specpi 0.29.0 → 0.30.1

package/extensions/jev-advisor/consent.mjs

@@ -20,6 +20,12 @@ import { MAX_STATE_BYTES } from "./sanitize.mjs";
  * and a TYPESAFE_BASE_URL override) names itself, and so the rule below is true rather than
  * aspirational: a grant is keyed on this string, so repointing the client really does ask again.
  */
+export const CONSENT_SCHEMA = 2;
+
+export function endpointOrigin() {
+    return new URL(baseUrl()).origin;
+}
+
 export function endpointLabel() {
     const base = baseUrl();
     try {
@@ -43,12 +49,12 @@ export function loadConsent() {
         }
 
         const stored = JSON.parse(fs.readFileSync(file, "utf8"));
-        if (stored?.schema !== 1 || stored.granted !== true || typeof stored.endpoint !== "string") {
+        if (stored?.schema !== CONSENT_SCHEMA || stored.granted !== true || typeof stored.endpoint !== "string") {
             return undefined;
         }
 
         // A grant is for the endpoint it was given for. Repointing the client asks again.
-        return stored.endpoint === endpointLabel() ? stored : undefined;
+        return stored.endpoint === endpointLabel() && stored.origin === endpointOrigin() ? stored : undefined;
     } catch {
         return undefined;
     }
@@ -65,9 +71,11 @@ export function saveConsent() {
     }
 
     const stored = {
-        schema: 1,
+        // Earlier consent incorrectly excluded file/command samples. It cannot authorize them.
+        schema: CONSENT_SCHEMA,
         granted: true,
         endpoint: endpointLabel(),
+        origin: endpointOrigin(),
         maxStateBytes: MAX_STATE_BYTES,
         grantedAt: new Date().toISOString(),
     };
@@ -89,15 +97,19 @@ export function consentPath() {
     return consentFile();
 }
 
-export const CONSENT_TITLE = "Allow SpecPi to send task summaries to Jev?";
+export const CONSENT_TITLE = "Allow SpecPi to send task summaries and text samples to Jev?";
 
 export function consentBody(systemLabel) {
     return [
         `${systemLabel} wants to ask TypeSafe's Jev classifier a question about this session.`,
         "",
-        `What is sent: a summary object of at most ${MAX_STATE_BYTES} bytes to ${endpointLabel()}, over HTTPS.`,
-        "It carries tool names, byte counts, relative paths and short descriptions.",
-        "It never carries file contents, command output, credentials, URLs or session history.",
+        `What is sent: a state object of at most ${MAX_STATE_BYTES} bytes to ${endpointLabel()}, with classifier questions.`,
+        `Transport: ${new URL(baseUrl()).protocol === "https:" ? "HTTPS" : "not HTTPS (configured endpoint override)"}.`,
+        "It may include the current request or objective, relative paths, capability-gap summaries,",
+        "and short sampled lines from file contents, command output and fetched content.",
+        "Known credential, email, URL and outside-workspace path patterns are redacted before sending.",
+        "Redaction is best effort, not a guarantee that every sensitive detail is removed. No session history is read.",
+        "Collection stays local unless you separately allow this advisor transmission.",
         "",
         "Every call is recorded locally in transmissions.jsonl with a hash of exactly what was sent,",
         "which you can read with /jev ledger. Turn this off at any time with /jev off.",

package/extensions/jev-advisor/gate.mjs

@@ -50,12 +50,11 @@
 //    was unreachable, and running the layer could never have revealed it, because a system that
 //    never fires looks exactly like a system whose advice was always to do nothing.
 //
-//    compaction's `unresolved_thread` had the same problem at high 0.85: the clearest open
-//    investigation the fixture can express scores 0.63-0.65. It is lowered to 0.60, which is
-//    defensible only because of what that branch does -- add one sentence to a summariser prompt
-//    that is being rebuilt from scratch anyway. It is the cheapest action in the layer, so it can
-//    afford the loosest gate. gap keeps 0.85 because its Noul reaches 0.96 on the case that matters
-//    and because a firing there blocks a write.
+//    Compaction guidance had the same problem and was lowered to 0.60 to clear it. That system has
+//    since been withdrawn -- two tier-6 runs measured the arm carrying it solving fewer long-session
+//    tasks than plain SpecPi -- so the loosest gate in the layer is gone with it. gap keeps 0.85
+//    because its Noul reaches 0.96 on the case that matters and because a firing there blocks a
+//    write.
 
 /**
  * The figures the comment above cites, in a form a test can check against the artifact. A citation
@@ -91,16 +90,6 @@ export const THRESHOLDS = Object.freeze({
         // Measured 0.06-0.07 on the spent case, so this clears with room.
         low: 0.1,
     }),
-    compaction: Object.freeze({
-        scoreConfidence: 0.6,
-        boundary: 0.3,
-        choiceConfidence: 0.75,
-        margin: 0.25,
-        // Lowered from 0.85: the clearest open investigation scores 0.63-0.65, and the action is
-        // one sentence added to a prompt that is being rebuilt regardless.
-        high: 0.6,
-        low: 0.15,
-    }),
     gap: Object.freeze({
         scoreConfidence: 0.6,
         boundary: 0.3,
@@ -178,14 +167,14 @@ export function thresholdsFor(system) {
 export function nounTrue(answer, system) {
     const limits = thresholdsFor(system);
 
-    return answer?.kind === "noul" && answer.value >= limits.high;
+    return answer?.kind === "noul" && Number.isFinite(answer.value) && answer.value >= limits.high && answer.value <= 1;
 }
 
 /** True when the Noul is confidently no. Not the negation of nounTrue: the middle band is silence. */
 export function nounFalse(answer, system) {
     const limits = thresholdsFor(system);
 
-    return answer?.kind === "noul" && answer.value <= limits.low;
+    return answer?.kind === "noul" && Number.isFinite(answer.value) && answer.value >= 0 && answer.value <= limits.low;
 }
 
 function topTwo(probabilities) {
@@ -204,7 +193,15 @@ function topTwo(probabilities) {
  * the margin is a live test rather than a branch that silently never runs.
  */
 export function choiceValue(answer, system) {
-    if (answer?.kind !== "choice" || typeof answer.confidence !== "number") {
+    if (
+        answer?.kind !== "choice" ||
+        !Number.isFinite(answer.confidence) ||
+        answer.confidence > 1 ||
+        !answer.probabilities ||
+        Array.isArray(answer.probabilities) ||
+        !Object.hasOwn(answer.probabilities, answer.value) ||
+        !Object.values(answer.probabilities).every((value) => Number.isFinite(value) && value >= 0 && value <= 1)
+    ) {
         return undefined;
     }
 
@@ -214,7 +211,7 @@ export function choiceValue(answer, system) {
     }
 
     const { first, second } = topTwo(answer.probabilities);
-    if (answer.probabilities && first - second < limits.margin) {
+    if (answer.probabilities[answer.value] !== first || first - second < limits.margin) {
         return undefined;
     }
 
@@ -226,7 +223,14 @@ export function choiceValue(answer, system) {
  * returned level is the rounded band; callers compare against their own rubric.
  */
 export function scoreLevel(answer, system) {
-    if (answer?.kind !== "score" || typeof answer.confidence !== "number") {
+    if (
+        answer?.kind !== "score" ||
+        !Number.isFinite(answer.confidence) ||
+        answer.confidence > 1 ||
+        !Number.isFinite(answer.value) ||
+        answer.value < 0 ||
+        answer.value > 9
+    ) {
         return undefined;
     }