crewly 1.6.4 → 1.6.5

package/dist/backend/backend/src/services/intent-task/intent-classifier.rules.d.ts

@@ -0,0 +1,331 @@
+/**
+ * Intent classifier rules — code-of-record per Mia spec §5 Q4.
+ *
+ * **Purpose**
+ *
+ * Centralises every keyword set and structural pattern the L0/L1/L2/L3
+ * classifier consumes. Both production classifier paths
+ * (`backend/src/types/intent-task.types.ts:classifyIntentLevel/Category`
+ * and `backend/src/services/v3/v3-data.service.ts:classifyIntent`) read
+ * from this file so a rule update lands in exactly one place.
+ *
+ * **Why a separate file?**
+ *
+ * Per Mia: "Easier review than buried regex." A reviewer auditing rule
+ * coverage scans this file's exports rather than walking the classifier's
+ * inline regex. Adding ZH parity to a category is a single-edit operation
+ * here; in the inline shape it would be a ~10-line walk.
+ *
+ * **Spec source-of-truth**
+ *
+ * `specs/2026-05-06-intent-classifier-rules.md` (Mia, 2026-05-06).
+ * Section references (§2.1, §2.2, etc.) below correspond to that doc.
+ *
+ * @module services/intent-task/intent-classifier.rules
+ */
+/**
+ * Quantifier tokens that, when paired with a state-changing verb AND a
+ * plural deliverable noun, force L1 → L2 promotion (§2.1).
+ *
+ * Anti-pattern guard (§3 last paragraph): quantifier alone is NOT enough.
+ * "Show me all 3 statuses" stays L0 because the verb is read-only.
+ * Promotion requires quantifier + state-changing verb + plural deliverable
+ * noun — see {@link L2_QUANTIFIER_PROMOTERS_TRIGRAM}.
+ */
+export declare const QUANTIFIER_TOKENS: {
+    /** EN: both/all/every/each + numeric + numeric-words. */
+    readonly en: RegExp;
+    /** ZH: 这两/三/四/五/.../多个/所有/全部/每个/几个. */
+    readonly zh: RegExp;
+};
+/**
+ * Plural deliverable noun tokens that, paired with a quantifier + state-
+ * changing verb, force L2 promotion (§2.1).
+ *
+ * The ZH measure-words (份/个/张/篇/段/章/项/条) double as deliverable-noun
+ * proxies because Chinese uses measure-words in lieu of pluralisation.
+ */
+export declare const DELIVERABLE_NOUN_TOKENS: {
+    readonly en: RegExp;
+    /**
+     * ZH: 文档/文件/规范 + measure-words (份/个/张/篇/段/章/项/条). Combined
+     * because Chinese parses noun + measure-word as the deliverable proxy.
+     */
+    readonly zh: RegExp;
+};
+/**
+ * Quantifier-trigram tokens (§2.1 + §3 anti-pattern guard).
+ *
+ * **Implementation note:** rather than constraining token order in a
+ * single regex, the classifier checks all three axes (quantifier +
+ * state-change verb + deliverable noun) for presence within the same
+ * input. This handles all natural orderings (VERB-QUANT-NOUN as in
+ * "create three reports" / "做这两份 md 文档"; QUANT-NOUN-VERB as in
+ * "three reports to make" / "三份报告要做"; QUANT-VERB-NOUN; etc.)
+ * without explosion of permutations.
+ *
+ * The anti-promotion guard in `classifyIntentLevel` ensures
+ * "Show me all 3 statuses" stays L0 by checking that the verb axis is
+ * NOT exclusively read-only ({@link L0_READ_VERB_PATTERN}) before
+ * promoting on this trigram.
+ */
+export declare const L2_QUANTIFIER_TRIGRAM: {
+    /**
+     * EN: requires all three axes — quantifier + state-change verb +
+     * deliverable noun — present in the input. Order-agnostic.
+     *
+     * The exposed regex matches the QUANTIFIER axis only; the consumer
+     * checks the other two axes via the helper below.
+     */
+    readonly en: RegExp;
+    /** ZH quantifier axis. */
+    readonly zh: RegExp;
+};
+/**
+ * State-change verb axis for the quantifier-trigram. Distinct from
+ * {@link ACTION_VERB_EN}/{@link ACTION_VERB_ZH} because the trigram
+ * specifically excludes ambiguous-shape verbs (e.g. `run`, `check`)
+ * that are common in read-only contexts.
+ */
+export declare const L2_TRIGRAM_STATE_CHANGE_VERB: {
+    readonly en: RegExp;
+    readonly zh: RegExp;
+};
+/**
+ * Plural-deliverable axis for the quantifier-trigram. Mirrors
+ * {@link DELIVERABLE_NOUN_TOKENS} but kept distinct so a future tuning
+ * of "what counts as deliverable in a trigram context" doesn't drift.
+ */
+export declare const L2_TRIGRAM_DELIVERABLE: {
+    readonly en: RegExp;
+    readonly zh: RegExp;
+};
+/**
+ * Doc/spec creation triggers (§2.2 row 1). These promote to L2 even with
+ * a tiny verb because authoring a doc is multi-step by nature.
+ */
+export declare const L2_DOC_CREATION_TRIGGER: {
+    readonly en: RegExp;
+    readonly zh: RegExp;
+};
+/**
+ * File-with-extension reference (§2.2 row 2). Any path with extension
+ * promotes to L2 unless the verb is read-only — the read-only guard is
+ * applied at the classifier level, not here, because read-only verbs are
+ * a separate axis (see {@link L0_READ_VERB_PATTERN}).
+ */
+export declare const L2_FILE_EXTENSION_TRIGGER: RegExp;
+/**
+ * PR / commit / branch artifact (§2.2 row 3). Order-agnostic — both
+ * "open a PR" (VERB-NOUN) and "PR to open" (NOUN-VERB) match, with the
+ * VERB-NOUN form being far more common.
+ */
+export declare const L2_PR_ARTIFACT_TRIGGER: {
+    readonly en: RegExp;
+    readonly zh: RegExp;
+};
+/**
+ * Migration / schema (§2.2 row 4).
+ */
+export declare const L2_MIGRATION_TRIGGER: {
+    readonly en: RegExp;
+    readonly zh: RegExp;
+};
+/**
+ * Build/implement + system-noun (§2.3 row 1). The canonical L2 shape —
+ * pre-existing in the EN classifier; ZH parity is the new contribution.
+ */
+export declare const L2_BUILD_SYSTEM_NOUN: {
+    readonly en: RegExp;
+    readonly zh: RegExp;
+};
+/**
+ * Deploy/migrate + env-noun (§2.3 row 2).
+ */
+export declare const L2_DEPLOY_ENV_NOUN: {
+    readonly en: RegExp;
+    readonly zh: RegExp;
+};
+/**
+ * End-to-end / full-stack scope (§2.3 row 3).
+ */
+export declare const L2_E2E_SCOPE: {
+    readonly en: RegExp;
+    readonly zh: RegExp;
+};
+/**
+ * Coordinate / orchestrate (§2.3 row 4). Single-token signal — by the
+ * time someone says "coordinate", the work is multi-agent by definition.
+ */
+export declare const L2_COORDINATE: {
+    readonly en: RegExp;
+    readonly zh: RegExp;
+};
+/**
+ * Sequenced verbs (§2.4 row 1). "X then Y" / "X 然后 Y".
+ */
+export declare const L2_SEQUENCED_VERBS: {
+    readonly en: RegExp;
+    readonly zh: RegExp;
+};
+/**
+ * Numbered/bulleted list (§2.4 row 3). Two or more numbered items in the
+ * same input is structurally L2.
+ */
+/**
+ * Numbered list trigger. Matches "1. … 2. …" / "(1) … (2) …" / "① … ②".
+ * The `\s?` after the marker is OPTIONAL because Chinese-encoded numbered
+ * lists ("①修复 ②测试") have no whitespace between the marker and the
+ * content.
+ */
+export declare const L2_NUMBERED_LIST: RegExp;
+/**
+ * Restart/recovery + redirect (§2.5). Status-update clause followed by a
+ * directive that references prior multi-step context. The "this/这个"
+ * referent points to a prior plan the classifier can't resolve, but the
+ * SHAPE is reliable enough to promote.
+ */
+export declare const L2_CROSS_SYSTEM_PIVOT: {
+    readonly en: RegExp;
+    /**
+     * ZH: status-clause (重启/回来/上线/跑起来/好了/恢复) + ≤16-char gap +
+     * directive prefix (现在/那/你?) + (再/继续/接着/开始) + verb (做/搞/跑/继续).
+     */
+    readonly zh: RegExp;
+};
+/**
+ * Sprint / milestone / epic markers (§2.6 row 1). These are L3 CANDIDATES
+ * — combined with §2.3 they classify L3; alone they classify L2.
+ */
+export declare const L3_SPRINT_OKR_MARKERS: {
+    readonly en: RegExp;
+    readonly zh: RegExp;
+};
+/**
+ * Multi-day timebox (§2.6 row 2). Time markers like "this week" /
+ * "by Friday" / "这周" / "下周" combined with a verb suggest sprint-level
+ * scope.
+ */
+export declare const L3_TIMEBOX_MARKERS: {
+    readonly en: RegExp;
+    readonly zh: RegExp;
+};
+/**
+ * Deliver / ship verbs (§2.6 row 3). Distinct from §2.3 deploy axis —
+ * "ship" / "上线" carry a sprint-completion semantic.
+ */
+export declare const L3_DELIVER_SHIP_VERBS: {
+    readonly en: RegExp;
+    readonly zh: RegExp;
+};
+/**
+ * Read-only / status-lookup verbs. If the only verb in the input is a
+ * read verb, do NOT promote even if a quantifier is present
+ * (§3 last paragraph: "Show me all 3 statuses" stays L0).
+ */
+export declare const L0_READ_VERB_PATTERN: {
+    readonly en: RegExp;
+    readonly zh: RegExp;
+};
+/**
+ * EN action verb pattern. Used as a positive signal for actionability
+ * when no non-actionable pattern matches. Pre-existing — kept here to
+ * keep all classifier inputs in one file.
+ */
+export declare const ACTION_VERB_EN: RegExp;
+/**
+ * ZH action verb pattern. Mirrors §2.3 + §2.6 verb sets plus the canonical
+ * Chinese imperatives Mia called out in the spec preamble (§0 root cause
+ * 1): 做/写/搞/搞定/重启/检查 etc.
+ *
+ * **Strictness on dual-use verbs (Mia 2026-05-06):** the bare imperatives
+ * 做 and 搞 are conversational-filler-ambiguous (做了 = "did already",
+ * 搞砸 = "messed up", 搞不定 = "can't do"). To follow Mia's
+ * stricter-promotion default ("false-negative is fixable; false-positive
+ * over-decomposes"), we use negative lookaheads on those two so the
+ * completed/denial forms don't match. Other actionable forms (做完, 搞定,
+ * 做+deliverable) are explicit fall-through tokens further in the pattern.
+ *
+ * The companion {@link NON_ACTIONABLE_PATTERNS_ZH} set includes
+ * sentence-shape filters for "I just did X" / "I can't do X" cases that
+ * the lookahead alone cannot catch — both layers together produce the
+ * stricter behaviour Mia asked for.
+ */
+export declare const ACTION_VERB_ZH: RegExp;
+/**
+ * Non-actionable ZH patterns that should be filtered out by
+ * `isActionableIntent` before classification (§3 filtered rows).
+ *
+ * Exported as an array for composition with the existing EN
+ * NON_ACTIONABLE_PATTERNS in `intent-task.types.ts`.
+ */
+export declare const NON_ACTIONABLE_PATTERNS_ZH: RegExp[];
+/**
+ * Length floors for L2 promotion when no rule fires. ZH has no spaces, so
+ * word-count alone under-estimates message density — char-count is the
+ * complementary axis.
+ */
+export declare const L2_LENGTH_FLOORS: {
+    /** EN/space-tokenised word count above which the message is L2. */
+    readonly wordCount: 35;
+    /** Total character count above which the message is L2 (catches ZH). */
+    readonly charCount: 80;
+};
+/**
+ * Category-specific keyword sets. Each category combines an EN pattern
+ * with a ZH pattern; the classifier matches if EITHER fires.
+ *
+ * Exported as a structured object so the consumer code reads
+ * `CATEGORY_KEYWORDS.debugging.zh.test(input)` rather than digging into
+ * nested ternaries.
+ *
+ * Order is significant — see `classifyIntentCategory` in
+ * `intent-task.types.ts`. The most-specific categories run first.
+ */
+export declare const CATEGORY_KEYWORDS: {
+    readonly debugging: {
+        readonly en: RegExp;
+        readonly zh: RegExp;
+    };
+    readonly deployment: {
+        readonly en: RegExp;
+        readonly zh: RegExp;
+    };
+    readonly release: {
+        readonly en: RegExp;
+        readonly zh: RegExp;
+    };
+    readonly ciCdContext: {
+        readonly en: RegExp;
+        readonly zh: RegExp;
+    };
+    readonly ciCdAction: {
+        readonly en: RegExp;
+        readonly zh: RegExp;
+    };
+    readonly review: {
+        readonly en: RegExp;
+        readonly zh: RegExp;
+    };
+    readonly research: {
+        readonly en: RegExp;
+        readonly zh: RegExp;
+    };
+    readonly planning: {
+        readonly en: RegExp;
+        readonly zh: RegExp;
+    };
+    readonly communication: {
+        readonly en: RegExp;
+        readonly zh: RegExp;
+    };
+    readonly codeChange: {
+        readonly en: RegExp;
+        readonly zh: RegExp;
+    };
+    readonly query: {
+        readonly en: RegExp;
+        readonly zh: RegExp;
+    };
+};
+//# sourceMappingURL=intent-classifier.rules.d.ts.map

package/dist/backend/backend/src/services/intent-task/intent-classifier.rules.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"intent-classifier.rules.d.ts","sourceRoot":"","sources":["../../../../../../backend/src/services/intent-task/intent-classifier.rules.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;GAwBG;AAMH;;;;;;;;GAQG;AACH,eAAO,MAAM,iBAAiB;IAC5B,yDAAyD;;IAEzD,uCAAuC;;CAE/B,CAAC;AAEX;;;;;;GAMG;AACH,eAAO,MAAM,uBAAuB;;IAElC;;;OAGG;;CAEK,CAAC;AAEX;;;;;;;;;;;;;;;GAeG;AACH,eAAO,MAAM,qBAAqB;IAChC;;;;;;OAMG;;IAEH,0BAA0B;;CAElB,CAAC;AAEX;;;;;GAKG;AACH,eAAO,MAAM,4BAA4B;;;CAG/B,CAAC;AAEX;;;;GAIG;AACH,eAAO,MAAM,sBAAsB;;;CAGzB,CAAC;AAMX;;;GAGG;AACH,eAAO,MAAM,uBAAuB;;;CAG1B,CAAC;AAEX;;;;;GAKG;AACH,eAAO,MAAM,yBAAyB,QAA+H,CAAC;AAEtK;;;;GAIG;AACH,eAAO,MAAM,sBAAsB;;;CAGzB,CAAC;AAEX;;GAEG;AACH,eAAO,MAAM,oBAAoB;;;CAGvB,CAAC;AAMX;;;GAGG;AACH,eAAO,MAAM,oBAAoB;;;CAGvB,CAAC;AAEX;;GAEG;AACH,eAAO,MAAM,kBAAkB;;;CAGrB,CAAC;AAEX;;GAEG;AACH,eAAO,MAAM,YAAY;;;CAGf,CAAC;AAEX;;;GAGG;AACH,eAAO,MAAM,aAAa;;;CAGhB,CAAC;AAMX;;GAEG;AACH,eAAO,MAAM,kBAAkB;;;CAGrB,CAAC;AAEX;;;GAGG;AACH;;;;;GAKG;AACH,eAAO,MAAM,gBAAgB,QAAgE,CAAC;AAM9F;;;;;GAKG;AACH,eAAO,MAAM,qBAAqB;;IAEhC;;;OAGG;;CAEK,CAAC;AAMX;;;GAGG;AACH,eAAO,MAAM,qBAAqB;;;CAGxB,CAAC;AAEX;;;;GAIG;AACH,eAAO,MAAM,kBAAkB;;;CAGrB,CAAC;AAEX;;;GAGG;AACH,eAAO,MAAM,qBAAqB;;;CAGxB,CAAC;AAMX;;;;GAIG;AACH,eAAO,MAAM,oBAAoB;;;CAGvB,CAAC;AAMX;;;;GAIG;AACH,eAAO,MAAM,cAAc,QAAmjB,CAAC;AAE/kB;;;;;;;;;;;;;;;;;GAiBG;AAQH,eAAO,MAAM,cAAc,QAAmS,CAAC;AAM/T;;;;;;GAMG;AACH,eAAO,MAAM,0BAA0B,EAAE,MAAM,EA8B9C,CAAC;AAMF;;;;GAIG;AACH,eAAO,MAAM,gBAAgB;IAC3B,mEAAmE;;IAEnE,wEAAwE;;CAEhE,CAAC;AAMX;;;;;;;;;;GAUG;AACH,eAAO,MAAM,iBAAiB;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;CA0DpB,CAAC"}