mohdel 1.5.0 → 1.6.0

package/README.md

@@ -355,7 +355,7 @@ With no session-bin configured, thin-gate runs in demo mode: `POST /v1/call` ret
 
 ### Calling from JS
 
-The client snippet under [Library Usage](#library-usage) above is the full surface: `call(envelope, { socketPath, signal? })` returns an async iterable of events. Pass an `AbortSignal` to cancel in flight; thin-gate forwards a cancel control message to the session and reuses it on the pool, and the client ends the stream with the same cancelled `done` the in-process path returns (status `incomplete`, warning `cancelled`, the partial output it relayed, zero tokens). The envelope is the flat `answer(prompt, options)` surface plus transport metadata (`callId`, `authId`, `auth.key`, optional `traceparent`); see [`js/core/envelope.js`](js/core/envelope.js) for the full field list.
+The client snippet under [Library Usage](#library-usage) above is the full surface: `call(envelope, { socketPath, signal?, headers? })` returns an async iterable of events. `headers` go with the request, for a router in front of the gate that authenticates its callers. Pass an `AbortSignal` to cancel in flight; thin-gate forwards a cancel control message to the session and reuses it on the pool, and the client ends the stream with the same cancelled `done` the in-process path returns (status `incomplete`, warning `cancelled`, the partial output it relayed, zero tokens). The envelope is the flat `answer(prompt, options)` surface plus transport metadata (`callId`, `authId`, `auth.key`, optional `traceparent`); see [`js/core/envelope.js`](js/core/envelope.js) for the full field list.
 
 ### Other languages
 

package/js/client/call.js

@@ -44,9 +44,11 @@ function cancelledDone (start, first, output) {
  * @param {string} options.socketPath
  * @param {AbortSignal} [options.signal]
  * @param {string} [options.path]  HTTP path; defaults to '/v1/call'
+ * @param {Record<string, string>} [options.headers]  sent with the request, for a router in front of the gate;
+ *   `content-type`, `content-length`, `transfer-encoding`, `connection` and `host` are the transport's
  * @returns {AsyncGenerator<import('#core/events.js').Event>}
  */
-export async function * call (envelope, { socketPath, signal, path = '/v1/call' }) {
+export async function * call (envelope, { socketPath, signal, path = '/v1/call', headers }) {
   const start = String(process.hrtime.bigint())
   if (signal?.aborted) {
     yield cancelledDone(start, null, '')
@@ -60,7 +62,8 @@ export async function * call (envelope, { socketPath, signal, path = '/v1/call'
       path,
       method: 'POST',
       body: envelope,
-      signal
+      signal,
+      headers
     })
   } catch (e) {
     if (signal?.aborted) {

package/js/client/call_embedding.js

@@ -19,15 +19,18 @@ import { MohdelError } from '#core'
  * @param {string} options.socketPath
  * @param {AbortSignal} [options.signal]
  * @param {string} [options.path]  HTTP path; defaults to '/v1/embed'
+ * @param {Record<string, string>} [options.headers]  sent with the request, for a router in front of the gate;
+ *   `content-type`, `content-length`, `transfer-encoding`, `connection` and `host` are the transport's
  * @returns {Promise<import('#core/embedding.js').EmbedResult>}
  */
-export async function callEmbedding (envelope, { socketPath, signal, path = '/v1/embed' }) {
+export async function callEmbedding (envelope, { socketPath, signal, path = '/v1/embed', headers }) {
   const res = await requestUnix({
     socketPath,
     path,
     method: 'POST',
     body: envelope,
-    signal
+    signal,
+    headers
   })
 
   const body = await readAll(res)

package/js/client/call_image.js

@@ -17,15 +17,18 @@ import { MohdelError } from '#core'
  * @param {string} options.socketPath
  * @param {AbortSignal} [options.signal]
  * @param {string} [options.path]  HTTP path; defaults to '/v1/image'
+ * @param {Record<string, string>} [options.headers]  sent with the request, for a router in front of the gate;
+ *   `content-type`, `content-length`, `transfer-encoding`, `connection` and `host` are the transport's
  * @returns {Promise<import('#core/image.js').ImageResult>}
  */
-export async function callImage (envelope, { socketPath, signal, path = '/v1/image' }) {
+export async function callImage (envelope, { socketPath, signal, path = '/v1/image', headers }) {
   const res = await requestUnix({
     socketPath,
     path,
     method: 'POST',
     body: envelope,
-    signal
+    signal,
+    headers
   })
 
   const body = await readAll(res)

package/js/client/call_transcription.js

@@ -20,15 +20,18 @@ import { MohdelError } from '#core'
  * @param {string} options.socketPath
  * @param {AbortSignal} [options.signal]
  * @param {string} [options.path]  HTTP path; defaults to '/v1/transcription'
+ * @param {Record<string, string>} [options.headers]  sent with the request, for a router in front of the gate;
+ *   `content-type`, `content-length`, `transfer-encoding`, `connection` and `host` are the transport's
  * @returns {Promise<import('#core/transcription.js').TranscriptionResult>}
  */
-export async function callTranscription (envelope, { socketPath, signal, path = '/v1/transcription' }) {
+export async function callTranscription (envelope, { socketPath, signal, path = '/v1/transcription', headers }) {
   const res = await requestUnix({
     socketPath,
     path,
     method: 'POST',
     body: envelope,
-    signal
+    signal,
+    headers
   })
 
   const body = await readAll(res)

package/js/session/run.js

@@ -85,6 +85,9 @@ export async function * run (envelope, {
   if (norm.error) { yield norm.error; return }
   envelope = norm.envelope
 
+  const cacheErr = cacheMarkerError(envelope.prompt)
+  if (cacheErr) { yield cacheErr; return }
+
   const provider = providerOf(envelope.model)
   const span = openSpan(envelope)
   const log = scopedLogger(logger, envelope, span)
@@ -357,6 +360,37 @@ export function normalizeModelId (envelope, resolveSpec) {
   return { envelope: next, key: base, spec: baseSpec }
 }
 
+/**
+ * A `cache` marker is honoured on `text` parts only, with a TTL the
+ * adapters know. The adapters would ignore any other marker, so it is
+ * refused rather than dropped.
+ *
+ * @param {import('#core/envelope.js').CallEnvelope['prompt']} prompt
+ * @returns {import('#core/events.js').ErrorEvent | undefined}
+ */
+function cacheMarkerError (prompt) {
+  if (!Array.isArray(prompt)) return undefined
+  for (const m of prompt) {
+    if (!Array.isArray(m?.content)) continue
+    for (const p of m.content) {
+      if (p?.cache == null) continue
+      if (p.type !== 'text') {
+        return errorEvent(
+          `\`cache\` is accepted on text parts only, not on a ${p.type} part (${m.role} message)`,
+          'SESSION_INVALID_PROMPT'
+        )
+      }
+      if (p.cache !== '5m' && p.cache !== '1h') {
+        return errorEvent(
+          `\`cache\` on a text part must be '5m' or '1h' (got ${JSON.stringify(p.cache)})`,
+          'SESSION_INVALID_PROMPT'
+        )
+      }
+    }
+  }
+  return undefined
+}
+
 /**
  * The two lane guards, in caller-then-internal order. Returns an
  * error event, or `undefined` when the lane is usable.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "mohdel",
-  "version": "1.5.0",
+  "version": "1.6.0",
   "license": "MIT",
   "author": {
     "name": "Christophe Le Bars",
@@ -136,7 +136,7 @@
     "@opentelemetry/exporter-trace-otlp-grpc": "^0.222.0",
     "@opentelemetry/sdk-node": "^0.222.0",
     "chalk": "^6.0.0",
-    "mohdel-thin-gate-linux-x64-gnu": "1.5.0"
+    "mohdel-thin-gate-linux-x64-gnu": "1.6.0"
   },
   "dependencies": {
     "@anthropic-ai/sdk": "^0.128.0",