pi-ghost-autocomplete 0.3.1 → 0.3.2

package/dist/edit-history-ring.js

@@ -0,0 +1,262 @@
+/**
+ * Edit-history ring + aggregated index. Enhancement #1
+ * (Next-Edit-Prediction motivated). See `docs/edit-history-signals-spec.md`.
+ *
+ * Records the most recent K edits (default 32) made in-session, then
+ * exposes:
+ *  - a `snapshot()` of newest-first events with age filtering, and
+ *  - an `index()` of decayed token frequencies + per-URI line proximity.
+ *
+ * The reranker reads `index()` to compute `editProximityBoost`; the M6
+ * context selector reads it (opt-in) for a token-overlap locality bit.
+ *
+ * Privacy: events are pre-sanitized at capture (`containsSensitive` on
+ * the raw added/removed text — caller is responsible for invoking it
+ * before calling `record`). Only tokens + scalars + URIs survive. No
+ * raw text is ever stored on the ring. The ring lives in-process only —
+ * no persistence in v0.5.
+ */
+const TOKEN_RE = /\b[\w][\w-]*\b/g;
+export const TOKEN_CAP_PER_EVENT = 16;
+export const DEFAULT_CAPACITY = 32;
+export const DEFAULT_MAX_AGE_MS = 10 * 60 * 1000;
+export const DEFAULT_HALF_LIFE_MS = 2 * 60 * 1000;
+/**
+ * Tokenize text using the same regex as `profile.ts` (`\b[\w][\w-]*\b`).
+ * Code identifiers are case-sensitive — we do NOT lowercase. Result is
+ * capped at `TOKEN_CAP_PER_EVENT` to bound memory per event.
+ */
+export function tokenize(text) {
+    if (!text)
+        return [];
+    const out = [];
+    for (const m of text.matchAll(TOKEN_RE)) {
+        if (m[0])
+            out.push(m[0]);
+        if (out.length >= TOKEN_CAP_PER_EVENT)
+            break;
+    }
+    return out;
+}
+/**
+ * Exponential time-decay weight: 1.0 at `ageMs=0`, halves every
+ * `halfLifeMs`. Pure helper for tests + index().
+ */
+export function timeDecayWeight(ageMs, halfLifeMs = DEFAULT_HALF_LIFE_MS) {
+    if (ageMs <= 0)
+        return 1;
+    if (halfLifeMs <= 0)
+        return 0;
+    return Math.exp((-ageMs * Math.LN2) / halfLifeMs);
+}
+export class EditHistoryRing {
+    events = [];
+    capacity;
+    maxAgeMs;
+    halfLifeMs;
+    constructor(opts = {}) {
+        this.capacity = Math.max(1, Math.floor(opts.capacity ?? DEFAULT_CAPACITY));
+        this.maxAgeMs = Math.max(0, opts.maxAgeMs ?? DEFAULT_MAX_AGE_MS);
+        this.halfLifeMs = Math.max(1, opts.halfLifeMs ?? DEFAULT_HALF_LIFE_MS);
+    }
+    /**
+     * Append an event. `ts` defaults to `Date.now()`. Caller is responsible
+     * for sanitization — the ring does not re-validate. Overflow drops the
+     * oldest event.
+     */
+    record(ev) {
+        const stored = {
+            ts: ev.ts ?? Date.now(),
+            uri: ev.uri,
+            rangeStartLine: ev.rangeStartLine,
+            addedTokens: ev.addedTokens.slice(0, TOKEN_CAP_PER_EVENT),
+            removedTokens: ev.removedTokens.slice(0, TOKEN_CAP_PER_EVENT),
+            netCharDelta: ev.netCharDelta,
+        };
+        this.events.push(stored);
+        while (this.events.length > this.capacity)
+            this.events.shift();
+    }
+    /**
+     * Newest-first list of events not older than `maxAgeMs`. Returns a
+     * copy — safe for the caller to retain.
+     *
+     * Scans without early-break: we cannot assume `record()` was called in
+     * ts-monotonic order (a backfill or test clock could insert older
+     * events after newer ones). At capacity=32 the scan cost is trivial.
+     */
+    snapshot(now = Date.now()) {
+        const cutoff = now - this.maxAgeMs;
+        const out = [];
+        for (let i = this.events.length - 1; i >= 0; i--) {
+            const ev = this.events[i];
+            if (ev.ts >= cutoff)
+                out.push(ev);
+        }
+        return out;
+    }
+    /**
+     * Aggregated index of recent edit signals, with exponential time-decay
+     * applied to token frequencies. URIs touched within `maxAgeMs` are
+     * collected; the *most recent* edit line per URI is kept (used by the
+     * reranker for line-proximity scoring).
+     */
+    index(now = Date.now()) {
+        const addedFreq = new Map();
+        const removedFreq = new Map();
+        const recentUris = new Set();
+        const lastEditLineByUri = new Map();
+        let totalEvents = 0;
+        const events = this.snapshot(now);
+        for (const ev of events) {
+            const w = timeDecayWeight(now - ev.ts, this.halfLifeMs);
+            for (const t of ev.addedTokens)
+                addedFreq.set(t, (addedFreq.get(t) ?? 0) + w);
+            for (const t of ev.removedTokens)
+                removedFreq.set(t, (removedFreq.get(t) ?? 0) + w);
+            if (ev.uri) {
+                recentUris.add(ev.uri);
+                if (!lastEditLineByUri.has(ev.uri)) {
+                    lastEditLineByUri.set(ev.uri, ev.rangeStartLine);
+                }
+            }
+            totalEvents++;
+        }
+        return { addedFreq, removedFreq, recentUris, lastEditLineByUri, totalEvents };
+    }
+    clear() {
+        this.events.length = 0;
+    }
+    get size() {
+        return this.events.length;
+    }
+}
+/** Default per-entry size cap. 1 MB — large enough for any realistic
+ *  source file, small enough that 16 × 1MB = 16MB worst-case is bounded. */
+export const DEFAULT_PRE_EDIT_MAX_BYTES = 1024 * 1024;
+/**
+ * Per-document pre-edit text cache. Holds the most recent `capacity`
+ * document texts so we can derive the *removed* slice of an edit before
+ * the on-disk text mutates. LRU-bounded to avoid pinning every project
+ * file in RAM. v0.5 stores transiently in process only — never persisted.
+ *
+ * Per-entry size is capped at `maxBytesPerEntry` (default 1 MB). Texts
+ * exceeding the cap are stored as empty string — the next edit on that
+ * document will derive empty `removedTokens`, which is the same
+ * degradation as a fresh LRU miss. This bounds worst-case memory at
+ * `capacity * maxBytesPerEntry` regardless of project file sizes.
+ */
+export class PreEditTextCache {
+    map = new Map();
+    capacity;
+    maxBytesPerEntry;
+    constructor(capacity = 16, maxBytesPerEntry = DEFAULT_PRE_EDIT_MAX_BYTES) {
+        this.capacity = Math.max(1, Math.floor(capacity));
+        this.maxBytesPerEntry = Math.max(0, Math.floor(maxBytesPerEntry));
+    }
+    /** Returns the snapshot for `uri` (or `""` if absent) without touching LRU order. */
+    peek(uri) {
+        return this.map.get(uri) ?? "";
+    }
+    /**
+     * Update the snapshot. Move-to-front semantics; oldest is evicted when
+     * capacity is exceeded. Texts exceeding `maxBytesPerEntry` are stored
+     * as empty string (the next edit on this URI gets no `removedTokens`,
+     * matching the LRU-miss degradation path).
+     *
+     * Byte length uses UTF-16 code units (`text.length`), not UTF-8 bytes —
+     * the cap is approximate but stable across machines.
+     */
+    update(uri, text) {
+        const safe = text.length <= this.maxBytesPerEntry ? text : "";
+        if (this.map.has(uri))
+            this.map.delete(uri);
+        this.map.set(uri, safe);
+        while (this.map.size > this.capacity) {
+            const oldest = this.map.keys().next().value;
+            if (oldest === undefined)
+                break;
+            this.map.delete(oldest);
+        }
+    }
+    clear() {
+        this.map.clear();
+    }
+    get size() {
+        return this.map.size;
+    }
+}
+/**
+ * Parse positive-integer env. Returns fallback when unset or invalid.
+ * Bounded by `[1, max]`.
+ */
+function parsePosIntEnv(raw, fallback, max) {
+    if (raw === undefined)
+        return fallback;
+    const v = Number(raw);
+    if (!Number.isFinite(v) || v < 1 || v > max)
+        return fallback;
+    return Math.floor(v);
+}
+/**
+ * Spec §5.1 — compute the raw `editProximityScore` for a candidate. Pure
+ * function. Returns 0 when the ring is empty (zero events). Output is
+ * un-normalized: the reranker's min-max normalization re-maps the value
+ * across the candidate set so absolute scale doesn't matter — only the
+ * relative ordering does. The same un-normalized value is reported on
+ * the `MetricRecord` for observability.
+ *
+ * `currentUri` / `currentLine` describe the user's cursor at predict
+ * time. The prompt-buffer editor uses a static `"prompt"` URI sentinel
+ * and the current cursor line.
+ */
+export function editProximityScore(text, index, currentUri, currentLine) {
+    if (index.totalEvents === 0)
+        return 0;
+    const tokens = tokenize(text);
+    let added = 0;
+    let removed = 0;
+    for (const t of tokens) {
+        added += index.addedFreq.get(t) ?? 0;
+        removed += index.removedFreq.get(t) ?? 0;
+    }
+    const tokenSignal = Math.log(1 + added) - 0.5 * Math.log(1 + removed);
+    const uriBoost = index.recentUris.has(currentUri) ? 1 : 0;
+    const lastLine = index.lastEditLineByUri.get(currentUri);
+    const lineProximity = lastLine === undefined ? 0 : Math.exp(-Math.abs(lastLine - currentLine) / 8);
+    return tokenSignal + 0.5 * uriBoost + 0.3 * lineProximity;
+}
+/**
+ * Pure character-diff helper. Returns the minimal `{ added, removed }`
+ * slice between two strings by stripping the common prefix + suffix.
+ * Caller is responsible for `containsSensitive` checks before tokenizing
+ * the result. Returns empty strings for both sides when texts are equal.
+ */
+export function diffStrings(prev, next) {
+    if (prev === next)
+        return { added: "", removed: "" };
+    let p = 0;
+    const maxP = Math.min(prev.length, next.length);
+    while (p < maxP && prev.charCodeAt(p) === next.charCodeAt(p))
+        p++;
+    let s = 0;
+    while (s < maxP - p &&
+        prev.charCodeAt(prev.length - 1 - s) === next.charCodeAt(next.length - 1 - s)) {
+        s++;
+    }
+    return {
+        added: next.slice(p, next.length - s),
+        removed: prev.slice(p, prev.length - s),
+    };
+}
+export function readEditHistoryEnv(env = process.env) {
+    const enabledRaw = env.PI_GHOST_EDIT_HISTORY;
+    const enabled = enabledRaw === undefined || enabledRaw === "1";
+    return {
+        enabled,
+        capacity: parsePosIntEnv(env.PI_GHOST_EDIT_HISTORY_CAP, DEFAULT_CAPACITY, 1024),
+        maxAgeMs: parsePosIntEnv(env.PI_GHOST_EDIT_HISTORY_MAX_AGE_MS, DEFAULT_MAX_AGE_MS, 24 * 60 * 60 * 1000),
+        halfLifeMs: parsePosIntEnv(env.PI_GHOST_EDIT_HISTORY_HALF_LIFE_MS, DEFAULT_HALF_LIFE_MS, 24 * 60 * 60 * 1000),
+    };
+}
+//# sourceMappingURL=edit-history-ring.js.map

package/dist/edit-history-ring.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"edit-history-ring.js","sourceRoot":"","sources":["../src/edit-history-ring.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;GAiBG;AAEH,MAAM,QAAQ,GAAG,iBAAiB,CAAC;AACnC,MAAM,CAAC,MAAM,mBAAmB,GAAG,EAAE,CAAC;AACtC,MAAM,CAAC,MAAM,gBAAgB,GAAG,EAAE,CAAC;AACnC,MAAM,CAAC,MAAM,kBAAkB,GAAG,EAAE,GAAG,EAAE,GAAG,IAAI,CAAC;AACjD,MAAM,CAAC,MAAM,oBAAoB,GAAG,CAAC,GAAG,EAAE,GAAG,IAAI,CAAC;AAyBlD;;;;GAIG;AACH,MAAM,UAAU,QAAQ,CAAC,IAAY;IACpC,IAAI,CAAC,IAAI;QAAE,OAAO,EAAE,CAAC;IACrB,MAAM,GAAG,GAAa,EAAE,CAAC;IACzB,KAAK,MAAM,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;QACzC,IAAI,CAAC,CAAC,CAAC,CAAC;YAAE,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QACzB,IAAI,GAAG,CAAC,MAAM,IAAI,mBAAmB;YAAE,MAAM;IAC9C,CAAC;IACD,OAAO,GAAG,CAAC;AACZ,CAAC;AAED;;;GAGG;AACH,MAAM,UAAU,eAAe,CAAC,KAAa,EAAE,aAAqB,oBAAoB;IACvF,IAAI,KAAK,IAAI,CAAC;QAAE,OAAO,CAAC,CAAC;IACzB,IAAI,UAAU,IAAI,CAAC;QAAE,OAAO,CAAC,CAAC;IAC9B,OAAO,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,KAAK,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,UAAU,CAAC,CAAC;AACnD,CAAC;AAED,MAAM,OAAO,eAAe;IACV,MAAM,GAAgB,EAAE,CAAC;IACjC,QAAQ,CAAS;IACjB,QAAQ,CAAS;IACjB,UAAU,CAAS;IAE5B,YAAY,OAAwB,EAAE;QACrC,IAAI,CAAC,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,QAAQ,IAAI,gBAAgB,CAAC,CAAC,CAAC;QAC3E,IAAI,CAAC,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,QAAQ,IAAI,kBAAkB,CAAC,CAAC;QACjE,IAAI,CAAC,UAAU,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,UAAU,IAAI,oBAAoB,CAAC,CAAC;IACxE,CAAC;IAED;;;;OAIG;IACH,MAAM,CAAC,EAA2C;QACjD,MAAM,MAAM,GAAc;YACzB,EAAE,EAAE,EAAE,CAAC,EAAE,IAAI,IAAI,CAAC,GAAG,EAAE;YACvB,GAAG,EAAE,EAAE,CAAC,GAAG;YACX,cAAc,EAAE,EAAE,CAAC,cAAc;YACjC,WAAW,EAAE,EAAE,CAAC,WAAW,CAAC,KAAK,CAAC,CAAC,EAAE,mBAAmB,CAAC;YACzD,aAAa,EAAE,EAAE,CAAC,aAAa,CAAC,KAAK,CAAC,CAAC,EAAE,mBAAmB,CAAC;YAC7D,YAAY,EAAE,EAAE,CAAC,YAAY;SAC7B,CAAC;QACF,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;QACzB,OAAO,IAAI,CAAC,MAAM,CAAC,MAAM,GAAG,IAAI,CAAC,QAAQ;YAAE,IAAI,CAAC,MAAM,CAAC,KAAK,EAAE,CAAC;IAChE,CAAC;IAED;;;;;;;OAOG;IACH,QAAQ,CAAC,MAAc,IAAI,CAAC,GAAG,EAAE;QAChC,MAAM,MAAM,GAAG,GAAG,GAAG,IAAI,CAAC,QAAQ,CAAC;QACnC,MAAM,GAAG,GAAgB,EAAE,CAAC;QAC5B,KAAK,IAAI,CAAC,GAAG,IAAI,CAAC,MAAM,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;YAClD,MAAM,EAAE,GAAG,IAAI,CAAC,MAAM,CAAC,CAAC,CAAE,CAAC;YAC3B,IAAI,EAAE,CAAC,EAAE,IAAI,MAAM;gBAAE,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;QACnC,CAAC;QACD,OAAO,GAAG,CAAC;IACZ,CAAC;IAED;;;;;OAKG;IACH,KAAK,CAAC,MAAc,IAAI,CAAC,GAAG,EAAE;QAC7B,MAAM,SAAS,GAAG,IAAI,GAAG,EAAkB,CAAC;QAC5C,MAAM,WAAW,GAAG,IAAI,GAAG,EAAkB,CAAC;QAC9C,MAAM,UAAU,GAAG,IAAI,GAAG,EAAU,CAAC;QACrC,MAAM,iBAAiB,GAAG,IAAI,GAAG,EAAkB,CAAC;QACpD,IAAI,WAAW,GAAG,CAAC,CAAC;QACpB,MAAM,MAAM,GAAG,IAAI,CAAC,QAAQ,CAAC,GAAG,CAAC,CAAC;QAClC,KAAK,MAAM,EAAE,IAAI,MAAM,EAAE,CAAC;YACzB,MAAM,CAAC,GAAG,eAAe,CAAC,GAAG,GAAG,EAAE,CAAC,EAAE,EAAE,IAAI,CAAC,UAAU,CAAC,CAAC;YACxD,KAAK,MAAM,CAAC,IAAI,EAAE,CAAC,WAAW;gBAAE,SAAS,CAAC,GAAG,CAAC,CAAC,EAAE,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;YAC9E,KAAK,MAAM,CAAC,IAAI,EAAE,CAAC,aAAa;gBAAE,WAAW,CAAC,GAAG,CAAC,CAAC,EAAE,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;YACpF,IAAI,EAAE,CAAC,GAAG,EAAE,CAAC;gBACZ,UAAU,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC;gBACvB,IAAI,CAAC,iBAAiB,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,EAAE,CAAC;oBACpC,iBAAiB,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,EAAE,EAAE,CAAC,cAAc,CAAC,CAAC;gBAClD,CAAC;YACF,CAAC;YACD,WAAW,EAAE,CAAC;QACf,CAAC;QACD,OAAO,EAAE,SAAS,EAAE,WAAW,EAAE,UAAU,EAAE,iBAAiB,EAAE,WAAW,EAAE,CAAC;IAC/E,CAAC;IAED,KAAK;QACJ,IAAI,CAAC,MAAM,CAAC,MAAM,GAAG,CAAC,CAAC;IACxB,CAAC;IAED,IAAI,IAAI;QACP,OAAO,IAAI,CAAC,MAAM,CAAC,MAAM,CAAC;IAC3B,CAAC;CACD;AAED;4EAC4E;AAC5E,MAAM,CAAC,MAAM,0BAA0B,GAAG,IAAI,GAAG,IAAI,CAAC;AAEtD;;;;;;;;;;;GAWG;AACH,MAAM,OAAO,gBAAgB;IACX,GAAG,GAAG,IAAI,GAAG,EAAkB,CAAC;IACxC,QAAQ,CAAS;IACjB,gBAAgB,CAAS;IAElC,YAAY,QAAQ,GAAG,EAAE,EAAE,mBAA2B,0BAA0B;QAC/E,IAAI,CAAC,QAAQ,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC,CAAC;QAClD,IAAI,CAAC,gBAAgB,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,IAAI,CAAC,KAAK,CAAC,gBAAgB,CAAC,CAAC,CAAC;IACnE,CAAC;IAED,qFAAqF;IACrF,IAAI,CAAC,GAAW;QACf,OAAO,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC;IAChC,CAAC;IAED;;;;;;;;OAQG;IACH,MAAM,CAAC,GAAW,EAAE,IAAY;QAC/B,MAAM,IAAI,GAAG,IAAI,CAAC,MAAM,IAAI,IAAI,CAAC,gBAAgB,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC;QAC9D,IAAI,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,GAAG,CAAC;YAAE,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;QAC5C,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,GAAG,EAAE,IAAI,CAAC,CAAC;QACxB,OAAO,IAAI,CAAC,GAAG,CAAC,IAAI,GAAG,IAAI,CAAC,QAAQ,EAAE,CAAC;YACtC,MAAM,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,EAAE,CAAC,IAAI,EAAE,CAAC,KAAK,CAAC;YAC5C,IAAI,MAAM,KAAK,SAAS;gBAAE,MAAM;YAChC,IAAI,CAAC,GAAG,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC;QACzB,CAAC;IACF,CAAC;IAED,KAAK;QACJ,IAAI,CAAC,GAAG,CAAC,KAAK,EAAE,CAAC;IAClB,CAAC;IAED,IAAI,IAAI;QACP,OAAO,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC;IACtB,CAAC;CACD;AAED;;;GAGG;AACH,SAAS,cAAc,CAAC,GAAuB,EAAE,QAAgB,EAAE,GAAW;IAC7E,IAAI,GAAG,KAAK,SAAS;QAAE,OAAO,QAAQ,CAAC;IACvC,MAAM,CAAC,GAAG,MAAM,CAAC,GAAG,CAAC,CAAC;IACtB,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,GAAG,GAAG;QAAE,OAAO,QAAQ,CAAC;IAC7D,OAAO,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;AACtB,CAAC;AAED;;;;;;;;;;;GAWG;AACH,MAAM,UAAU,kBAAkB,CACjC,IAAY,EACZ,KAAuB,EACvB,UAAkB,EAClB,WAAmB;IAEnB,IAAI,KAAK,CAAC,WAAW,KAAK,CAAC;QAAE,OAAO,CAAC,CAAC;IACtC,MAAM,MAAM,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC;IAC9B,IAAI,KAAK,GAAG,CAAC,CAAC;IACd,IAAI,OAAO,GAAG,CAAC,CAAC;IAChB,KAAK,MAAM,CAAC,IAAI,MAAM,EAAE,CAAC;QACxB,KAAK,IAAI,KAAK,CAAC,SAAS,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;QACrC,OAAO,IAAI,KAAK,CAAC,WAAW,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC;IAC1C,CAAC;IACD,MAAM,WAAW,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,KAAK,CAAC,GAAG,GAAG,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,GAAG,OAAO,CAAC,CAAC;IACtE,MAAM,QAAQ,GAAG,KAAK,CAAC,UAAU,CAAC,GAAG,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAC1D,MAAM,QAAQ,GAAG,KAAK,CAAC,iBAAiB,CAAC,GAAG,CAAC,UAAU,CAAC,CAAC;IACzD,MAAM,aAAa,GAClB,QAAQ,KAAK,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,QAAQ,GAAG,WAAW,CAAC,GAAG,CAAC,CAAC,CAAC;IAC9E,OAAO,WAAW,GAAG,GAAG,GAAG,QAAQ,GAAG,GAAG,GAAG,aAAa,CAAC;AAC3D,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,WAAW,CAAC,IAAY,EAAE,IAAY;IACrD,IAAI,IAAI,KAAK,IAAI;QAAE,OAAO,EAAE,KAAK,EAAE,EAAE,EAAE,OAAO,EAAE,EAAE,EAAE,CAAC;IACrD,IAAI,CAAC,GAAG,CAAC,CAAC;IACV,MAAM,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,MAAM,EAAE,IAAI,CAAC,MAAM,CAAC,CAAC;IAChD,OAAO,CAAC,GAAG,IAAI,IAAI,IAAI,CAAC,UAAU,CAAC,CAAC,CAAC,KAAK,IAAI,CAAC,UAAU,CAAC,CAAC,CAAC;QAAE,CAAC,EAAE,CAAC;IAClE,IAAI,CAAC,GAAG,CAAC,CAAC;IACV,OACC,CAAC,GAAG,IAAI,GAAG,CAAC;QACZ,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,GAAG,CAAC,CAAC,KAAK,IAAI,CAAC,UAAU,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,GAAG,CAAC,CAAC,EAC5E,CAAC;QACF,CAAC,EAAE,CAAC;IACL,CAAC;IACD,OAAO;QACN,KAAK,EAAE,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC;QACrC,OAAO,EAAE,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC;KACvC,CAAC;AACH,CAAC;AAED,MAAM,UAAU,kBAAkB,CACjC,MAA0C,OAAO,CAAC,GAAG;IAErD,MAAM,UAAU,GAAG,GAAG,CAAC,qBAAqB,CAAC;IAC7C,MAAM,OAAO,GAAG,UAAU,KAAK,SAAS,IAAI,UAAU,KAAK,GAAG,CAAC;IAC/D,OAAO;QACN,OAAO;QACP,QAAQ,EAAE,cAAc,CAAC,GAAG,CAAC,yBAAyB,EAAE,gBAAgB,EAAE,IAAI,CAAC;QAC/E,QAAQ,EAAE,cAAc,CACvB,GAAG,CAAC,gCAAgC,EACpC,kBAAkB,EAClB,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,IAAI,CACnB;QACD,UAAU,EAAE,cAAc,CACzB,GAAG,CAAC,kCAAkC,EACtC,oBAAoB,EACpB,EAAE,GAAG,EAAE,GAAG,EAAE,GAAG,IAAI,CACnB;KACD,CAAC;AACH,CAAC"}

package/dist/entropy-stop.d.ts

@@ -0,0 +1,102 @@
+/**
+ * Entropy-based early stop. PRD §M7.
+ *
+ * Streams `(token, -logprob)` pairs; cuts when a token's `-logprob` spikes
+ * above a learned baseline AND the token is whitespace/punctuation
+ * (natural cut points). Returns the text up to (not including) the
+ * cutting token.
+ *
+ * Pure state machine — no I/O, no provider coupling. Caller drives the
+ * stream and acts on the returned `StopDecision`.
+ *
+ * Capability gate (PRD §M7):
+ *   Mercury Edit FIM: no logprobs → caller skips this module
+ *   OpenAI chat:      supported
+ *   Groq/Ollama:      model-dependent
+ *
+ * Disabled when fewer than `warmup` tokens have been seen — no statistical
+ * baseline yet. Hard cap at `maxTokens` regardless.
+ */
+export declare const DEFAULT_WARMUP_TOKENS = 5;
+export declare const DEFAULT_SIGMA_MULT = 1.5;
+export declare const DEFAULT_MAX_TOKENS = 64;
+export interface EntropyStopperOpts {
+    /** Number of leading tokens used to seed the mean+std. */
+    warmupTokens?: number;
+    /** Sigma multiplier for the spike threshold. */
+    sigmaMult?: number;
+    /** Hard cap on total tokens observed before forced stop. */
+    maxTokens?: number;
+}
+export interface StopDecision {
+    /** True iff caller should stop streaming now. */
+    stop: boolean;
+    /**
+     * When `stop`, this is the reason. `entropy-spike` means a natural
+     * cut token tripped the threshold; `max-tokens` means the hard cap
+     * fired regardless of entropy.
+     */
+    reason?: "entropy-spike" | "max-tokens";
+    /**
+     * Running mean of `-logprob` after warmup. Useful for telemetry;
+     * undefined while still in warmup.
+     */
+    mean?: number;
+    /** Running standard deviation; undefined during warmup. */
+    std?: number;
+}
+export declare class EntropyStopper {
+    private readonly warmup;
+    private readonly sigma;
+    private readonly cap;
+    private readonly stats;
+    private tokensSeen;
+    constructor(opts?: EntropyStopperOpts);
+    /**
+     * Feed one streaming token + its `-logprob` (higher = more
+     * surprising / less confident). Returns whether the caller should
+     * stop the stream BEFORE this token. The caller's accumulated text
+     * up to (not including) this token is the "accept" boundary.
+     *
+     * Order of evaluation:
+     *   1. Hard cap: token index >= cap → stop
+     *   2. Warmup: token index < warmup → feed stats, no stop
+     *   3. Spike: -logprob > mean + sigma*std AND token is natural cut
+     *      → stop
+     *   4. Else: feed stats, no stop
+     */
+    observe(token: string, neglogprob: number): StopDecision;
+    /**
+     * Diagnostic snapshot — count + current mean/std. Returns `null`
+     * mean/std until the warmup window has filled enough samples for a
+     * meaningful estimate.
+     */
+    snapshot(): {
+        tokensSeen: number;
+        mean: number | null;
+        std: number | null;
+    };
+}
+/**
+ * True iff `token` is a natural cut point — whitespace, end-of-clause
+ * punctuation, or a closing bracket. PRD §M7 limits entropy-driven
+ * cutoffs to these to avoid mid-identifier truncation.
+ */
+export declare function isNaturalCutToken(token: string): boolean;
+/**
+ * Run a stream through an entropy stopper. Returns the prefix of joined
+ * tokens up to (but not including) the stop point. When the stream ends
+ * naturally without a stop, returns the full concatenation.
+ *
+ * Exposed as a convenience for tests + future provider wiring; editor
+ * integration is deferred until pi-ai exposes per-token logprob events.
+ */
+export declare function runWithStop(tokens: Iterable<{
+    token: string;
+    neglogprob: number;
+}>, opts?: EntropyStopperOpts): {
+    text: string;
+    stopped: boolean;
+    reason?: "entropy-spike" | "max-tokens";
+};
+//# sourceMappingURL=entropy-stop.d.ts.map

package/dist/entropy-stop.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"entropy-stop.d.ts","sourceRoot":"","sources":["../src/entropy-stop.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;GAkBG;AAEH,eAAO,MAAM,qBAAqB,IAAI,CAAC;AACvC,eAAO,MAAM,kBAAkB,MAAM,CAAC;AACtC,eAAO,MAAM,kBAAkB,KAAK,CAAC;AAKrC,MAAM,WAAW,kBAAkB;IAClC,0DAA0D;IAC1D,YAAY,CAAC,EAAE,MAAM,CAAC;IACtB,gDAAgD;IAChD,SAAS,CAAC,EAAE,MAAM,CAAC;IACnB,4DAA4D;IAC5D,SAAS,CAAC,EAAE,MAAM,CAAC;CACnB;AAED,MAAM,WAAW,YAAY;IAC5B,iDAAiD;IACjD,IAAI,EAAE,OAAO,CAAC;IACd;;;;OAIG;IACH,MAAM,CAAC,EAAE,eAAe,GAAG,YAAY,CAAC;IACxC;;;OAGG;IACH,IAAI,CAAC,EAAE,MAAM,CAAC;IACd,2DAA2D;IAC3D,GAAG,CAAC,EAAE,MAAM,CAAC;CACb;AAiCD,qBAAa,cAAc;IAC1B,OAAO,CAAC,QAAQ,CAAC,MAAM,CAAS;IAChC,OAAO,CAAC,QAAQ,CAAC,KAAK,CAAS;IAC/B,OAAO,CAAC,QAAQ,CAAC,GAAG,CAAS;IAC7B,OAAO,CAAC,QAAQ,CAAC,KAAK,CAAsB;IAC5C,OAAO,CAAC,UAAU,CAAK;gBAEX,IAAI,GAAE,kBAAuB;IAMzC;;;;;;;;;;;;OAYG;IACH,OAAO,CAAC,KAAK,EAAE,MAAM,EAAE,UAAU,EAAE,MAAM,GAAG,YAAY;IAoBxD;;;;OAIG;IACH,QAAQ,IAAI;QAAE,UAAU,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,GAAG,IAAI,CAAC;QAAC,GAAG,EAAE,MAAM,GAAG,IAAI,CAAA;KAAE;CAO3E;AAED;;;;GAIG;AACH,wBAAgB,iBAAiB,CAAC,KAAK,EAAE,MAAM,GAAG,OAAO,CAOxD;AAED;;;;;;;GAOG;AACH,wBAAgB,WAAW,CAC1B,MAAM,EAAE,QAAQ,CAAC;IAAE,KAAK,EAAE,MAAM,CAAC;IAAC,UAAU,EAAE,MAAM,CAAA;CAAE,CAAC,EACvD,IAAI,GAAE,kBAAuB,GAC3B;IAAE,IAAI,EAAE,MAAM,CAAC;IAAC,OAAO,EAAE,OAAO,CAAC;IAAC,MAAM,CAAC,EAAE,eAAe,GAAG,YAAY,CAAA;CAAE,CAS7E"}

package/dist/entropy-stop.js

@@ -0,0 +1,141 @@
+/**
+ * Entropy-based early stop. PRD §M7.
+ *
+ * Streams `(token, -logprob)` pairs; cuts when a token's `-logprob` spikes
+ * above a learned baseline AND the token is whitespace/punctuation
+ * (natural cut points). Returns the text up to (not including) the
+ * cutting token.
+ *
+ * Pure state machine — no I/O, no provider coupling. Caller drives the
+ * stream and acts on the returned `StopDecision`.
+ *
+ * Capability gate (PRD §M7):
+ *   Mercury Edit FIM: no logprobs → caller skips this module
+ *   OpenAI chat:      supported
+ *   Groq/Ollama:      model-dependent
+ *
+ * Disabled when fewer than `warmup` tokens have been seen — no statistical
+ * baseline yet. Hard cap at `maxTokens` regardless.
+ */
+export const DEFAULT_WARMUP_TOKENS = 5;
+export const DEFAULT_SIGMA_MULT = 1.5;
+export const DEFAULT_MAX_TOKENS = 64;
+/** Characters at which an entropy spike is allowed to terminate. */
+const NATURAL_CUT_PATTERN = /[\s.,;:!?)\]}>'"`]/;
+/**
+ * Simple incremental mean+std (Welford's algorithm). One-pass over the
+ * warmup window, stable for the small N=5 case the PRD specifies.
+ */
+class RunningStats {
+    n = 0;
+    mean = 0;
+    m2 = 0;
+    push(x) {
+        this.n++;
+        const delta = x - this.mean;
+        this.mean += delta / this.n;
+        const delta2 = x - this.mean;
+        this.m2 += delta * delta2;
+    }
+    get count() {
+        return this.n;
+    }
+    get currentMean() {
+        return this.mean;
+    }
+    get currentStd() {
+        if (this.n < 2)
+            return 0;
+        return Math.sqrt(this.m2 / (this.n - 1));
+    }
+}
+export class EntropyStopper {
+    warmup;
+    sigma;
+    cap;
+    stats = new RunningStats();
+    tokensSeen = 0;
+    constructor(opts = {}) {
+        this.warmup = opts.warmupTokens ?? DEFAULT_WARMUP_TOKENS;
+        this.sigma = opts.sigmaMult ?? DEFAULT_SIGMA_MULT;
+        this.cap = opts.maxTokens ?? DEFAULT_MAX_TOKENS;
+    }
+    /**
+     * Feed one streaming token + its `-logprob` (higher = more
+     * surprising / less confident). Returns whether the caller should
+     * stop the stream BEFORE this token. The caller's accumulated text
+     * up to (not including) this token is the "accept" boundary.
+     *
+     * Order of evaluation:
+     *   1. Hard cap: token index >= cap → stop
+     *   2. Warmup: token index < warmup → feed stats, no stop
+     *   3. Spike: -logprob > mean + sigma*std AND token is natural cut
+     *      → stop
+     *   4. Else: feed stats, no stop
+     */
+    observe(token, neglogprob) {
+        const idx = this.tokensSeen;
+        this.tokensSeen++;
+        if (idx >= this.cap) {
+            return { stop: true, reason: "max-tokens" };
+        }
+        if (idx < this.warmup) {
+            this.stats.push(neglogprob);
+            return { stop: false };
+        }
+        const mean = this.stats.currentMean;
+        const std = this.stats.currentStd;
+        const threshold = mean + this.sigma * std;
+        if (neglogprob > threshold && isNaturalCutToken(token)) {
+            return { stop: true, reason: "entropy-spike", mean, std };
+        }
+        this.stats.push(neglogprob);
+        return { stop: false, mean, std };
+    }
+    /**
+     * Diagnostic snapshot — count + current mean/std. Returns `null`
+     * mean/std until the warmup window has filled enough samples for a
+     * meaningful estimate.
+     */
+    snapshot() {
+        return {
+            tokensSeen: this.tokensSeen,
+            mean: this.stats.count >= 1 ? this.stats.currentMean : null,
+            std: this.stats.count >= 2 ? this.stats.currentStd : null,
+        };
+    }
+}
+/**
+ * True iff `token` is a natural cut point — whitespace, end-of-clause
+ * punctuation, or a closing bracket. PRD §M7 limits entropy-driven
+ * cutoffs to these to avoid mid-identifier truncation.
+ */
+export function isNaturalCutToken(token) {
+    if (!token)
+        return false;
+    // Token may be multiple characters; check the FIRST char so the
+    // algorithm cuts at the natural boundary inside the token rather
+    // than waiting for a pure-punctuation token.
+    const ch = token[0];
+    return NATURAL_CUT_PATTERN.test(ch);
+}
+/**
+ * Run a stream through an entropy stopper. Returns the prefix of joined
+ * tokens up to (but not including) the stop point. When the stream ends
+ * naturally without a stop, returns the full concatenation.
+ *
+ * Exposed as a convenience for tests + future provider wiring; editor
+ * integration is deferred until pi-ai exposes per-token logprob events.
+ */
+export function runWithStop(tokens, opts = {}) {
+    const stopper = new EntropyStopper(opts);
+    let text = "";
+    for (const { token, neglogprob } of tokens) {
+        const d = stopper.observe(token, neglogprob);
+        if (d.stop)
+            return { text, stopped: true, reason: d.reason };
+        text += token;
+    }
+    return { text, stopped: false };
+}
+//# sourceMappingURL=entropy-stop.js.map

package/dist/entropy-stop.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"entropy-stop.js","sourceRoot":"","sources":["../src/entropy-stop.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;GAkBG;AAEH,MAAM,CAAC,MAAM,qBAAqB,GAAG,CAAC,CAAC;AACvC,MAAM,CAAC,MAAM,kBAAkB,GAAG,GAAG,CAAC;AACtC,MAAM,CAAC,MAAM,kBAAkB,GAAG,EAAE,CAAC;AAErC,oEAAoE;AACpE,MAAM,mBAAmB,GAAG,oBAAoB,CAAC;AA6BjD;;;GAGG;AACH,MAAM,YAAY;IACT,CAAC,GAAG,CAAC,CAAC;IACN,IAAI,GAAG,CAAC,CAAC;IACT,EAAE,GAAG,CAAC,CAAC;IAEf,IAAI,CAAC,CAAS;QACb,IAAI,CAAC,CAAC,EAAE,CAAC;QACT,MAAM,KAAK,GAAG,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC;QAC5B,IAAI,CAAC,IAAI,IAAI,KAAK,GAAG,IAAI,CAAC,CAAC,CAAC;QAC5B,MAAM,MAAM,GAAG,CAAC,GAAG,IAAI,CAAC,IAAI,CAAC;QAC7B,IAAI,CAAC,EAAE,IAAI,KAAK,GAAG,MAAM,CAAC;IAC3B,CAAC;IAED,IAAI,KAAK;QACR,OAAO,IAAI,CAAC,CAAC,CAAC;IACf,CAAC;IAED,IAAI,WAAW;QACd,OAAO,IAAI,CAAC,IAAI,CAAC;IAClB,CAAC;IAED,IAAI,UAAU;QACb,IAAI,IAAI,CAAC,CAAC,GAAG,CAAC;YAAE,OAAO,CAAC,CAAC;QACzB,OAAO,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,GAAG,CAAC,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;IAC1C,CAAC;CACD;AAED,MAAM,OAAO,cAAc;IACT,MAAM,CAAS;IACf,KAAK,CAAS;IACd,GAAG,CAAS;IACZ,KAAK,GAAG,IAAI,YAAY,EAAE,CAAC;IACpC,UAAU,GAAG,CAAC,CAAC;IAEvB,YAAY,OAA2B,EAAE;QACxC,IAAI,CAAC,MAAM,GAAG,IAAI,CAAC,YAAY,IAAI,qBAAqB,CAAC;QACzD,IAAI,CAAC,KAAK,GAAG,IAAI,CAAC,SAAS,IAAI,kBAAkB,CAAC;QAClD,IAAI,CAAC,GAAG,GAAG,IAAI,CAAC,SAAS,IAAI,kBAAkB,CAAC;IACjD,CAAC;IAED;;;;;;;;;;;;OAYG;IACH,OAAO,CAAC,KAAa,EAAE,UAAkB;QACxC,MAAM,GAAG,GAAG,IAAI,CAAC,UAAU,CAAC;QAC5B,IAAI,CAAC,UAAU,EAAE,CAAC;QAClB,IAAI,GAAG,IAAI,IAAI,CAAC,GAAG,EAAE,CAAC;YACrB,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE,YAAY,EAAE,CAAC;QAC7C,CAAC;QACD,IAAI,GAAG,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC;YACvB,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;YAC5B,OAAO,EAAE,IAAI,EAAE,KAAK,EAAE,CAAC;QACxB,CAAC;QACD,MAAM,IAAI,GAAG,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC;QACpC,MAAM,GAAG,GAAG,IAAI,CAAC,KAAK,CAAC,UAAU,CAAC;QAClC,MAAM,SAAS,GAAG,IAAI,GAAG,IAAI,CAAC,KAAK,GAAG,GAAG,CAAC;QAC1C,IAAI,UAAU,GAAG,SAAS,IAAI,iBAAiB,CAAC,KAAK,CAAC,EAAE,CAAC;YACxD,OAAO,EAAE,IAAI,EAAE,IAAI,EAAE,MAAM,EAAE,eAAe,EAAE,IAAI,EAAE,GAAG,EAAE,CAAC;QAC3D,CAAC;QACD,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;QAC5B,OAAO,EAAE,IAAI,EAAE,KAAK,EAAE,IAAI,EAAE,GAAG,EAAE,CAAC;IACnC,CAAC;IAED;;;;OAIG;IACH,QAAQ;QACP,OAAO;YACN,UAAU,EAAE,IAAI,CAAC,UAAU;YAC3B,IAAI,EAAE,IAAI,CAAC,KAAK,CAAC,KAAK,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC,CAAC,IAAI;YAC3D,GAAG,EAAE,IAAI,CAAC,KAAK,CAAC,KAAK,IAAI,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,KAAK,CAAC,UAAU,CAAC,CAAC,CAAC,IAAI;SACzD,CAAC;IACH,CAAC;CACD;AAED;;;;GAIG;AACH,MAAM,UAAU,iBAAiB,CAAC,KAAa;IAC9C,IAAI,CAAC,KAAK;QAAE,OAAO,KAAK,CAAC;IACzB,gEAAgE;IAChE,iEAAiE;IACjE,6CAA6C;IAC7C,MAAM,EAAE,GAAG,KAAK,CAAC,CAAC,CAAE,CAAC;IACrB,OAAO,mBAAmB,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;AACrC,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,WAAW,CAC1B,MAAuD,EACvD,OAA2B,EAAE;IAE7B,MAAM,OAAO,GAAG,IAAI,cAAc,CAAC,IAAI,CAAC,CAAC;IACzC,IAAI,IAAI,GAAG,EAAE,CAAC;IACd,KAAK,MAAM,EAAE,KAAK,EAAE,UAAU,EAAE,IAAI,MAAM,EAAE,CAAC;QAC5C,MAAM,CAAC,GAAG,OAAO,CAAC,OAAO,CAAC,KAAK,EAAE,UAAU,CAAC,CAAC;QAC7C,IAAI,CAAC,CAAC,IAAI;YAAE,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,IAAI,EAAE,MAAM,EAAE,CAAC,CAAC,MAAM,EAAE,CAAC;QAC7D,IAAI,IAAI,KAAK,CAAC;IACf,CAAC;IACD,OAAO,EAAE,IAAI,EAAE,OAAO,EAAE,KAAK,EAAE,CAAC;AACjC,CAAC"}

package/dist/eval-dataset.d.ts

@@ -0,0 +1,66 @@
+/**
+ * Offline-eval dataset — capture + load. PRD §M5c.
+ *
+ * One JSONL example per committed prompt sample. The shape captures
+ * everything a replay run needs to reproduce the original ghost call:
+ * `(context, prefix, expected continuation, provider fingerprint)`.
+ *
+ * Privacy: examples DO contain raw prompt text. Capture is gated on the
+ * `PI_GHOST_EVAL_CAPTURE=1` env (not a default behavior — opt-in only).
+ * Sanitization still applies through `containsSensitive` so messages
+ * with API keys are dropped at capture time. Downstream tooling (CLI
+ * replay) ships a `--redact` flag that emits aggregate metrics without
+ * raw examples for publication.
+ */
+export declare const EVAL_REL = ".pi/ghost-autocomplete/eval.jsonl";
+/** Char positions at which to sample prefixes from a committed line. */
+export declare const PREFIX_SAMPLE_POSITIONS: readonly number[];
+/** Beyond `LONG_LINE_START`, sample every `LONG_LINE_STEP` chars. */
+export declare const LONG_LINE_START = 20;
+export declare const LONG_LINE_STEP = 10;
+export interface EvalExample {
+    ts: number;
+    /** Single-line committed text the sample was drawn from. */
+    committed: string;
+    /** Char-position sample point on `committed`. */
+    prefixLen: number;
+    /** Substring `committed[0..prefixLen]` — the LLM input. */
+    prefix: string;
+    /** Substring `committed[prefixLen..end]` — the ground truth continuation. */
+    expected: string;
+    /** Provider context (renderable). May be the empty string. */
+    context: string;
+    /** Provider/version fingerprints active at capture. */
+    providerFingerprint: string;
+    versionFingerprint: string;
+    /** Provider name (e.g. "ghost-mercury-edit"). */
+    providerName: string;
+}
+/**
+ * Generate sample (prefix, expected) splits from a committed line. PRD
+ * §M5c: positions `[3, 5, 8, 12, 20]` plus every 10 chars beyond 20.
+ * Returns empty when the committed text is too short for any sample.
+ */
+export declare function samplePrefixes(committed: string): Array<{
+    prefixLen: number;
+    prefix: string;
+    expected: string;
+}>;
+/**
+ * Build eval examples for a single commit. Drops sensitive messages
+ * entirely. `context` should already be rendered by the caller (the
+ * provider's recent-messages block joined into a single string) — keeps
+ * this module decoupled from provider context shape.
+ */
+export declare function buildExamplesFromCommit(committed: string, meta: {
+    context: string;
+    providerFingerprint: string;
+    versionFingerprint: string;
+    providerName: string;
+}, ts?: number): EvalExample[];
+export declare function appendExamples(cwd: string, examples: EvalExample[]): Promise<void>;
+export declare function loadExamples(cwd: string): Promise<EvalExample[]>;
+export declare function parseExamples(body: string): EvalExample[];
+/** Parse `PI_GHOST_EVAL_CAPTURE`. Returns true only when set to "1". */
+export declare function parseEvalCaptureEnv(raw: string | undefined): boolean;
+//# sourceMappingURL=eval-dataset.d.ts.map

package/dist/eval-dataset.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"eval-dataset.d.ts","sourceRoot":"","sources":["../src/eval-dataset.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;GAaG;AAMH,eAAO,MAAM,QAAQ,sCAAsC,CAAC;AAE5D,wEAAwE;AACxE,eAAO,MAAM,uBAAuB,EAAE,SAAS,MAAM,EAAsB,CAAC;AAC5E,qEAAqE;AACrE,eAAO,MAAM,eAAe,KAAK,CAAC;AAClC,eAAO,MAAM,cAAc,KAAK,CAAC;AAEjC,MAAM,WAAW,WAAW;IAC3B,EAAE,EAAE,MAAM,CAAC;IACX,4DAA4D;IAC5D,SAAS,EAAE,MAAM,CAAC;IAClB,iDAAiD;IACjD,SAAS,EAAE,MAAM,CAAC;IAClB,2DAA2D;IAC3D,MAAM,EAAE,MAAM,CAAC;IACf,6EAA6E;IAC7E,QAAQ,EAAE,MAAM,CAAC;IACjB,8DAA8D;IAC9D,OAAO,EAAE,MAAM,CAAC;IAChB,uDAAuD;IACvD,mBAAmB,EAAE,MAAM,CAAC;IAC5B,kBAAkB,EAAE,MAAM,CAAC;IAC3B,iDAAiD;IACjD,YAAY,EAAE,MAAM,CAAC;CACrB;AAED;;;;GAIG;AACH,wBAAgB,cAAc,CAAC,SAAS,EAAE,MAAM,GAAG,KAAK,CAAC;IAAE,SAAS,EAAE,MAAM,CAAC;IAAC,MAAM,EAAE,MAAM,CAAC;IAAC,QAAQ,EAAE,MAAM,CAAA;CAAE,CAAC,CAehH;AAED;;;;;GAKG;AACH,wBAAgB,uBAAuB,CACtC,SAAS,EAAE,MAAM,EACjB,IAAI,EAAE;IACL,OAAO,EAAE,MAAM,CAAC;IAChB,mBAAmB,EAAE,MAAM,CAAC;IAC5B,kBAAkB,EAAE,MAAM,CAAC;IAC3B,YAAY,EAAE,MAAM,CAAC;CACrB,EACD,EAAE,GAAE,MAAmB,GACrB,WAAW,EAAE,CAef;AAED,wBAAsB,cAAc,CAAC,GAAG,EAAE,MAAM,EAAE,QAAQ,EAAE,WAAW,EAAE,GAAG,OAAO,CAAC,IAAI,CAAC,CAUxF;AAED,wBAAsB,YAAY,CAAC,GAAG,EAAE,MAAM,GAAG,OAAO,CAAC,WAAW,EAAE,CAAC,CAStE;AAED,wBAAgB,aAAa,CAAC,IAAI,EAAE,MAAM,GAAG,WAAW,EAAE,CAwBzD;AAED,wEAAwE;AACxE,wBAAgB,mBAAmB,CAAC,GAAG,EAAE,MAAM,GAAG,SAAS,GAAG,OAAO,CAEpE"}