pi-harness-runtime 1.1.422 → 1.1.423

package/packages/skills/dist/sanitize.js

@@ -9,60 +9,26 @@
  *
  * Both writers therefore build their points with the same rules, and the
  * name of a point is the sanitized name. One skill, one point.
- */
-/**
- * The patterns of the client specific names.
  *
- * The list is about the names of the clients and the projects of the
- * users of the harness - not about the skills themselves.
- *
- * The patterns match a plain fragment, without a word boundary. A client
- * name rarely stands alone: it appears inside a directory name
- * (`th_digisoft_erp_chart.json`), inside a Python module (`inpac_pharma`)
- * or in a sentence ("integrated with Paperclip"). A word boundary would
- * miss all of them - and `_` is a word character, so a boundary never
- * follows an underscore.
+ * The rules themselves are not in this file. The names of the clients of
+ * the users of the harness are not in this repository - it is public, and
+ * a list of client names in it is a list of client names in the world.
+ * The rules live in a local file, and `sanitize-rules.ts` reads them. See
+ * that module for the format.
  */
-const SANITIZATION_RULES = [
-    // Client company names (case-insensitive)
-    { pattern: /inpac/gi, replacement: "[Client A]" },
-    { pattern: /m-capital/gi, replacement: "[Client A]" },
-    { pattern: /mcapital/gi, replacement: "[Client A]" },
-    { pattern: /digisoft/gi, replacement: "[Client B]" },
-    // Project/app specific names
-    { pattern: /gse-insurance/gi, replacement: "[Insurance Project]" },
-    { pattern: /tbs-import/gi, replacement: "[Import Project]" },
-    { pattern: /paperclip/gi, replacement: "[Client D]" },
-    { pattern: /openclaw/gi, replacement: "[Client F]" },
-    // Generic replacements
-    { pattern: /Company Name/gi, replacement: "[Client Name]" },
-    { pattern: /your company/gi, replacement: "[Client]" },
-    { pattern: /our client/gi, replacement: "[Client]" },
-    { pattern: /client-specific/gi, replacement: "[Project-specific]" },
-];
-/**
- * The names that are renamed as a whole.
- *
- * A name that would be unreadable after the pattern rules - a name that
- * carries the client name in its middle - is renamed here.
- */
-const SKILL_NAME_MAP = {
-    "gse-insurance-step5-slot-filter": "insurance-project-step-filter",
-    "inpac-pe-approval-workflow": "client-a-pe-approval-workflow",
-    "inpac-po-approval-workflow": "client-a-po-approval-workflow",
-    "inpac-pr-approval-workflow": "client-a-pr-approval-workflow",
-    "tbs-import-clearance-lcv": "import-project-clearance",
-    whispertool: "whisper-tool",
-};
+import { loadSanitizationRules } from "./sanitize-rules.js";
+export { loadSanitizationRules, resetSanitizationRulesCache, sanitizationFilePath, sanitizationStatus, } from "./sanitize-rules.js";
 /**
  * Replace the client specific names of a text.
  *
  * The text stays the same length, so a vector that was built from it
  * keeps its shape.
+ *
+ * The rules are the ones of this installation unless a set is passed in.
  */
-export function sanitizeText(text) {
+export function sanitizeText(text, rules = loadSanitizationRules()) {
     let result = text;
-    for (const rule of SANITIZATION_RULES) {
+    for (const rule of rules.rules) {
         result = result.replace(rule.pattern, rule.replacement);
     }
     return result;
@@ -78,11 +44,11 @@ export function sanitizeText(text) {
  * returns. The replacements of the rules carry a space and brackets,
  * which no skill name may hold.
  */
-export function skillPointName(name) {
-    const mapped = SKILL_NAME_MAP[name];
+export function skillPointName(name, rules = loadSanitizationRules()) {
+    const mapped = rules.nameMap[name];
     if (mapped)
         return mapped;
-    return kebabName(sanitizeText(name));
+    return kebabName(sanitizeText(name, rules));
 }
 /**
  * Bring a name into the shape of a skill name.

package/packages/skills/dist/sanitize.js.map

@@ -1 +1 @@
-{"version":3,"file":"sanitize.js","sourceRoot":"","sources":["../src/sanitize.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;GAWG;AAQH;;;;;;;;;;;;GAYG;AACH,MAAM,kBAAkB,GAAuB;IAC7C,0CAA0C;IAC1C,EAAE,OAAO,EAAE,SAAS,EAAE,WAAW,EAAE,YAAY,EAAE;IACjD,EAAE,OAAO,EAAE,aAAa,EAAE,WAAW,EAAE,YAAY,EAAE;IACrD,EAAE,OAAO,EAAE,YAAY,EAAE,WAAW,EAAE,YAAY,EAAE;IACpD,EAAE,OAAO,EAAE,YAAY,EAAE,WAAW,EAAE,YAAY,EAAE;IAEpD,6BAA6B;IAC7B,EAAE,OAAO,EAAE,iBAAiB,EAAE,WAAW,EAAE,qBAAqB,EAAE;IAClE,EAAE,OAAO,EAAE,cAAc,EAAE,WAAW,EAAE,kBAAkB,EAAE;IAC5D,EAAE,OAAO,EAAE,aAAa,EAAE,WAAW,EAAE,YAAY,EAAE;IACrD,EAAE,OAAO,EAAE,YAAY,EAAE,WAAW,EAAE,YAAY,EAAE;IAEpD,uBAAuB;IACvB,EAAE,OAAO,EAAE,gBAAgB,EAAE,WAAW,EAAE,eAAe,EAAE;IAC3D,EAAE,OAAO,EAAE,gBAAgB,EAAE,WAAW,EAAE,UAAU,EAAE;IACtD,EAAE,OAAO,EAAE,cAAc,EAAE,WAAW,EAAE,UAAU,EAAE;IACpD,EAAE,OAAO,EAAE,mBAAmB,EAAE,WAAW,EAAE,oBAAoB,EAAE;CACpE,CAAC;AAEF;;;;;GAKG;AACH,MAAM,cAAc,GAA2B;IAC9C,iCAAiC,EAAE,+BAA+B;IAClE,4BAA4B,EAAE,+BAA+B;IAC7D,4BAA4B,EAAE,+BAA+B;IAC7D,4BAA4B,EAAE,+BAA+B;IAC7D,0BAA0B,EAAE,0BAA0B;IACtD,WAAW,EAAE,cAAc;CAC3B,CAAC;AAEF;;;;;GAKG;AACH,MAAM,UAAU,YAAY,CAAC,IAAY;IACxC,IAAI,MAAM,GAAG,IAAI,CAAC;IAClB,KAAK,MAAM,IAAI,IAAI,kBAAkB,EAAE,CAAC;QACvC,MAAM,GAAG,MAAM,CAAC,OAAO,CAAC,IAAI,CAAC,OAAO,EAAE,IAAI,CAAC,WAAW,CAAC,CAAC;IACzD,CAAC;IACD,OAAO,MAAM,CAAC;AACf,CAAC;AAED;;;;;;;;;;GAUG;AACH,MAAM,UAAU,cAAc,CAAC,IAAY;IAC1C,MAAM,MAAM,GAAG,cAAc,CAAC,IAAI,CAAC,CAAC;IACpC,IAAI,MAAM;QAAE,OAAO,MAAM,CAAC;IAC1B,OAAO,SAAS,CAAC,YAAY,CAAC,IAAI,CAAC,CAAC,CAAC;AACtC,CAAC;AAED;;;;;GAKG;AACH,SAAS,SAAS,CAAC,IAAY;IAC9B,OAAO,IAAI;SACT,WAAW,EAAE;SACb,OAAO,CAAC,aAAa,EAAE,GAAG,CAAC;SAC3B,OAAO,CAAC,QAAQ,EAAE,GAAG,CAAC;SACtB,OAAO,CAAC,UAAU,EAAE,EAAE,CAAC,CAAC;AAC3B,CAAC"}
+{"version":3,"file":"sanitize.js","sourceRoot":"","sources":["../src/sanitize.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;GAiBG;AAEH,OAAO,EAAE,qBAAqB,EAA4B,MAAM,qBAAqB,CAAC;AAGtF,OAAO,EACL,qBAAqB,EACrB,2BAA2B,EAC3B,oBAAoB,EACpB,kBAAkB,GACnB,MAAM,qBAAqB,CAAC;AAE7B;;;;;;;GAOG;AACH,MAAM,UAAU,YAAY,CAC1B,IAAY,EACZ,KAAK,GAAwB,qBAAqB,EAAE;IAErD,IAAI,MAAM,GAAG,IAAI,CAAC;IAClB,KAAK,MAAM,IAAI,IAAI,KAAK,CAAC,KAAK,EAAE,CAAC;QAChC,MAAM,GAAG,MAAM,CAAC,OAAO,CAAC,IAAI,CAAC,OAAO,EAAE,IAAI,CAAC,WAAW,CAAC,CAAC;IACzD,CAAC;IACD,OAAO,MAAM,CAAC;AACf,CAAC;AAED;;;;;;;;;;GAUG;AACH,MAAM,UAAU,cAAc,CAC5B,IAAY,EACZ,KAAK,GAAwB,qBAAqB,EAAE;IAErD,MAAM,MAAM,GAAG,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC;IACnC,IAAI,MAAM;QAAE,OAAO,MAAM,CAAC;IAC1B,OAAO,SAAS,CAAC,YAAY,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC,CAAC;AAC7C,CAAC;AAED;;;;;GAKG;AACH,SAAS,SAAS,CAAC,IAAY;IAC9B,OAAO,IAAI;SACT,WAAW,EAAE;SACb,OAAO,CAAC,aAAa,EAAE,GAAG,CAAC;SAC3B,OAAO,CAAC,QAAQ,EAAE,GAAG,CAAC;SACtB,OAAO,CAAC,UAAU,EAAE,EAAE,CAAC,CAAC;AAC3B,CAAC"}

package/packages/skills/dist/vector-schema.d.ts

@@ -15,6 +15,7 @@
  * module owns the collection - that is a decision of the skills domain, not
  * of the vector store.
  */
+import type { SanitizationRuleSet } from "./sanitize-rules.js";
 /**
  * The collection the runtime indexes into and searches.
  *
@@ -167,7 +168,12 @@ export declare function skillPointPayload(input: {
     description: string;
     /** The body of the skill. */
     body: string;
-}): SkillPointPayload;
+}, 
+/**
+ * The rules of the installation. They default to the rules of this
+ * machine, and a test passes its own rules in.
+ */
+rules?: SanitizationRuleSet): SkillPointPayload;
 export declare function describeSkillSchemaMismatch(actual: {
     dimensions?: number;
     distance?: string;

package/packages/skills/dist/vector-schema.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"vector-schema.d.ts","sourceRoot":"","sources":["../src/vector-schema.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;GAgBG;AAIH;;;;;GAKG;AACH,eAAO,MAAM,gBAAgB,gCAAgC,CAAC;AAE9D;;;;;;;GAOG;AACH,eAAO,MAAM,uBAAuB,sBAAsB,CAAC;AAE3D,0CAA0C;AAC1C,eAAO,MAAM,6BAA6B,2CAA2C,CAAC;AAEtF,gDAAgD;AAChD,eAAO,MAAM,kCAAkC,MAAM,CAAC;AAEtD;;;;;;;;;;;;;;;;;;GAkBG;AACH,eAAO,MAAM,eAAe,OAAO,CAAC;AAEpC;;;;;;GAMG;AACH,eAAO,MAAM,sBAAsB,MAAM,CAAC;AAE1C,yDAAyD;AACzD,MAAM,WAAW,iBAAiB;IACjC,uBAAuB;IACvB,UAAU,EAAE,MAAM,CAAC;IACnB,2BAA2B;IAC3B,QAAQ,EAAE,MAAM,CAAC;IACjB,6CAA6C;IAC7C,KAAK,CAAC,EAAE,MAAM,CAAC;CACf;AAED;;;;;;;;;GASG;AACH,MAAM,MAAM,iBAAiB,GAAG;IAC9B,8DAA8D;IAC9D,IAAI,EAAE,MAAM,CAAC;IACb,2BAA2B;IAC3B,WAAW,EAAE,MAAM,CAAC;IACpB,6BAA6B;IAC7B,IAAI,EAAE,MAAM,CAAC;IACb,0CAA0C;IAC1C,IAAI,EAAE,MAAM,CAAC;CACd,CAAC;AAEF;;;;;GAKG;AACH,eAAO,MAAM,wBAAwB,OAAO,CAAC;AAE7C,kDAAkD;AAClD,eAAO,MAAM,wBAAwB,OAAO,CAAC;AAE7C;;;;;;;;;;;GAWG;AACH,wBAAgB,UAAU,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,CAMhD;AAED;;;;;GAKG;AACH,wBAAgB,cAAc,CAAC,KAAK,EAAE;IACpC,gEAAgE;IAChE,IAAI,EAAE,MAAM,CAAC;IACb,6DAA6D;IAC7D,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,2BAA2B;IAC3B,WAAW,EAAE,MAAM,CAAC;IACpB,6BAA6B;IAC7B,IAAI,EAAE,MAAM,CAAC;CACd,GAAG,MAAM,CAMT;AAED;;;;;;GAMG;AACH,MAAM,WAAW,gBAAgB;IAC/B,EAAE,EAAE,MAAM,CAAC;IACX,OAAO,EAAE,iBAAiB,CAAC;CAC5B;AAED;;;;;;;;GAQG;AACH,wBAAgB,gBAAgB,CAAC,OAAO,EAAE,iBAAiB,GAAG,gBAAgB,CAE7E;AAED;;;;;;GAMG;AACH,wBAAgB,iBAAiB,CAAC,KAAK,EAAE;IACvC,gEAAgE;IAChE,IAAI,EAAE,MAAM,CAAC;IACb,6DAA6D;IAC7D,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,2BAA2B;IAC3B,WAAW,EAAE,MAAM,CAAC;IACpB,6BAA6B;IAC7B,IAAI,EAAE,MAAM,CAAC;CACd,GAAG,iBAAiB,CAWpB;AACD,wBAAgB,2BAA2B,CAC1C,MAAM,EAAE;IAAE,UAAU,CAAC,EAAE,MAAM,CAAC;IAAC,QAAQ,CAAC,EAAE,MAAM,CAAA;CAAE,EAClD,QAAQ,EAAE,iBAAiB,GACzB,MAAM,GAAG,IAAI,CASf"}
+{"version":3,"file":"vector-schema.d.ts","sourceRoot":"","sources":["../src/vector-schema.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;GAgBG;AAGH,OAAO,KAAK,EAAE,mBAAmB,EAAE,MAAM,qBAAqB,CAAC;AAE/D;;;;;GAKG;AACH,eAAO,MAAM,gBAAgB,gCAAgC,CAAC;AAE9D;;;;;;;GAOG;AACH,eAAO,MAAM,uBAAuB,sBAAsB,CAAC;AAE3D,0CAA0C;AAC1C,eAAO,MAAM,6BAA6B,2CAA2C,CAAC;AAEtF,gDAAgD;AAChD,eAAO,MAAM,kCAAkC,MAAM,CAAC;AAEtD;;;;;;;;;;;;;;;;;;GAkBG;AACH,eAAO,MAAM,eAAe,OAAO,CAAC;AAEpC;;;;;;GAMG;AACH,eAAO,MAAM,sBAAsB,MAAM,CAAC;AAE1C,yDAAyD;AACzD,MAAM,WAAW,iBAAiB;IACjC,uBAAuB;IACvB,UAAU,EAAE,MAAM,CAAC;IACnB,2BAA2B;IAC3B,QAAQ,EAAE,MAAM,CAAC;IACjB,6CAA6C;IAC7C,KAAK,CAAC,EAAE,MAAM,CAAC;CACf;AAED;;;;;;;;;GASG;AACH,MAAM,MAAM,iBAAiB,GAAG;IAC9B,8DAA8D;IAC9D,IAAI,EAAE,MAAM,CAAC;IACb,2BAA2B;IAC3B,WAAW,EAAE,MAAM,CAAC;IACpB,6BAA6B;IAC7B,IAAI,EAAE,MAAM,CAAC;IACb,0CAA0C;IAC1C,IAAI,EAAE,MAAM,CAAC;CACd,CAAC;AAEF;;;;;GAKG;AACH,eAAO,MAAM,wBAAwB,OAAO,CAAC;AAE7C,kDAAkD;AAClD,eAAO,MAAM,wBAAwB,OAAO,CAAC;AAE7C;;;;;;;;;;;GAWG;AACH,wBAAgB,UAAU,CAAC,KAAK,EAAE,MAAM,GAAG,MAAM,CAMhD;AAED;;;;;GAKG;AACH,wBAAgB,cAAc,CAAC,KAAK,EAAE;IACpC,gEAAgE;IAChE,IAAI,EAAE,MAAM,CAAC;IACb,6DAA6D;IAC7D,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,2BAA2B;IAC3B,WAAW,EAAE,MAAM,CAAC;IACpB,6BAA6B;IAC7B,IAAI,EAAE,MAAM,CAAC;CACd,GAAG,MAAM,CAMT;AAED;;;;;;GAMG;AACH,MAAM,WAAW,gBAAgB;IAC/B,EAAE,EAAE,MAAM,CAAC;IACX,OAAO,EAAE,iBAAiB,CAAC;CAC5B;AAED;;;;;;;;GAQG;AACH,wBAAgB,gBAAgB,CAAC,OAAO,EAAE,iBAAiB,GAAG,gBAAgB,CAE7E;AAED;;;;;;GAMG;AACH,wBAAgB,iBAAiB,CAC/B,KAAK,EAAE;IACL,gEAAgE;IAChE,IAAI,EAAE,MAAM,CAAC;IACb,6DAA6D;IAC7D,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,2BAA2B;IAC3B,WAAW,EAAE,MAAM,CAAC;IACpB,6BAA6B;IAC7B,IAAI,EAAE,MAAM,CAAC;CACd;AACD;;;GAGG;AACH,KAAK,CAAC,EAAE,mBAAmB,GAC1B,iBAAiB,CAWnB;AACD,wBAAgB,2BAA2B,CAC1C,MAAM,EAAE;IAAE,UAAU,CAAC,EAAE,MAAM,CAAC;IAAC,QAAQ,CAAC,EAAE,MAAM,CAAA;CAAE,EAClD,QAAQ,EAAE,iBAAiB,GACzB,MAAM,GAAG,IAAI,CASf"}

package/packages/skills/dist/vector-schema.js

@@ -125,11 +125,16 @@ export function skillQdrantPoint(payload) {
  * of a user must not travel into it. Both writers use this function, so
  * both write the same sanitized payload for the same skill.
  */
-export function skillPointPayload(input) {
-    const name = skillPointName(input.name);
-    const title = input.title ? skillPointName(input.title) : undefined;
-    const description = sanitizeText(input.description);
-    const body = sanitizeText(input.body).slice(0, SKILL_PAYLOAD_BODY_LIMIT);
+export function skillPointPayload(input, 
+/**
+ * The rules of the installation. They default to the rules of this
+ * machine, and a test passes its own rules in.
+ */
+rules) {
+    const name = skillPointName(input.name, rules);
+    const title = input.title ? skillPointName(input.title, rules) : undefined;
+    const description = sanitizeText(input.description, rules);
+    const body = sanitizeText(input.body, rules).slice(0, SKILL_PAYLOAD_BODY_LIMIT);
     return {
         name,
         description,

package/packages/skills/dist/vector-schema.js.map

@@ -1 +1 @@
-{"version":3,"file":"vector-schema.js","sourceRoot":"","sources":["../src/vector-schema.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;GAgBG;AAEH,OAAO,EAAE,YAAY,EAAE,cAAc,EAAE,MAAM,eAAe,CAAC;AAE7D;;;;;GAKG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAAG,6BAA6B,CAAC;AAE9D;;;;;;;GAOG;AACH,MAAM,CAAC,MAAM,uBAAuB,GAAG,mBAAmB,CAAC;AAE3D,0CAA0C;AAC1C,MAAM,CAAC,MAAM,6BAA6B,GAAG,wCAAwC,CAAC;AAEtF,gDAAgD;AAChD,MAAM,CAAC,MAAM,kCAAkC,GAAG,GAAG,CAAC;AAEtD;;;;;;;;;;;;;;;;;;GAkBG;AACH,MAAM,CAAC,MAAM,eAAe,GAAG,IAAI,CAAC;AAEpC;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,sBAAsB,GAAG,GAAG,CAAC;AAiC1C;;;;;GAKG;AACH,MAAM,CAAC,MAAM,wBAAwB,GAAG,IAAI,CAAC;AAE7C,kDAAkD;AAClD,MAAM,CAAC,MAAM,wBAAwB,GAAG,IAAI,CAAC;AAE7C;;;;;;;;;;;GAWG;AACH,MAAM,UAAU,UAAU,CAAC,KAAa;IACtC,IAAI,IAAI,GAAG,CAAC,CAAC;IACb,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACtC,IAAI,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,IAAI,CAAC,GAAG,KAAK,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC;IACzD,CAAC;IACD,OAAO,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;AACxB,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,cAAc,CAAC,KAS9B;IACC,MAAM,KAAK,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IAC3B,IAAI,KAAK,CAAC,KAAK,IAAI,KAAK,CAAC,KAAK,KAAK,KAAK,CAAC,IAAI;QAAE,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC;IACvE,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC;IAC9B,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,wBAAwB,CAAC,CAAC,CAAC;IAC1D,OAAO,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;AACzB,CAAC;AAcD;;;;;;;;GAQG;AACH,MAAM,UAAU,gBAAgB,CAAC,OAA0B;IACzD,OAAO,EAAE,EAAE,EAAE,UAAU,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE,OAAO,EAAE,CAAC;AACnD,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,iBAAiB,CAAC,KASjC;IACC,MAAM,IAAI,GAAG,cAAc,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IACxC,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,cAAc,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;IACpE,MAAM,WAAW,GAAG,YAAY,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC;IACpD,MAAM,IAAI,GAAG,YAAY,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,KAAK,CAAC,CAAC,EAAE,wBAAwB,CAAC,CAAC;IACzE,OAAO;QACL,IAAI;QACJ,WAAW;QACX,IAAI;QACJ,IAAI,EAAE,cAAc,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,WAAW,EAAE,IAAI,EAAE,CAAC;KACzD,CAAC;AACJ,CAAC;AACD,MAAM,UAAU,2BAA2B,CAC1C,MAAkD,EAClD,QAA2B;IAE3B,IAAI,MAAM,CAAC,UAAU,KAAK,SAAS,IAAI,MAAM,CAAC,UAAU,KAAK,QAAQ,CAAC,UAAU,EAAE,CAAC;QAClF,MAAM,KAAK,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,KAAK,QAAQ,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC;QAC3D,OAAO,wBAAwB,MAAM,CAAC,UAAU,iDAAiD,QAAQ,CAAC,UAAU,GAAG,KAAK,EAAE,CAAC;IAChI,CAAC;IACD,IAAI,MAAM,CAAC,QAAQ,KAAK,SAAS,IAAI,MAAM,CAAC,QAAQ,KAAK,QAAQ,CAAC,QAAQ,EAAE,CAAC;QAC5E,OAAO,2BAA2B,MAAM,CAAC,QAAQ,kCAAkC,QAAQ,CAAC,QAAQ,EAAE,CAAC;IACxG,CAAC;IACD,OAAO,IAAI,CAAC;AACb,CAAC"}
+{"version":3,"file":"vector-schema.js","sourceRoot":"","sources":["../src/vector-schema.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;GAgBG;AAEH,OAAO,EAAE,YAAY,EAAE,cAAc,EAAE,MAAM,eAAe,CAAC;AAG7D;;;;;GAKG;AACH,MAAM,CAAC,MAAM,gBAAgB,GAAG,6BAA6B,CAAC;AAE9D;;;;;;;GAOG;AACH,MAAM,CAAC,MAAM,uBAAuB,GAAG,mBAAmB,CAAC;AAE3D,0CAA0C;AAC1C,MAAM,CAAC,MAAM,6BAA6B,GAAG,wCAAwC,CAAC;AAEtF,gDAAgD;AAChD,MAAM,CAAC,MAAM,kCAAkC,GAAG,GAAG,CAAC;AAEtD;;;;;;;;;;;;;;;;;;GAkBG;AACH,MAAM,CAAC,MAAM,eAAe,GAAG,IAAI,CAAC;AAEpC;;;;;;GAMG;AACH,MAAM,CAAC,MAAM,sBAAsB,GAAG,GAAG,CAAC;AAiC1C;;;;;GAKG;AACH,MAAM,CAAC,MAAM,wBAAwB,GAAG,IAAI,CAAC;AAE7C,kDAAkD;AAClD,MAAM,CAAC,MAAM,wBAAwB,GAAG,IAAI,CAAC;AAE7C;;;;;;;;;;;GAWG;AACH,MAAM,UAAU,UAAU,CAAC,KAAa;IACtC,IAAI,IAAI,GAAG,CAAC,CAAC;IACb,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,KAAK,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACtC,IAAI,GAAG,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,EAAE,IAAI,CAAC,GAAG,KAAK,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC;IACzD,CAAC;IACD,OAAO,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;AACxB,CAAC;AAED;;;;;GAKG;AACH,MAAM,UAAU,cAAc,CAAC,KAS9B;IACC,MAAM,KAAK,GAAG,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IAC3B,IAAI,KAAK,CAAC,KAAK,IAAI,KAAK,CAAC,KAAK,KAAK,KAAK,CAAC,IAAI;QAAE,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,KAAK,CAAC,CAAC;IACvE,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,WAAW,CAAC,CAAC;IAC9B,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,EAAE,wBAAwB,CAAC,CAAC,CAAC;IAC1D,OAAO,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;AACzB,CAAC;AAcD;;;;;;;;GAQG;AACH,MAAM,UAAU,gBAAgB,CAAC,OAA0B;IACzD,OAAO,EAAE,EAAE,EAAE,UAAU,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE,OAAO,EAAE,CAAC;AACnD,CAAC;AAED;;;;;;GAMG;AACH,MAAM,UAAU,iBAAiB,CAC/B,KASC;AACD;;;GAGG;AACH,KAA2B;IAE3B,MAAM,IAAI,GAAG,cAAc,CAAC,KAAK,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC;IAC/C,MAAM,KAAK,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,cAAc,CAAC,KAAK,CAAC,KAAK,EAAE,KAAK,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;IAC3E,MAAM,WAAW,GAAG,YAAY,CAAC,KAAK,CAAC,WAAW,EAAE,KAAK,CAAC,CAAC;IAC3D,MAAM,IAAI,GAAG,YAAY,CAAC,KAAK,CAAC,IAAI,EAAE,KAAK,CAAC,CAAC,KAAK,CAAC,CAAC,EAAE,wBAAwB,CAAC,CAAC;IAChF,OAAO;QACL,IAAI;QACJ,WAAW;QACX,IAAI;QACJ,IAAI,EAAE,cAAc,CAAC,EAAE,IAAI,EAAE,KAAK,EAAE,WAAW,EAAE,IAAI,EAAE,CAAC;KACzD,CAAC;AACJ,CAAC;AACD,MAAM,UAAU,2BAA2B,CAC1C,MAAkD,EAClD,QAA2B;IAE3B,IAAI,MAAM,CAAC,UAAU,KAAK,SAAS,IAAI,MAAM,CAAC,UAAU,KAAK,QAAQ,CAAC,UAAU,EAAE,CAAC;QAClF,MAAM,KAAK,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,KAAK,QAAQ,CAAC,KAAK,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC;QAC3D,OAAO,wBAAwB,MAAM,CAAC,UAAU,iDAAiD,QAAQ,CAAC,UAAU,GAAG,KAAK,EAAE,CAAC;IAChI,CAAC;IACD,IAAI,MAAM,CAAC,QAAQ,KAAK,SAAS,IAAI,MAAM,CAAC,QAAQ,KAAK,QAAQ,CAAC,QAAQ,EAAE,CAAC;QAC5E,OAAO,2BAA2B,MAAM,CAAC,QAAQ,kCAAkC,QAAQ,CAAC,QAAQ,EAAE,CAAC;IACxG,CAAC;IACD,OAAO,IAAI,CAAC;AACb,CAAC"}

package/packages/skills/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@pi-harness/skills",
-  "version": "1.1.422",
+  "version": "1.1.423",
   "description": "File-based skill system with pi.dev compatibility and Hermes self-improvement",
   "type": "module",
   "main": "src/index.ts",

package/packages/skills/src/index.ts

@@ -24,6 +24,9 @@ export * from "./linter.js";
 import type { SkillsConfig, ScanResult, ScanOptions } from "./types.js";
 import { scanForSkillsSync } from "./scanner.js";
 import { getGlobalSkillRegistry, type SkillRegistry } from "./loader.js";
+import { existsSync, readdirSync } from "node:fs";
+import { homedir } from "node:os";
+import { join } from "node:path";
 
 /**
  * The result of the skill initialization.
@@ -73,16 +76,20 @@ export function initSkills(
   // Hint: if ~/.pi-harness-runtime/skills/ is empty, suggest the sync script
   checkEmptySkillsDir();
 
-  return { ...result, registry };
+  // The scan can find the same skill twice: a symlinked directory is
+  // scanned twice, because the scanner follows the path it is given. The
+  // registry keeps one entry per id, so the count of the local skills is
+  // the count of the DISTINCT skills - and not the count of the files
+  // that were read.
+  const distinct = new Map<string, (typeof result.skills)[number]>();
+  for (const skill of result.skills) distinct.set(skill.id, skill);
+
+  return { ...result, skills: [...distinct.values()], registry };
 }
 
 /** Show sync hint if the harness skills dir is empty. */
 function checkEmptySkillsDir(): void {
   try {
-    const { readdirSync, existsSync } = require("node:fs");
-    const { join } = require("node:path");
-    const { homedir } = require("node:os");
-
     const harnessDir = join(homedir(), ".pi-harness-runtime", "skills");
     if (!existsSync(harnessDir)) return;
 

package/packages/skills/src/sanitize-rules.ts

@@ -0,0 +1,218 @@
+/**
+ * The rules that keep a client name out of the shared collection.
+ *
+ * The names of the clients of the users of the harness are **not** in this
+ * repository. The repository is public, and a list of client names in it
+ * is a list of client names in the world - the names of the projects and
+ * of the clients of every person who installs the package.
+ *
+ * The rules therefore live in a local file:
+ *
+ *   ~/.pi-harness-runtime/skill-sanitization.json
+ *
+ * and this module is the mechanism that reads them and applies them. The
+ * file is per installation, so a machine holds the rules of its own
+ * clients and nobody else's.
+ *
+ * The file is optional. Without it only the generic rules apply, and the
+ * name of a client would travel into the collection of that installation.
+ * `sanitizationStatus()` tells the caller which of the two is the case,
+ * so a missing file is a visible state and not a silent one.
+ *
+ * The format of the file:
+ *
+ * ```json
+ * {
+ *   "rules": [
+ *     { "pattern": "acme", "replacement": "[Client A]", "flags": "gi" }
+ *   ],
+ *   "nameMap": {
+ *     "acme-approval-workflow": "client-a-approval-workflow"
+ *   }
+ * }
+ * ```
+ *
+ * The pattern is the source of a regular expression and the flags default
+ * to "gi". The rules match a plain fragment, without a word boundary: a
+ * client name rarely stands alone, it appears inside a directory name
+ * (`erp_report_2024.json`), inside a module name (`acme_pharma`) or in a
+ * sentence ("integrated with Acme"). A word boundary would miss all of
+ * them - and `_` is a word character, so a boundary never follows an
+ * underscore.
+ */
+
+import { readFileSync } from "node:fs";
+import { homedir } from "node:os";
+import { join } from "node:path";
+
+/** A rule as it is written in the local file. */
+export interface SanitizationRuleSpec {
+  /** The source of a regular expression. */
+  pattern: string;
+  /** What a match becomes. */
+  replacement: string;
+  /** The flags of the regular expression. Defaults to "gi". */
+  flags?: string;
+}
+
+/** A compiled rule. */
+export interface SanitizationRule {
+  pattern: RegExp;
+  replacement: string;
+}
+
+/** The whole rule set of one installation. */
+export interface SanitizationRuleSet {
+  /** The rules, in the order they are applied. */
+  rules: SanitizationRule[];
+  /** The names that are renamed as a whole. */
+  nameMap: Record<string, string>;
+  /** Where the rules came from. */
+  source: "local-file" | "generic-only";
+  /** Why the local file did not contribute its rules. */
+  error?: string;
+}
+
+/**
+ * The rules that hold for every installation.
+ *
+ * They are about the way a text talks about a client, not about the name
+ * of a client: "Company Name" is a placeholder in a template, and it
+ * must not become a client name in the collection.
+ */
+const GENERIC_RULES: SanitizationRule[] = [
+  { pattern: /Company Name/gi, replacement: "[Client Name]" },
+  { pattern: /your company/gi, replacement: "[Client]" },
+  { pattern: /our client/gi, replacement: "[Client]" },
+  { pattern: /client-specific/gi, replacement: "[Project-specific]" },
+];
+
+/**
+ * The names that are renamed as a whole, for every installation.
+ *
+ * A name that would be unreadable after the pattern rules - a name that
+ * carries the client name in its middle - is renamed here.
+ */
+const GENERIC_NAME_MAP: Record<string, string> = {
+  whispertool: "whisper-tool",
+};
+
+/** Where the rules of this installation live. */
+export function sanitizationFilePath(): string {
+  const override = process.env.PI_SKILL_SANITIZATION_FILE;
+  if (override && override.trim()) return override.trim();
+  const home = process.env.HOME || process.env.USERPROFILE || homedir();
+  return join(home, ".pi-harness-runtime", "skill-sanitization.json");
+}
+
+/**
+ * The state of the rules of this installation.
+ *
+ * `configured` is false when the local file is missing or unreadable. The
+ * caller shows it, because a missing file is a weaker privacy guarantee
+ * and not a detail.
+ */
+export function sanitizationStatus(): {
+  configured: boolean;
+  path: string;
+  ruleCount: number;
+  error?: string;
+} {
+  const path = sanitizationFilePath();
+  const set = loadSanitizationRules(path);
+  return {
+    configured: set.source === "local-file",
+    path,
+    ruleCount: set.rules.length,
+    ...(set.error ? { error: set.error } : {}),
+  };
+}
+
+let cached: { path: string; set: SanitizationRuleSet } | undefined;
+
+/**
+ * The rules of this installation.
+ *
+ * The file is read once per process: the rules do not change while the
+ * harness runs, and a skill that is indexed a hundred times must not read
+ * a file a hundred times.
+ */
+export function loadSanitizationRules(path = sanitizationFilePath()): SanitizationRuleSet {
+  if (cached && cached.path === path) return cached.set;
+
+  const set = readSanitizationRules(path);
+  cached = { path, set };
+  return set;
+}
+
+/** Forget the cached rules. A test needs it, a new file needs it. */
+export function resetSanitizationRulesCache(): void {
+  cached = undefined;
+}
+
+function readSanitizationRules(path: string): SanitizationRuleSet {
+  let raw: string;
+  try {
+    raw = readFileSync(path, "utf8");
+  } catch (err) {
+    const code = (err as NodeJS.ErrnoException)?.code;
+    if (code !== "ENOENT") {
+      return {
+        rules: [...GENERIC_RULES],
+        nameMap: { ...GENERIC_NAME_MAP },
+        source: "generic-only",
+        error: `The sanitization rules at ${path} could not be read: ${err instanceof Error ? err.message : String(err)}`,
+      };
+    }
+    return {
+      rules: [...GENERIC_RULES],
+      nameMap: { ...GENERIC_NAME_MAP },
+      source: "generic-only",
+    };
+  }
+
+  let parsed: unknown;
+  try {
+    parsed = JSON.parse(raw);
+  } catch (err) {
+    return {
+      rules: [...GENERIC_RULES],
+      nameMap: { ...GENERIC_NAME_MAP },
+      source: "generic-only",
+      error: `The sanitization rules at ${path} are not valid JSON: ${err instanceof Error ? err.message : String(err)}`,
+    };
+  }
+
+  const rules: SanitizationRule[] = [...GENERIC_RULES];
+  const nameMap: Record<string, string> = { ...GENERIC_NAME_MAP };
+
+  const file = parsed as {
+    rules?: SanitizationRuleSpec[];
+    nameMap?: Record<string, string>;
+  };
+
+  for (const spec of file.rules ?? []) {
+    if (!spec || typeof spec.pattern !== "string" || typeof spec.replacement !== "string") {
+      continue;
+    }
+    try {
+      rules.push({
+        pattern: new RegExp(spec.pattern, spec.flags ?? "gi"),
+        replacement: spec.replacement,
+      });
+    } catch {
+      // A broken pattern must not take the whole rule set down: the other
+      // rules still hold, and the broken one is reported below.
+      rules.push({
+        pattern: /$^/,
+        replacement: spec.replacement,
+      });
+    }
+  }
+
+  for (const [from, to] of Object.entries(file.nameMap ?? {})) {
+    if (typeof from === "string" && typeof to === "string") nameMap[from] = to;
+  }
+
+  return { rules, nameMap, source: "local-file" };
+}

package/packages/skills/src/sanitize.ts

@@ -9,71 +9,38 @@
  *
  * Both writers therefore build their points with the same rules, and the
  * name of a point is the sanitized name. One skill, one point.
- */
-
-/** A rule that replaces a client specific text. */
-interface SanitizationRule {
-  pattern: RegExp;
-  replacement: string;
-}
-
-/**
- * The patterns of the client specific names.
- *
- * The list is about the names of the clients and the projects of the
- * users of the harness - not about the skills themselves.
  *
- * The patterns match a plain fragment, without a word boundary. A client
- * name rarely stands alone: it appears inside a directory name
- * (`th_digisoft_erp_chart.json`), inside a Python module (`inpac_pharma`)
- * or in a sentence ("integrated with Paperclip"). A word boundary would
- * miss all of them - and `_` is a word character, so a boundary never
- * follows an underscore.
+ * The rules themselves are not in this file. The names of the clients of
+ * the users of the harness are not in this repository - it is public, and
+ * a list of client names in it is a list of client names in the world.
+ * The rules live in a local file, and `sanitize-rules.ts` reads them. See
+ * that module for the format.
  */
-const SANITIZATION_RULES: SanitizationRule[] = [
-  // Client company names (case-insensitive)
-  { pattern: /inpac/gi, replacement: "[Client A]" },
-  { pattern: /m-capital/gi, replacement: "[Client A]" },
-  { pattern: /mcapital/gi, replacement: "[Client A]" },
-  { pattern: /digisoft/gi, replacement: "[Client B]" },
 
-  // Project/app specific names
-  { pattern: /gse-insurance/gi, replacement: "[Insurance Project]" },
-  { pattern: /tbs-import/gi, replacement: "[Import Project]" },
-  { pattern: /paperclip/gi, replacement: "[Client D]" },
-  { pattern: /openclaw/gi, replacement: "[Client F]" },
+import { loadSanitizationRules, type SanitizationRuleSet } from "./sanitize-rules.js";
 
-  // Generic replacements
-  { pattern: /Company Name/gi, replacement: "[Client Name]" },
-  { pattern: /your company/gi, replacement: "[Client]" },
-  { pattern: /our client/gi, replacement: "[Client]" },
-  { pattern: /client-specific/gi, replacement: "[Project-specific]" },
-];
-
-/**
- * The names that are renamed as a whole.
- *
- * A name that would be unreadable after the pattern rules - a name that
- * carries the client name in its middle - is renamed here.
- */
-const SKILL_NAME_MAP: Record<string, string> = {
-	"gse-insurance-step5-slot-filter": "insurance-project-step-filter",
-	"inpac-pe-approval-workflow": "client-a-pe-approval-workflow",
-	"inpac-po-approval-workflow": "client-a-po-approval-workflow",
-	"inpac-pr-approval-workflow": "client-a-pr-approval-workflow",
-	"tbs-import-clearance-lcv": "import-project-clearance",
-	whispertool: "whisper-tool",
-};
+export type { SanitizationRule, SanitizationRuleSet } from "./sanitize-rules.js";
+export {
+  loadSanitizationRules,
+  resetSanitizationRulesCache,
+  sanitizationFilePath,
+  sanitizationStatus,
+} from "./sanitize-rules.js";
 
 /**
  * Replace the client specific names of a text.
  *
  * The text stays the same length, so a vector that was built from it
  * keeps its shape.
+ *
+ * The rules are the ones of this installation unless a set is passed in.
  */
-export function sanitizeText(text: string): string {
+export function sanitizeText(
+  text: string,
+  rules: SanitizationRuleSet = loadSanitizationRules(),
+): string {
 	let result = text;
-	for (const rule of SANITIZATION_RULES) {
+	for (const rule of rules.rules) {
 		result = result.replace(rule.pattern, rule.replacement);
 	}
 	return result;
@@ -90,10 +57,13 @@ export function sanitizeText(text: string): string {
  * returns. The replacements of the rules carry a space and brackets,
  * which no skill name may hold.
  */
-export function skillPointName(name: string): string {
-	const mapped = SKILL_NAME_MAP[name];
+export function skillPointName(
+  name: string,
+  rules: SanitizationRuleSet = loadSanitizationRules(),
+): string {
+	const mapped = rules.nameMap[name];
 	if (mapped) return mapped;
-	return kebabName(sanitizeText(name));
+	return kebabName(sanitizeText(name, rules));
 }
 
 /**

package/packages/skills/src/vector-schema.ts

@@ -17,6 +17,7 @@
  */
 
 import { sanitizeText, skillPointName } from "./sanitize.js";
+import type { SanitizationRuleSet } from "./sanitize-rules.js";
 
 /**
  * The collection the runtime indexes into and searches.
@@ -189,20 +190,27 @@ export function skillQdrantPoint(payload: SkillPointPayload): SkillQdrantPoint {
  * of a user must not travel into it. Both writers use this function, so
  * both write the same sanitized payload for the same skill.
  */
-export function skillPointPayload(input: {
-  /** The name the payload carries - the identity of the skill. */
-  name: string;
-  /** The title of the skill, when it differs from the name. */
-  title?: string;
-  /** What the skill does. */
-  description: string;
-  /** The body of the skill. */
-  body: string;
-}): SkillPointPayload {
-  const name = skillPointName(input.name);
-  const title = input.title ? skillPointName(input.title) : undefined;
-  const description = sanitizeText(input.description);
-  const body = sanitizeText(input.body).slice(0, SKILL_PAYLOAD_BODY_LIMIT);
+export function skillPointPayload(
+  input: {
+    /** The name the payload carries - the identity of the skill. */
+    name: string;
+    /** The title of the skill, when it differs from the name. */
+    title?: string;
+    /** What the skill does. */
+    description: string;
+    /** The body of the skill. */
+    body: string;
+  },
+  /**
+   * The rules of the installation. They default to the rules of this
+   * machine, and a test passes its own rules in.
+   */
+  rules?: SanitizationRuleSet,
+): SkillPointPayload {
+  const name = skillPointName(input.name, rules);
+  const title = input.title ? skillPointName(input.title, rules) : undefined;
+  const description = sanitizeText(input.description, rules);
+  const body = sanitizeText(input.body, rules).slice(0, SKILL_PAYLOAD_BODY_LIMIT);
   return {
     name,
     description,