer the JS agent when you want automatic scrubbing + global hooks; use this HTTP contract for every other language

What is the Sentinel HTTP ingest API?

1. Endpoint & authentication

2. JSON payload contract

3. Phase A — cURL smoke test

Bearer alternative

4. Phase B — Call the same endpoint from your MAIN app

5. Success response

6. Rate limits & tips

Preparing your secure workspace…

What is the Sentinel HTTP ingest API?

1. Endpoint & authentication

2. JSON payload contract

3. Phase A — cURL smoke test

Bearer alternative

4. Phase B — Call the same endpoint from your MAIN app

5. Success response

6. Rate limits & tips