@remnic/core 9.3.608 → 9.3.609
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/access-cli.js +2 -2
- package/dist/{chunk-LCR46JY5.js → chunk-7YX23JBA.js} +40 -3
- package/dist/chunk-7YX23JBA.js.map +1 -0
- package/dist/{chunk-YJ6QCQNE.js → chunk-HJNQQICM.js} +2 -2
- package/dist/extraction-judge-training.js +1 -1
- package/dist/index.js +2 -2
- package/dist/orchestrator.js +2 -2
- package/dist/schemas.d.ts +22 -22
- package/dist/transfer/types.d.ts +12 -12
- package/package.json +1 -1
- package/src/extraction-judge-training.ts +49 -21
- package/dist/chunk-LCR46JY5.js.map +0 -1
- /package/dist/{chunk-YJ6QCQNE.js.map → chunk-HJNQQICM.js.map} +0 -0
package/dist/access-cli.js
CHANGED
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
import {
|
|
2
2
|
Orchestrator
|
|
3
|
-
} from "./chunk-
|
|
3
|
+
} from "./chunk-HJNQQICM.js";
|
|
4
4
|
import "./chunk-5RIRL3XL.js";
|
|
5
5
|
import "./chunk-KVEVLBKC.js";
|
|
6
6
|
import "./chunk-BFBF3XEF.js";
|
|
@@ -47,7 +47,7 @@ import "./chunk-VU3SVYMA.js";
|
|
|
47
47
|
import "./chunk-H63EDPFJ.js";
|
|
48
48
|
import "./chunk-PD6O7AXF.js";
|
|
49
49
|
import "./chunk-YAZNBMNF.js";
|
|
50
|
-
import "./chunk-
|
|
50
|
+
import "./chunk-7YX23JBA.js";
|
|
51
51
|
import "./chunk-C4SQJZAF.js";
|
|
52
52
|
import "./chunk-KM2A35EO.js";
|
|
53
53
|
import "./chunk-4RA3C3EV.js";
|
|
@@ -5,7 +5,7 @@ import {
|
|
|
5
5
|
// src/extraction-judge-training.ts
|
|
6
6
|
import path from "path";
|
|
7
7
|
import { homedir } from "os";
|
|
8
|
-
import { appendFile, chmod, mkdir, readFile, readdir } from "fs/promises";
|
|
8
|
+
import { appendFile, chmod, lstat, mkdir, readFile, readdir, realpath } from "fs/promises";
|
|
9
9
|
function expandTilde(p) {
|
|
10
10
|
const home = homedir();
|
|
11
11
|
if (p === "~" || p.startsWith("~/") || p.startsWith("~\\")) {
|
|
@@ -36,6 +36,10 @@ function dateStamp(iso) {
|
|
|
36
36
|
function trainingFilePathFor(directory, iso) {
|
|
37
37
|
return path.join(directory, `${dateStamp(iso)}.jsonl`);
|
|
38
38
|
}
|
|
39
|
+
function isPathInsideDirectory(filePath, directory) {
|
|
40
|
+
const relative = path.relative(directory, filePath);
|
|
41
|
+
return relative === "" || !relative.startsWith("..") && !path.isAbsolute(relative);
|
|
42
|
+
}
|
|
39
43
|
async function recordJudgeTrainingPair(row, options) {
|
|
40
44
|
if (!options.enabled) return;
|
|
41
45
|
const dir = resolveTrainingDir(options);
|
|
@@ -56,6 +60,19 @@ async function recordJudgeTrainingPair(row, options) {
|
|
|
56
60
|
}
|
|
57
61
|
async function readJudgeTrainingPairs(options) {
|
|
58
62
|
const dir = resolveTrainingDir({ enabled: true, ...options });
|
|
63
|
+
try {
|
|
64
|
+
const dirStat = await lstat(dir);
|
|
65
|
+
if (dirStat.isSymbolicLink()) {
|
|
66
|
+
throw new Error("Judge training directory must not be a symlink");
|
|
67
|
+
}
|
|
68
|
+
if (!dirStat.isDirectory()) {
|
|
69
|
+
throw new Error("Judge training path must be a directory");
|
|
70
|
+
}
|
|
71
|
+
} catch (err) {
|
|
72
|
+
const code = err.code;
|
|
73
|
+
if (code === "ENOENT") return { rows: [], malformed: 0 };
|
|
74
|
+
throw err;
|
|
75
|
+
}
|
|
59
76
|
let entries;
|
|
60
77
|
try {
|
|
61
78
|
entries = await readdir(dir);
|
|
@@ -66,10 +83,30 @@ async function readJudgeTrainingPairs(options) {
|
|
|
66
83
|
}
|
|
67
84
|
const rows = [];
|
|
68
85
|
let malformed = 0;
|
|
86
|
+
const resolvedDir = await realpath(dir);
|
|
69
87
|
entries.sort();
|
|
70
88
|
for (const name of entries) {
|
|
71
89
|
if (!name.endsWith(".jsonl")) continue;
|
|
72
|
-
const
|
|
90
|
+
const filePath = path.join(dir, name);
|
|
91
|
+
let fileStat;
|
|
92
|
+
try {
|
|
93
|
+
fileStat = await lstat(filePath);
|
|
94
|
+
} catch (err) {
|
|
95
|
+
const code = err.code;
|
|
96
|
+
if (code === "ENOENT") continue;
|
|
97
|
+
throw err;
|
|
98
|
+
}
|
|
99
|
+
if (fileStat.isSymbolicLink() || !fileStat.isFile()) continue;
|
|
100
|
+
let resolvedFilePath;
|
|
101
|
+
try {
|
|
102
|
+
resolvedFilePath = await realpath(filePath);
|
|
103
|
+
} catch (err) {
|
|
104
|
+
const code = err.code;
|
|
105
|
+
if (code === "ENOENT") continue;
|
|
106
|
+
throw err;
|
|
107
|
+
}
|
|
108
|
+
if (!isPathInsideDirectory(resolvedFilePath, resolvedDir)) continue;
|
|
109
|
+
const raw = await readFile(resolvedFilePath, "utf-8");
|
|
73
110
|
for (const line of raw.split("\n")) {
|
|
74
111
|
if (!line.trim()) continue;
|
|
75
112
|
let parsed;
|
|
@@ -120,4 +157,4 @@ export {
|
|
|
120
157
|
readJudgeTrainingPairs,
|
|
121
158
|
isValidTrainingPair
|
|
122
159
|
};
|
|
123
|
-
//# sourceMappingURL=chunk-
|
|
160
|
+
//# sourceMappingURL=chunk-7YX23JBA.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"sources":["../src/extraction-judge-training.ts"],"sourcesContent":["/**\n * Extraction Judge Training Data Shim (issue #562, PR 4).\n *\n * Opt-in collector for `(candidate_text, verdict_kind, reason,\n * ground_truth_label?)` tuples. Rows are appended to JSONL files under\n * `~/.remnic/judge-training/<YYYY-MM-DD>.jsonl` so operators can ship the\n * data into a future GRPO training pipeline without exfiltrating live\n * memory content through the regular observation ledger.\n *\n * Gating:\n * - Off by default. Must be explicitly enabled via\n * `collectJudgeTrainingPairs: true` in plugin config.\n * - The ground-truth label is always optional — labels are added out-of-\n * band once reviewers disambiguate the candidate's fate.\n *\n * Privacy: the row carries only what the judge already sees — the\n * candidate text and its metadata. It does NOT carry session keys,\n * principal IDs, or any user identifiers. The file lives in the user's\n * home directory rather than the shared memory directory so it is never\n * committed, sync'd, or bundled into exports.\n */\n\nimport path from \"node:path\";\nimport { homedir } from \"node:os\";\nimport { appendFile, chmod, lstat, mkdir, readFile, readdir, realpath } from \"node:fs/promises\";\nimport { log } from \"./logger.js\";\nimport type { JudgeVerdictKind } from \"./extraction-judge.js\";\n\n/**\n * Persisted training row. Intentionally minimal: just the signal needed\n * to train a judge replacement policy. Schema version is tagged so future\n * readers can migrate older rows.\n */\nexport interface JudgeTrainingPair {\n version: 1;\n ts: string; // ISO-8601\n candidateText: string;\n candidateCategory: string;\n candidateConfidence?: number;\n verdictKind: JudgeVerdictKind;\n reason: string;\n /**\n * Number of prior deferrals when the verdict was resolved. `0` for the\n * first resolution; only set when known (defer pathway).\n */\n priorDeferrals?: number;\n /**\n * Optional human-applied ground-truth label. Added after the fact by a\n * reviewer / labelling script; not present on fresh rows.\n */\n groundTruthLabel?: JudgeVerdictKind;\n}\n\nexport interface JudgeTrainingOptions {\n enabled: boolean;\n /**\n * Override for the output directory. Defaults to\n * `~/.remnic/judge-training`. Tests pass a temp path here.\n */\n directory?: string;\n}\n\n/**\n * Expand a leading `~` / `~/` / `$HOME/` / `${HOME}/` to the process home\n * directory. Node's `fs` APIs do not expand `~` themselves (CLAUDE.md\n * gotcha 17), so every user-facing path input must be funnelled through\n * this helper before it reaches the filesystem.\n */\nfunction expandTilde(p: string): string {\n const home = homedir();\n if (p === \"~\" || p.startsWith(\"~/\") || p.startsWith(\"~\\\\\")) {\n return home + p.slice(1);\n }\n if (p === \"$HOME\" || p.startsWith(\"$HOME/\") || p.startsWith(\"$HOME\\\\\")) {\n return home + p.slice(5);\n }\n if (p === \"${HOME}\" || p.startsWith(\"${HOME}/\") || p.startsWith(\"${HOME}\\\\\")) {\n return home + p.slice(7);\n }\n return p;\n}\n\nexport function resolveTrainingDir(options: JudgeTrainingOptions): string {\n if (options.directory && options.directory.length > 0) {\n // Expand `~` / `$HOME` in the override so operators can write the\n // config as the user sees it (CLAUDE.md gotcha 17).\n return expandTilde(options.directory);\n }\n return path.join(homedir(), \".remnic\", \"judge-training\");\n}\n\nfunction dateStamp(iso: string): string {\n // `YYYY-MM-DD` from an ISO-8601 string. Falls back to today on a parse\n // failure rather than throwing — the caller already wrote a row and the\n // timestamp is best-effort.\n const ms = Date.parse(iso);\n const d = Number.isFinite(ms) ? new Date(ms) : new Date();\n const yyyy = d.getUTCFullYear().toString().padStart(4, \"0\");\n const mm = (d.getUTCMonth() + 1).toString().padStart(2, \"0\");\n const dd = d.getUTCDate().toString().padStart(2, \"0\");\n return `${yyyy}-${mm}-${dd}`;\n}\n\nexport function trainingFilePathFor(directory: string, iso: string): string {\n return path.join(directory, `${dateStamp(iso)}.jsonl`);\n}\n\nfunction isPathInsideDirectory(filePath: string, directory: string): boolean {\n const relative = path.relative(directory, filePath);\n return relative === \"\" || (!relative.startsWith(\"..\") && !path.isAbsolute(relative));\n}\n\n/**\n * Append a single training row. Fails open — write errors are logged at\n * debug level and swallowed, same policy as the telemetry emitter.\n * No-op when `options.enabled` is false.\n */\nexport async function recordJudgeTrainingPair(row: JudgeTrainingPair, options: JudgeTrainingOptions): Promise<void> {\n if (!options.enabled) return;\n const dir = resolveTrainingDir(options);\n const filePath = trainingFilePathFor(dir, row.ts);\n try {\n await mkdir(dir, { recursive: true, mode: 0o700 });\n await appendFile(filePath, `${JSON.stringify(row)}\\n`, {\n encoding: \"utf-8\",\n mode: 0o600,\n });\n await chmod(filePath, 0o600);\n } catch (err) {\n log.debug(\n `extraction-judge-training: append failed (non-fatal): ${err instanceof Error ? err.message : String(err)}`\n );\n }\n}\n\n/**\n * Read all training rows from the configured directory. Returns an empty\n * array when the directory is missing. Malformed lines are skipped and\n * counted in the returned `malformed` tally.\n */\nexport async function readJudgeTrainingPairs(\n options: Pick<JudgeTrainingOptions, \"directory\">\n): Promise<{ rows: JudgeTrainingPair[]; malformed: number }> {\n const dir = resolveTrainingDir({ enabled: true, ...options });\n try {\n const dirStat = await lstat(dir);\n if (dirStat.isSymbolicLink()) {\n throw new Error(\"Judge training directory must not be a symlink\");\n }\n if (!dirStat.isDirectory()) {\n throw new Error(\"Judge training path must be a directory\");\n }\n } catch (err) {\n const code = (err as NodeJS.ErrnoException).code;\n if (code === \"ENOENT\") return { rows: [], malformed: 0 };\n throw err;\n }\n\n let entries: string[];\n try {\n entries = await readdir(dir);\n } catch (err) {\n const code = (err as NodeJS.ErrnoException).code;\n if (code === \"ENOENT\") return { rows: [], malformed: 0 };\n throw err;\n }\n\n const rows: JudgeTrainingPair[] = [];\n let malformed = 0;\n const resolvedDir = await realpath(dir);\n // Sort so reads are deterministic across platforms.\n entries.sort();\n for (const name of entries) {\n if (!name.endsWith(\".jsonl\")) continue;\n const filePath = path.join(dir, name);\n let fileStat: Awaited<ReturnType<typeof lstat>>;\n try {\n fileStat = await lstat(filePath);\n } catch (err) {\n const code = (err as NodeJS.ErrnoException).code;\n if (code === \"ENOENT\") continue;\n throw err;\n }\n if (fileStat.isSymbolicLink() || !fileStat.isFile()) continue;\n\n let resolvedFilePath: string;\n try {\n resolvedFilePath = await realpath(filePath);\n } catch (err) {\n const code = (err as NodeJS.ErrnoException).code;\n if (code === \"ENOENT\") continue;\n throw err;\n }\n if (!isPathInsideDirectory(resolvedFilePath, resolvedDir)) continue;\n\n const raw = await readFile(resolvedFilePath, \"utf-8\");\n for (const line of raw.split(\"\\n\")) {\n if (!line.trim()) continue;\n let parsed: unknown;\n try {\n parsed = JSON.parse(line);\n } catch {\n malformed += 1;\n continue;\n }\n if (!isValidTrainingPair(parsed)) {\n malformed += 1;\n continue;\n }\n rows.push(parsed);\n }\n }\n return { rows, malformed };\n}\n\n/**\n * Structural validator matching the persisted schema. Forward-compat: an\n * unknown `verdictKind` string is treated as malformed (strict training\n * signal — we do not want to admit unlabelled gibberish into a trainer).\n */\nexport function isValidTrainingPair(value: unknown): value is JudgeTrainingPair {\n if (typeof value !== \"object\" || value === null || Array.isArray(value)) {\n return false;\n }\n const p = value as Record<string, unknown>;\n if (p.version !== 1) return false;\n if (typeof p.ts !== \"string\") return false;\n if (typeof p.candidateText !== \"string\") return false;\n if (typeof p.candidateCategory !== \"string\") return false;\n if (p.verdictKind !== \"accept\" && p.verdictKind !== \"reject\" && p.verdictKind !== \"defer\") {\n return false;\n }\n if (typeof p.reason !== \"string\") return false;\n if (p.candidateConfidence !== undefined && typeof p.candidateConfidence !== \"number\") {\n return false;\n }\n if (p.priorDeferrals !== undefined && typeof p.priorDeferrals !== \"number\") {\n return false;\n }\n if (\n p.groundTruthLabel !== undefined &&\n p.groundTruthLabel !== \"accept\" &&\n p.groundTruthLabel !== \"reject\" &&\n p.groundTruthLabel !== \"defer\"\n ) {\n return false;\n }\n return true;\n}\n"],"mappings":";;;;;AAsBA,OAAO,UAAU;AACjB,SAAS,eAAe;AACxB,SAAS,YAAY,OAAO,OAAO,OAAO,UAAU,SAAS,gBAAgB;AA4C7E,SAAS,YAAY,GAAmB;AACtC,QAAM,OAAO,QAAQ;AACrB,MAAI,MAAM,OAAO,EAAE,WAAW,IAAI,KAAK,EAAE,WAAW,KAAK,GAAG;AAC1D,WAAO,OAAO,EAAE,MAAM,CAAC;AAAA,EACzB;AACA,MAAI,MAAM,WAAW,EAAE,WAAW,QAAQ,KAAK,EAAE,WAAW,SAAS,GAAG;AACtE,WAAO,OAAO,EAAE,MAAM,CAAC;AAAA,EACzB;AACA,MAAI,MAAM,aAAa,EAAE,WAAW,UAAU,KAAK,EAAE,WAAW,WAAW,GAAG;AAC5E,WAAO,OAAO,EAAE,MAAM,CAAC;AAAA,EACzB;AACA,SAAO;AACT;AAEO,SAAS,mBAAmB,SAAuC;AACxE,MAAI,QAAQ,aAAa,QAAQ,UAAU,SAAS,GAAG;AAGrD,WAAO,YAAY,QAAQ,SAAS;AAAA,EACtC;AACA,SAAO,KAAK,KAAK,QAAQ,GAAG,WAAW,gBAAgB;AACzD;AAEA,SAAS,UAAU,KAAqB;AAItC,QAAM,KAAK,KAAK,MAAM,GAAG;AACzB,QAAM,IAAI,OAAO,SAAS,EAAE,IAAI,IAAI,KAAK,EAAE,IAAI,oBAAI,KAAK;AACxD,QAAM,OAAO,EAAE,eAAe,EAAE,SAAS,EAAE,SAAS,GAAG,GAAG;AAC1D,QAAM,MAAM,EAAE,YAAY,IAAI,GAAG,SAAS,EAAE,SAAS,GAAG,GAAG;AAC3D,QAAM,KAAK,EAAE,WAAW,EAAE,SAAS,EAAE,SAAS,GAAG,GAAG;AACpD,SAAO,GAAG,IAAI,IAAI,EAAE,IAAI,EAAE;AAC5B;AAEO,SAAS,oBAAoB,WAAmB,KAAqB;AAC1E,SAAO,KAAK,KAAK,WAAW,GAAG,UAAU,GAAG,CAAC,QAAQ;AACvD;AAEA,SAAS,sBAAsB,UAAkB,WAA4B;AAC3E,QAAM,WAAW,KAAK,SAAS,WAAW,QAAQ;AAClD,SAAO,aAAa,MAAO,CAAC,SAAS,WAAW,IAAI,KAAK,CAAC,KAAK,WAAW,QAAQ;AACpF;AAOA,eAAsB,wBAAwB,KAAwB,SAA8C;AAClH,MAAI,CAAC,QAAQ,QAAS;AACtB,QAAM,MAAM,mBAAmB,OAAO;AACtC,QAAM,WAAW,oBAAoB,KAAK,IAAI,EAAE;AAChD,MAAI;AACF,UAAM,MAAM,KAAK,EAAE,WAAW,MAAM,MAAM,IAAM,CAAC;AACjD,UAAM,WAAW,UAAU,GAAG,KAAK,UAAU,GAAG,CAAC;AAAA,GAAM;AAAA,MACrD,UAAU;AAAA,MACV,MAAM;AAAA,IACR,CAAC;AACD,UAAM,MAAM,UAAU,GAAK;AAAA,EAC7B,SAAS,KAAK;AACZ,QAAI;AAAA,MACF,yDAAyD,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG,CAAC;AAAA,IAC3G;AAAA,EACF;AACF;AAOA,eAAsB,uBACpB,SAC2D;AAC3D,QAAM,MAAM,mBAAmB,EAAE,SAAS,MAAM,GAAG,QAAQ,CAAC;AAC5D,MAAI;AACF,UAAM,UAAU,MAAM,MAAM,GAAG;AAC/B,QAAI,QAAQ,eAAe,GAAG;AAC5B,YAAM,IAAI,MAAM,gDAAgD;AAAA,IAClE;AACA,QAAI,CAAC,QAAQ,YAAY,GAAG;AAC1B,YAAM,IAAI,MAAM,yCAAyC;AAAA,IAC3D;AAAA,EACF,SAAS,KAAK;AACZ,UAAM,OAAQ,IAA8B;AAC5C,QAAI,SAAS,SAAU,QAAO,EAAE,MAAM,CAAC,GAAG,WAAW,EAAE;AACvD,UAAM;AAAA,EACR;AAEA,MAAI;AACJ,MAAI;AACF,cAAU,MAAM,QAAQ,GAAG;AAAA,EAC7B,SAAS,KAAK;AACZ,UAAM,OAAQ,IAA8B;AAC5C,QAAI,SAAS,SAAU,QAAO,EAAE,MAAM,CAAC,GAAG,WAAW,EAAE;AACvD,UAAM;AAAA,EACR;AAEA,QAAM,OAA4B,CAAC;AACnC,MAAI,YAAY;AAChB,QAAM,cAAc,MAAM,SAAS,GAAG;AAEtC,UAAQ,KAAK;AACb,aAAW,QAAQ,SAAS;AAC1B,QAAI,CAAC,KAAK,SAAS,QAAQ,EAAG;AAC9B,UAAM,WAAW,KAAK,KAAK,KAAK,IAAI;AACpC,QAAI;AACJ,QAAI;AACF,iBAAW,MAAM,MAAM,QAAQ;AAAA,IACjC,SAAS,KAAK;AACZ,YAAM,OAAQ,IAA8B;AAC5C,UAAI,SAAS,SAAU;AACvB,YAAM;AAAA,IACR;AACA,QAAI,SAAS,eAAe,KAAK,CAAC,SAAS,OAAO,EAAG;AAErD,QAAI;AACJ,QAAI;AACF,yBAAmB,MAAM,SAAS,QAAQ;AAAA,IAC5C,SAAS,KAAK;AACZ,YAAM,OAAQ,IAA8B;AAC5C,UAAI,SAAS,SAAU;AACvB,YAAM;AAAA,IACR;AACA,QAAI,CAAC,sBAAsB,kBAAkB,WAAW,EAAG;AAE3D,UAAM,MAAM,MAAM,SAAS,kBAAkB,OAAO;AACpD,eAAW,QAAQ,IAAI,MAAM,IAAI,GAAG;AAClC,UAAI,CAAC,KAAK,KAAK,EAAG;AAClB,UAAI;AACJ,UAAI;AACF,iBAAS,KAAK,MAAM,IAAI;AAAA,MAC1B,QAAQ;AACN,qBAAa;AACb;AAAA,MACF;AACA,UAAI,CAAC,oBAAoB,MAAM,GAAG;AAChC,qBAAa;AACb;AAAA,MACF;AACA,WAAK,KAAK,MAAM;AAAA,IAClB;AAAA,EACF;AACA,SAAO,EAAE,MAAM,UAAU;AAC3B;AAOO,SAAS,oBAAoB,OAA4C;AAC9E,MAAI,OAAO,UAAU,YAAY,UAAU,QAAQ,MAAM,QAAQ,KAAK,GAAG;AACvE,WAAO;AAAA,EACT;AACA,QAAM,IAAI;AACV,MAAI,EAAE,YAAY,EAAG,QAAO;AAC5B,MAAI,OAAO,EAAE,OAAO,SAAU,QAAO;AACrC,MAAI,OAAO,EAAE,kBAAkB,SAAU,QAAO;AAChD,MAAI,OAAO,EAAE,sBAAsB,SAAU,QAAO;AACpD,MAAI,EAAE,gBAAgB,YAAY,EAAE,gBAAgB,YAAY,EAAE,gBAAgB,SAAS;AACzF,WAAO;AAAA,EACT;AACA,MAAI,OAAO,EAAE,WAAW,SAAU,QAAO;AACzC,MAAI,EAAE,wBAAwB,UAAa,OAAO,EAAE,wBAAwB,UAAU;AACpF,WAAO;AAAA,EACT;AACA,MAAI,EAAE,mBAAmB,UAAa,OAAO,EAAE,mBAAmB,UAAU;AAC1E,WAAO;AAAA,EACT;AACA,MACE,EAAE,qBAAqB,UACvB,EAAE,qBAAqB,YACvB,EAAE,qBAAqB,YACvB,EAAE,qBAAqB,SACvB;AACA,WAAO;AAAA,EACT;AACA,SAAO;AACT;","names":[]}
|
|
@@ -140,7 +140,7 @@ import {
|
|
|
140
140
|
} from "./chunk-YAZNBMNF.js";
|
|
141
141
|
import {
|
|
142
142
|
recordJudgeTrainingPair
|
|
143
|
-
} from "./chunk-
|
|
143
|
+
} from "./chunk-7YX23JBA.js";
|
|
144
144
|
import {
|
|
145
145
|
createDeferCountMap,
|
|
146
146
|
createVerdictCache,
|
|
@@ -12441,4 +12441,4 @@ export {
|
|
|
12441
12441
|
resolvePersistedMemoryRelativePath,
|
|
12442
12442
|
Orchestrator
|
|
12443
12443
|
};
|
|
12444
|
-
//# sourceMappingURL=chunk-
|
|
12444
|
+
//# sourceMappingURL=chunk-HJNQQICM.js.map
|
package/dist/index.js
CHANGED
|
@@ -182,7 +182,7 @@ import {
|
|
|
182
182
|
saveTaxonomy,
|
|
183
183
|
validateSlug,
|
|
184
184
|
validateTaxonomy
|
|
185
|
-
} from "./chunk-
|
|
185
|
+
} from "./chunk-HJNQQICM.js";
|
|
186
186
|
import "./chunk-5RIRL3XL.js";
|
|
187
187
|
import {
|
|
188
188
|
migrateFromEngram,
|
|
@@ -265,7 +265,7 @@ import {
|
|
|
265
265
|
planRecallMode
|
|
266
266
|
} from "./chunk-PD6O7AXF.js";
|
|
267
267
|
import "./chunk-YAZNBMNF.js";
|
|
268
|
-
import "./chunk-
|
|
268
|
+
import "./chunk-7YX23JBA.js";
|
|
269
269
|
import {
|
|
270
270
|
clearVerdictCache,
|
|
271
271
|
createVerdictCache,
|
package/dist/orchestrator.js
CHANGED
|
@@ -26,7 +26,7 @@ import {
|
|
|
26
26
|
sanitizeSessionKeyForFilename,
|
|
27
27
|
shouldFilterLifecycleRecallCandidate,
|
|
28
28
|
summarizeGraphShadowComparison
|
|
29
|
-
} from "./chunk-
|
|
29
|
+
} from "./chunk-HJNQQICM.js";
|
|
30
30
|
import "./chunk-5RIRL3XL.js";
|
|
31
31
|
import "./chunk-KVEVLBKC.js";
|
|
32
32
|
import "./chunk-BFBF3XEF.js";
|
|
@@ -73,7 +73,7 @@ import "./chunk-VU3SVYMA.js";
|
|
|
73
73
|
import "./chunk-H63EDPFJ.js";
|
|
74
74
|
import "./chunk-PD6O7AXF.js";
|
|
75
75
|
import "./chunk-YAZNBMNF.js";
|
|
76
|
-
import "./chunk-
|
|
76
|
+
import "./chunk-7YX23JBA.js";
|
|
77
77
|
import "./chunk-C4SQJZAF.js";
|
|
78
78
|
import "./chunk-KM2A35EO.js";
|
|
79
79
|
import "./chunk-4RA3C3EV.js";
|
package/dist/schemas.d.ts
CHANGED
|
@@ -275,12 +275,12 @@ declare const EntityMentionSchema: z.ZodObject<{
|
|
|
275
275
|
title: z.ZodString;
|
|
276
276
|
facts: z.ZodArray<z.ZodString, "many">;
|
|
277
277
|
}, "strip", z.ZodTypeAny, {
|
|
278
|
-
title: string;
|
|
279
278
|
key: string;
|
|
279
|
+
title: string;
|
|
280
280
|
facts: string[];
|
|
281
281
|
}, {
|
|
282
|
-
title: string;
|
|
283
282
|
key: string;
|
|
283
|
+
title: string;
|
|
284
284
|
facts: string[];
|
|
285
285
|
}>, "many">>>;
|
|
286
286
|
}, "strip", z.ZodTypeAny, {
|
|
@@ -288,8 +288,8 @@ declare const EntityMentionSchema: z.ZodObject<{
|
|
|
288
288
|
name: string;
|
|
289
289
|
facts: string[];
|
|
290
290
|
structuredSections?: {
|
|
291
|
-
title: string;
|
|
292
291
|
key: string;
|
|
292
|
+
title: string;
|
|
293
293
|
facts: string[];
|
|
294
294
|
}[] | null | undefined;
|
|
295
295
|
promptedByQuestion?: string | null | undefined;
|
|
@@ -298,8 +298,8 @@ declare const EntityMentionSchema: z.ZodObject<{
|
|
|
298
298
|
name: string;
|
|
299
299
|
facts: string[];
|
|
300
300
|
structuredSections?: {
|
|
301
|
-
title: string;
|
|
302
301
|
key: string;
|
|
302
|
+
title: string;
|
|
303
303
|
facts: string[];
|
|
304
304
|
}[] | null | undefined;
|
|
305
305
|
promptedByQuestion?: string | null | undefined;
|
|
@@ -584,12 +584,12 @@ declare const ProactiveExtractionResultSchema: z.ZodObject<{
|
|
|
584
584
|
title: z.ZodString;
|
|
585
585
|
facts: z.ZodArray<z.ZodString, "many">;
|
|
586
586
|
}, "strip", z.ZodTypeAny, {
|
|
587
|
-
title: string;
|
|
588
587
|
key: string;
|
|
588
|
+
title: string;
|
|
589
589
|
facts: string[];
|
|
590
590
|
}, {
|
|
591
|
-
title: string;
|
|
592
591
|
key: string;
|
|
592
|
+
title: string;
|
|
593
593
|
facts: string[];
|
|
594
594
|
}>, "many">>>;
|
|
595
595
|
}, "strip", z.ZodTypeAny, {
|
|
@@ -597,8 +597,8 @@ declare const ProactiveExtractionResultSchema: z.ZodObject<{
|
|
|
597
597
|
name: string;
|
|
598
598
|
facts: string[];
|
|
599
599
|
structuredSections?: {
|
|
600
|
-
title: string;
|
|
601
600
|
key: string;
|
|
601
|
+
title: string;
|
|
602
602
|
facts: string[];
|
|
603
603
|
}[] | null | undefined;
|
|
604
604
|
promptedByQuestion?: string | null | undefined;
|
|
@@ -607,8 +607,8 @@ declare const ProactiveExtractionResultSchema: z.ZodObject<{
|
|
|
607
607
|
name: string;
|
|
608
608
|
facts: string[];
|
|
609
609
|
structuredSections?: {
|
|
610
|
-
title: string;
|
|
611
610
|
key: string;
|
|
611
|
+
title: string;
|
|
612
612
|
facts: string[];
|
|
613
613
|
}[] | null | undefined;
|
|
614
614
|
promptedByQuestion?: string | null | undefined;
|
|
@@ -665,8 +665,8 @@ declare const ProactiveExtractionResultSchema: z.ZodObject<{
|
|
|
665
665
|
name: string;
|
|
666
666
|
facts: string[];
|
|
667
667
|
structuredSections?: {
|
|
668
|
-
title: string;
|
|
669
668
|
key: string;
|
|
669
|
+
title: string;
|
|
670
670
|
facts: string[];
|
|
671
671
|
}[] | null | undefined;
|
|
672
672
|
promptedByQuestion?: string | null | undefined;
|
|
@@ -714,8 +714,8 @@ declare const ProactiveExtractionResultSchema: z.ZodObject<{
|
|
|
714
714
|
name: string;
|
|
715
715
|
facts: string[];
|
|
716
716
|
structuredSections?: {
|
|
717
|
-
title: string;
|
|
718
717
|
key: string;
|
|
718
|
+
title: string;
|
|
719
719
|
facts: string[];
|
|
720
720
|
}[] | null | undefined;
|
|
721
721
|
promptedByQuestion?: string | null | undefined;
|
|
@@ -952,12 +952,12 @@ declare const ExtractionResultSchema: z.ZodObject<{
|
|
|
952
952
|
title: z.ZodString;
|
|
953
953
|
facts: z.ZodArray<z.ZodString, "many">;
|
|
954
954
|
}, "strip", z.ZodTypeAny, {
|
|
955
|
-
title: string;
|
|
956
955
|
key: string;
|
|
956
|
+
title: string;
|
|
957
957
|
facts: string[];
|
|
958
958
|
}, {
|
|
959
|
-
title: string;
|
|
960
959
|
key: string;
|
|
960
|
+
title: string;
|
|
961
961
|
facts: string[];
|
|
962
962
|
}>, "many">>>;
|
|
963
963
|
}, "strip", z.ZodTypeAny, {
|
|
@@ -965,8 +965,8 @@ declare const ExtractionResultSchema: z.ZodObject<{
|
|
|
965
965
|
name: string;
|
|
966
966
|
facts: string[];
|
|
967
967
|
structuredSections?: {
|
|
968
|
-
title: string;
|
|
969
968
|
key: string;
|
|
969
|
+
title: string;
|
|
970
970
|
facts: string[];
|
|
971
971
|
}[] | null | undefined;
|
|
972
972
|
promptedByQuestion?: string | null | undefined;
|
|
@@ -975,8 +975,8 @@ declare const ExtractionResultSchema: z.ZodObject<{
|
|
|
975
975
|
name: string;
|
|
976
976
|
facts: string[];
|
|
977
977
|
structuredSections?: {
|
|
978
|
-
title: string;
|
|
979
978
|
key: string;
|
|
979
|
+
title: string;
|
|
980
980
|
facts: string[];
|
|
981
981
|
}[] | null | undefined;
|
|
982
982
|
promptedByQuestion?: string | null | undefined;
|
|
@@ -1047,8 +1047,8 @@ declare const ExtractionResultSchema: z.ZodObject<{
|
|
|
1047
1047
|
name: string;
|
|
1048
1048
|
facts: string[];
|
|
1049
1049
|
structuredSections?: {
|
|
1050
|
-
title: string;
|
|
1051
1050
|
key: string;
|
|
1051
|
+
title: string;
|
|
1052
1052
|
facts: string[];
|
|
1053
1053
|
}[] | null | undefined;
|
|
1054
1054
|
promptedByQuestion?: string | null | undefined;
|
|
@@ -1102,8 +1102,8 @@ declare const ExtractionResultSchema: z.ZodObject<{
|
|
|
1102
1102
|
name: string;
|
|
1103
1103
|
facts: string[];
|
|
1104
1104
|
structuredSections?: {
|
|
1105
|
-
title: string;
|
|
1106
1105
|
key: string;
|
|
1106
|
+
title: string;
|
|
1107
1107
|
facts: string[];
|
|
1108
1108
|
}[] | null | undefined;
|
|
1109
1109
|
promptedByQuestion?: string | null | undefined;
|
|
@@ -1172,12 +1172,12 @@ declare const ConsolidationResultSchema: z.ZodObject<{
|
|
|
1172
1172
|
title: z.ZodString;
|
|
1173
1173
|
facts: z.ZodArray<z.ZodString, "many">;
|
|
1174
1174
|
}, "strip", z.ZodTypeAny, {
|
|
1175
|
-
title: string;
|
|
1176
1175
|
key: string;
|
|
1176
|
+
title: string;
|
|
1177
1177
|
facts: string[];
|
|
1178
1178
|
}, {
|
|
1179
|
-
title: string;
|
|
1180
1179
|
key: string;
|
|
1180
|
+
title: string;
|
|
1181
1181
|
facts: string[];
|
|
1182
1182
|
}>, "many">>>;
|
|
1183
1183
|
}, "strip", z.ZodTypeAny, {
|
|
@@ -1185,8 +1185,8 @@ declare const ConsolidationResultSchema: z.ZodObject<{
|
|
|
1185
1185
|
name: string;
|
|
1186
1186
|
facts: string[];
|
|
1187
1187
|
structuredSections?: {
|
|
1188
|
-
title: string;
|
|
1189
1188
|
key: string;
|
|
1189
|
+
title: string;
|
|
1190
1190
|
facts: string[];
|
|
1191
1191
|
}[] | null | undefined;
|
|
1192
1192
|
promptedByQuestion?: string | null | undefined;
|
|
@@ -1195,8 +1195,8 @@ declare const ConsolidationResultSchema: z.ZodObject<{
|
|
|
1195
1195
|
name: string;
|
|
1196
1196
|
facts: string[];
|
|
1197
1197
|
structuredSections?: {
|
|
1198
|
-
title: string;
|
|
1199
1198
|
key: string;
|
|
1199
|
+
title: string;
|
|
1200
1200
|
facts: string[];
|
|
1201
1201
|
}[] | null | undefined;
|
|
1202
1202
|
promptedByQuestion?: string | null | undefined;
|
|
@@ -1215,8 +1215,8 @@ declare const ConsolidationResultSchema: z.ZodObject<{
|
|
|
1215
1215
|
name: string;
|
|
1216
1216
|
facts: string[];
|
|
1217
1217
|
structuredSections?: {
|
|
1218
|
-
title: string;
|
|
1219
1218
|
key: string;
|
|
1219
|
+
title: string;
|
|
1220
1220
|
facts: string[];
|
|
1221
1221
|
}[] | null | undefined;
|
|
1222
1222
|
promptedByQuestion?: string | null | undefined;
|
|
@@ -1235,8 +1235,8 @@ declare const ConsolidationResultSchema: z.ZodObject<{
|
|
|
1235
1235
|
name: string;
|
|
1236
1236
|
facts: string[];
|
|
1237
1237
|
structuredSections?: {
|
|
1238
|
-
title: string;
|
|
1239
1238
|
key: string;
|
|
1239
|
+
title: string;
|
|
1240
1240
|
facts: string[];
|
|
1241
1241
|
}[] | null | undefined;
|
|
1242
1242
|
promptedByQuestion?: string | null | undefined;
|
package/dist/transfer/types.d.ts
CHANGED
|
@@ -313,13 +313,13 @@ declare const CapsuleBlockSchema: z.ZodObject<{
|
|
|
313
313
|
peerProfiles: boolean;
|
|
314
314
|
}>;
|
|
315
315
|
}, "strip", z.ZodTypeAny, {
|
|
316
|
-
schemaVersion: string;
|
|
317
316
|
includes: {
|
|
318
317
|
procedural: boolean;
|
|
319
318
|
taxonomy: boolean;
|
|
320
319
|
identityAnchors: boolean;
|
|
321
320
|
peerProfiles: boolean;
|
|
322
321
|
};
|
|
322
|
+
schemaVersion: string;
|
|
323
323
|
id: string;
|
|
324
324
|
description: string;
|
|
325
325
|
version: string;
|
|
@@ -334,13 +334,13 @@ declare const CapsuleBlockSchema: z.ZodObject<{
|
|
|
334
334
|
directAnswerEnabled: boolean;
|
|
335
335
|
};
|
|
336
336
|
}, {
|
|
337
|
-
schemaVersion: string;
|
|
338
337
|
includes: {
|
|
339
338
|
procedural: boolean;
|
|
340
339
|
taxonomy: boolean;
|
|
341
340
|
identityAnchors: boolean;
|
|
342
341
|
peerProfiles: boolean;
|
|
343
342
|
};
|
|
343
|
+
schemaVersion: string;
|
|
344
344
|
id: string;
|
|
345
345
|
description: string;
|
|
346
346
|
version: string;
|
|
@@ -464,13 +464,13 @@ declare const ExportManifestV2Schema: z.ZodObject<{
|
|
|
464
464
|
peerProfiles: boolean;
|
|
465
465
|
}>;
|
|
466
466
|
}, "strip", z.ZodTypeAny, {
|
|
467
|
-
schemaVersion: string;
|
|
468
467
|
includes: {
|
|
469
468
|
procedural: boolean;
|
|
470
469
|
taxonomy: boolean;
|
|
471
470
|
identityAnchors: boolean;
|
|
472
471
|
peerProfiles: boolean;
|
|
473
472
|
};
|
|
473
|
+
schemaVersion: string;
|
|
474
474
|
id: string;
|
|
475
475
|
description: string;
|
|
476
476
|
version: string;
|
|
@@ -485,13 +485,13 @@ declare const ExportManifestV2Schema: z.ZodObject<{
|
|
|
485
485
|
directAnswerEnabled: boolean;
|
|
486
486
|
};
|
|
487
487
|
}, {
|
|
488
|
-
schemaVersion: string;
|
|
489
488
|
includes: {
|
|
490
489
|
procedural: boolean;
|
|
491
490
|
taxonomy: boolean;
|
|
492
491
|
identityAnchors: boolean;
|
|
493
492
|
peerProfiles: boolean;
|
|
494
493
|
};
|
|
494
|
+
schemaVersion: string;
|
|
495
495
|
id: string;
|
|
496
496
|
description: string;
|
|
497
497
|
version: string;
|
|
@@ -518,13 +518,13 @@ declare const ExportManifestV2Schema: z.ZodObject<{
|
|
|
518
518
|
pluginVersion: string;
|
|
519
519
|
includesTranscripts: boolean;
|
|
520
520
|
capsule: {
|
|
521
|
-
schemaVersion: string;
|
|
522
521
|
includes: {
|
|
523
522
|
procedural: boolean;
|
|
524
523
|
taxonomy: boolean;
|
|
525
524
|
identityAnchors: boolean;
|
|
526
525
|
peerProfiles: boolean;
|
|
527
526
|
};
|
|
527
|
+
schemaVersion: string;
|
|
528
528
|
id: string;
|
|
529
529
|
description: string;
|
|
530
530
|
version: string;
|
|
@@ -551,13 +551,13 @@ declare const ExportManifestV2Schema: z.ZodObject<{
|
|
|
551
551
|
pluginVersion: string;
|
|
552
552
|
includesTranscripts: boolean;
|
|
553
553
|
capsule: {
|
|
554
|
-
schemaVersion: string;
|
|
555
554
|
includes: {
|
|
556
555
|
procedural: boolean;
|
|
557
556
|
taxonomy: boolean;
|
|
558
557
|
identityAnchors: boolean;
|
|
559
558
|
peerProfiles: boolean;
|
|
560
559
|
};
|
|
560
|
+
schemaVersion: string;
|
|
561
561
|
id: string;
|
|
562
562
|
description: string;
|
|
563
563
|
version: string;
|
|
@@ -683,13 +683,13 @@ declare const ExportBundleV2Schema: z.ZodObject<{
|
|
|
683
683
|
peerProfiles: boolean;
|
|
684
684
|
}>;
|
|
685
685
|
}, "strip", z.ZodTypeAny, {
|
|
686
|
-
schemaVersion: string;
|
|
687
686
|
includes: {
|
|
688
687
|
procedural: boolean;
|
|
689
688
|
taxonomy: boolean;
|
|
690
689
|
identityAnchors: boolean;
|
|
691
690
|
peerProfiles: boolean;
|
|
692
691
|
};
|
|
692
|
+
schemaVersion: string;
|
|
693
693
|
id: string;
|
|
694
694
|
description: string;
|
|
695
695
|
version: string;
|
|
@@ -704,13 +704,13 @@ declare const ExportBundleV2Schema: z.ZodObject<{
|
|
|
704
704
|
directAnswerEnabled: boolean;
|
|
705
705
|
};
|
|
706
706
|
}, {
|
|
707
|
-
schemaVersion: string;
|
|
708
707
|
includes: {
|
|
709
708
|
procedural: boolean;
|
|
710
709
|
taxonomy: boolean;
|
|
711
710
|
identityAnchors: boolean;
|
|
712
711
|
peerProfiles: boolean;
|
|
713
712
|
};
|
|
713
|
+
schemaVersion: string;
|
|
714
714
|
id: string;
|
|
715
715
|
description: string;
|
|
716
716
|
version: string;
|
|
@@ -737,13 +737,13 @@ declare const ExportBundleV2Schema: z.ZodObject<{
|
|
|
737
737
|
pluginVersion: string;
|
|
738
738
|
includesTranscripts: boolean;
|
|
739
739
|
capsule: {
|
|
740
|
-
schemaVersion: string;
|
|
741
740
|
includes: {
|
|
742
741
|
procedural: boolean;
|
|
743
742
|
taxonomy: boolean;
|
|
744
743
|
identityAnchors: boolean;
|
|
745
744
|
peerProfiles: boolean;
|
|
746
745
|
};
|
|
746
|
+
schemaVersion: string;
|
|
747
747
|
id: string;
|
|
748
748
|
description: string;
|
|
749
749
|
version: string;
|
|
@@ -770,13 +770,13 @@ declare const ExportBundleV2Schema: z.ZodObject<{
|
|
|
770
770
|
pluginVersion: string;
|
|
771
771
|
includesTranscripts: boolean;
|
|
772
772
|
capsule: {
|
|
773
|
-
schemaVersion: string;
|
|
774
773
|
includes: {
|
|
775
774
|
procedural: boolean;
|
|
776
775
|
taxonomy: boolean;
|
|
777
776
|
identityAnchors: boolean;
|
|
778
777
|
peerProfiles: boolean;
|
|
779
778
|
};
|
|
779
|
+
schemaVersion: string;
|
|
780
780
|
id: string;
|
|
781
781
|
description: string;
|
|
782
782
|
version: string;
|
|
@@ -815,13 +815,13 @@ declare const ExportBundleV2Schema: z.ZodObject<{
|
|
|
815
815
|
pluginVersion: string;
|
|
816
816
|
includesTranscripts: boolean;
|
|
817
817
|
capsule: {
|
|
818
|
-
schemaVersion: string;
|
|
819
818
|
includes: {
|
|
820
819
|
procedural: boolean;
|
|
821
820
|
taxonomy: boolean;
|
|
822
821
|
identityAnchors: boolean;
|
|
823
822
|
peerProfiles: boolean;
|
|
824
823
|
};
|
|
824
|
+
schemaVersion: string;
|
|
825
825
|
id: string;
|
|
826
826
|
description: string;
|
|
827
827
|
version: string;
|
|
@@ -854,13 +854,13 @@ declare const ExportBundleV2Schema: z.ZodObject<{
|
|
|
854
854
|
pluginVersion: string;
|
|
855
855
|
includesTranscripts: boolean;
|
|
856
856
|
capsule: {
|
|
857
|
-
schemaVersion: string;
|
|
858
857
|
includes: {
|
|
859
858
|
procedural: boolean;
|
|
860
859
|
taxonomy: boolean;
|
|
861
860
|
identityAnchors: boolean;
|
|
862
861
|
peerProfiles: boolean;
|
|
863
862
|
};
|
|
863
|
+
schemaVersion: string;
|
|
864
864
|
id: string;
|
|
865
865
|
description: string;
|
|
866
866
|
version: string;
|
package/package.json
CHANGED
|
@@ -22,7 +22,7 @@
|
|
|
22
22
|
|
|
23
23
|
import path from "node:path";
|
|
24
24
|
import { homedir } from "node:os";
|
|
25
|
-
import { appendFile, chmod, mkdir, readFile, readdir } from "node:fs/promises";
|
|
25
|
+
import { appendFile, chmod, lstat, mkdir, readFile, readdir, realpath } from "node:fs/promises";
|
|
26
26
|
import { log } from "./logger.js";
|
|
27
27
|
import type { JudgeVerdictKind } from "./extraction-judge.js";
|
|
28
28
|
|
|
@@ -101,22 +101,21 @@ function dateStamp(iso: string): string {
|
|
|
101
101
|
return `${yyyy}-${mm}-${dd}`;
|
|
102
102
|
}
|
|
103
103
|
|
|
104
|
-
export function trainingFilePathFor(
|
|
105
|
-
directory: string,
|
|
106
|
-
iso: string,
|
|
107
|
-
): string {
|
|
104
|
+
export function trainingFilePathFor(directory: string, iso: string): string {
|
|
108
105
|
return path.join(directory, `${dateStamp(iso)}.jsonl`);
|
|
109
106
|
}
|
|
110
107
|
|
|
108
|
+
function isPathInsideDirectory(filePath: string, directory: string): boolean {
|
|
109
|
+
const relative = path.relative(directory, filePath);
|
|
110
|
+
return relative === "" || (!relative.startsWith("..") && !path.isAbsolute(relative));
|
|
111
|
+
}
|
|
112
|
+
|
|
111
113
|
/**
|
|
112
114
|
* Append a single training row. Fails open — write errors are logged at
|
|
113
115
|
* debug level and swallowed, same policy as the telemetry emitter.
|
|
114
116
|
* No-op when `options.enabled` is false.
|
|
115
117
|
*/
|
|
116
|
-
export async function recordJudgeTrainingPair(
|
|
117
|
-
row: JudgeTrainingPair,
|
|
118
|
-
options: JudgeTrainingOptions,
|
|
119
|
-
): Promise<void> {
|
|
118
|
+
export async function recordJudgeTrainingPair(row: JudgeTrainingPair, options: JudgeTrainingOptions): Promise<void> {
|
|
120
119
|
if (!options.enabled) return;
|
|
121
120
|
const dir = resolveTrainingDir(options);
|
|
122
121
|
const filePath = trainingFilePathFor(dir, row.ts);
|
|
@@ -129,7 +128,7 @@ export async function recordJudgeTrainingPair(
|
|
|
129
128
|
await chmod(filePath, 0o600);
|
|
130
129
|
} catch (err) {
|
|
131
130
|
log.debug(
|
|
132
|
-
`extraction-judge-training: append failed (non-fatal): ${err instanceof Error ? err.message : String(err)}
|
|
131
|
+
`extraction-judge-training: append failed (non-fatal): ${err instanceof Error ? err.message : String(err)}`
|
|
133
132
|
);
|
|
134
133
|
}
|
|
135
134
|
}
|
|
@@ -140,9 +139,23 @@ export async function recordJudgeTrainingPair(
|
|
|
140
139
|
* counted in the returned `malformed` tally.
|
|
141
140
|
*/
|
|
142
141
|
export async function readJudgeTrainingPairs(
|
|
143
|
-
options: Pick<JudgeTrainingOptions, "directory"
|
|
142
|
+
options: Pick<JudgeTrainingOptions, "directory">
|
|
144
143
|
): Promise<{ rows: JudgeTrainingPair[]; malformed: number }> {
|
|
145
144
|
const dir = resolveTrainingDir({ enabled: true, ...options });
|
|
145
|
+
try {
|
|
146
|
+
const dirStat = await lstat(dir);
|
|
147
|
+
if (dirStat.isSymbolicLink()) {
|
|
148
|
+
throw new Error("Judge training directory must not be a symlink");
|
|
149
|
+
}
|
|
150
|
+
if (!dirStat.isDirectory()) {
|
|
151
|
+
throw new Error("Judge training path must be a directory");
|
|
152
|
+
}
|
|
153
|
+
} catch (err) {
|
|
154
|
+
const code = (err as NodeJS.ErrnoException).code;
|
|
155
|
+
if (code === "ENOENT") return { rows: [], malformed: 0 };
|
|
156
|
+
throw err;
|
|
157
|
+
}
|
|
158
|
+
|
|
146
159
|
let entries: string[];
|
|
147
160
|
try {
|
|
148
161
|
entries = await readdir(dir);
|
|
@@ -154,11 +167,33 @@ export async function readJudgeTrainingPairs(
|
|
|
154
167
|
|
|
155
168
|
const rows: JudgeTrainingPair[] = [];
|
|
156
169
|
let malformed = 0;
|
|
170
|
+
const resolvedDir = await realpath(dir);
|
|
157
171
|
// Sort so reads are deterministic across platforms.
|
|
158
172
|
entries.sort();
|
|
159
173
|
for (const name of entries) {
|
|
160
174
|
if (!name.endsWith(".jsonl")) continue;
|
|
161
|
-
const
|
|
175
|
+
const filePath = path.join(dir, name);
|
|
176
|
+
let fileStat: Awaited<ReturnType<typeof lstat>>;
|
|
177
|
+
try {
|
|
178
|
+
fileStat = await lstat(filePath);
|
|
179
|
+
} catch (err) {
|
|
180
|
+
const code = (err as NodeJS.ErrnoException).code;
|
|
181
|
+
if (code === "ENOENT") continue;
|
|
182
|
+
throw err;
|
|
183
|
+
}
|
|
184
|
+
if (fileStat.isSymbolicLink() || !fileStat.isFile()) continue;
|
|
185
|
+
|
|
186
|
+
let resolvedFilePath: string;
|
|
187
|
+
try {
|
|
188
|
+
resolvedFilePath = await realpath(filePath);
|
|
189
|
+
} catch (err) {
|
|
190
|
+
const code = (err as NodeJS.ErrnoException).code;
|
|
191
|
+
if (code === "ENOENT") continue;
|
|
192
|
+
throw err;
|
|
193
|
+
}
|
|
194
|
+
if (!isPathInsideDirectory(resolvedFilePath, resolvedDir)) continue;
|
|
195
|
+
|
|
196
|
+
const raw = await readFile(resolvedFilePath, "utf-8");
|
|
162
197
|
for (const line of raw.split("\n")) {
|
|
163
198
|
if (!line.trim()) continue;
|
|
164
199
|
let parsed: unknown;
|
|
@@ -192,18 +227,11 @@ export function isValidTrainingPair(value: unknown): value is JudgeTrainingPair
|
|
|
192
227
|
if (typeof p.ts !== "string") return false;
|
|
193
228
|
if (typeof p.candidateText !== "string") return false;
|
|
194
229
|
if (typeof p.candidateCategory !== "string") return false;
|
|
195
|
-
if (
|
|
196
|
-
p.verdictKind !== "accept" &&
|
|
197
|
-
p.verdictKind !== "reject" &&
|
|
198
|
-
p.verdictKind !== "defer"
|
|
199
|
-
) {
|
|
230
|
+
if (p.verdictKind !== "accept" && p.verdictKind !== "reject" && p.verdictKind !== "defer") {
|
|
200
231
|
return false;
|
|
201
232
|
}
|
|
202
233
|
if (typeof p.reason !== "string") return false;
|
|
203
|
-
if (
|
|
204
|
-
p.candidateConfidence !== undefined &&
|
|
205
|
-
typeof p.candidateConfidence !== "number"
|
|
206
|
-
) {
|
|
234
|
+
if (p.candidateConfidence !== undefined && typeof p.candidateConfidence !== "number") {
|
|
207
235
|
return false;
|
|
208
236
|
}
|
|
209
237
|
if (p.priorDeferrals !== undefined && typeof p.priorDeferrals !== "number") {
|
|
@@ -1 +0,0 @@
|
|
|
1
|
-
{"version":3,"sources":["../src/extraction-judge-training.ts"],"sourcesContent":["/**\n * Extraction Judge Training Data Shim (issue #562, PR 4).\n *\n * Opt-in collector for `(candidate_text, verdict_kind, reason,\n * ground_truth_label?)` tuples. Rows are appended to JSONL files under\n * `~/.remnic/judge-training/<YYYY-MM-DD>.jsonl` so operators can ship the\n * data into a future GRPO training pipeline without exfiltrating live\n * memory content through the regular observation ledger.\n *\n * Gating:\n * - Off by default. Must be explicitly enabled via\n * `collectJudgeTrainingPairs: true` in plugin config.\n * - The ground-truth label is always optional — labels are added out-of-\n * band once reviewers disambiguate the candidate's fate.\n *\n * Privacy: the row carries only what the judge already sees — the\n * candidate text and its metadata. It does NOT carry session keys,\n * principal IDs, or any user identifiers. The file lives in the user's\n * home directory rather than the shared memory directory so it is never\n * committed, sync'd, or bundled into exports.\n */\n\nimport path from \"node:path\";\nimport { homedir } from \"node:os\";\nimport { appendFile, chmod, mkdir, readFile, readdir } from \"node:fs/promises\";\nimport { log } from \"./logger.js\";\nimport type { JudgeVerdictKind } from \"./extraction-judge.js\";\n\n/**\n * Persisted training row. Intentionally minimal: just the signal needed\n * to train a judge replacement policy. Schema version is tagged so future\n * readers can migrate older rows.\n */\nexport interface JudgeTrainingPair {\n version: 1;\n ts: string; // ISO-8601\n candidateText: string;\n candidateCategory: string;\n candidateConfidence?: number;\n verdictKind: JudgeVerdictKind;\n reason: string;\n /**\n * Number of prior deferrals when the verdict was resolved. `0` for the\n * first resolution; only set when known (defer pathway).\n */\n priorDeferrals?: number;\n /**\n * Optional human-applied ground-truth label. Added after the fact by a\n * reviewer / labelling script; not present on fresh rows.\n */\n groundTruthLabel?: JudgeVerdictKind;\n}\n\nexport interface JudgeTrainingOptions {\n enabled: boolean;\n /**\n * Override for the output directory. Defaults to\n * `~/.remnic/judge-training`. Tests pass a temp path here.\n */\n directory?: string;\n}\n\n/**\n * Expand a leading `~` / `~/` / `$HOME/` / `${HOME}/` to the process home\n * directory. Node's `fs` APIs do not expand `~` themselves (CLAUDE.md\n * gotcha 17), so every user-facing path input must be funnelled through\n * this helper before it reaches the filesystem.\n */\nfunction expandTilde(p: string): string {\n const home = homedir();\n if (p === \"~\" || p.startsWith(\"~/\") || p.startsWith(\"~\\\\\")) {\n return home + p.slice(1);\n }\n if (p === \"$HOME\" || p.startsWith(\"$HOME/\") || p.startsWith(\"$HOME\\\\\")) {\n return home + p.slice(5);\n }\n if (p === \"${HOME}\" || p.startsWith(\"${HOME}/\") || p.startsWith(\"${HOME}\\\\\")) {\n return home + p.slice(7);\n }\n return p;\n}\n\nexport function resolveTrainingDir(options: JudgeTrainingOptions): string {\n if (options.directory && options.directory.length > 0) {\n // Expand `~` / `$HOME` in the override so operators can write the\n // config as the user sees it (CLAUDE.md gotcha 17).\n return expandTilde(options.directory);\n }\n return path.join(homedir(), \".remnic\", \"judge-training\");\n}\n\nfunction dateStamp(iso: string): string {\n // `YYYY-MM-DD` from an ISO-8601 string. Falls back to today on a parse\n // failure rather than throwing — the caller already wrote a row and the\n // timestamp is best-effort.\n const ms = Date.parse(iso);\n const d = Number.isFinite(ms) ? new Date(ms) : new Date();\n const yyyy = d.getUTCFullYear().toString().padStart(4, \"0\");\n const mm = (d.getUTCMonth() + 1).toString().padStart(2, \"0\");\n const dd = d.getUTCDate().toString().padStart(2, \"0\");\n return `${yyyy}-${mm}-${dd}`;\n}\n\nexport function trainingFilePathFor(\n directory: string,\n iso: string,\n): string {\n return path.join(directory, `${dateStamp(iso)}.jsonl`);\n}\n\n/**\n * Append a single training row. Fails open — write errors are logged at\n * debug level and swallowed, same policy as the telemetry emitter.\n * No-op when `options.enabled` is false.\n */\nexport async function recordJudgeTrainingPair(\n row: JudgeTrainingPair,\n options: JudgeTrainingOptions,\n): Promise<void> {\n if (!options.enabled) return;\n const dir = resolveTrainingDir(options);\n const filePath = trainingFilePathFor(dir, row.ts);\n try {\n await mkdir(dir, { recursive: true, mode: 0o700 });\n await appendFile(filePath, `${JSON.stringify(row)}\\n`, {\n encoding: \"utf-8\",\n mode: 0o600,\n });\n await chmod(filePath, 0o600);\n } catch (err) {\n log.debug(\n `extraction-judge-training: append failed (non-fatal): ${err instanceof Error ? err.message : String(err)}`,\n );\n }\n}\n\n/**\n * Read all training rows from the configured directory. Returns an empty\n * array when the directory is missing. Malformed lines are skipped and\n * counted in the returned `malformed` tally.\n */\nexport async function readJudgeTrainingPairs(\n options: Pick<JudgeTrainingOptions, \"directory\">,\n): Promise<{ rows: JudgeTrainingPair[]; malformed: number }> {\n const dir = resolveTrainingDir({ enabled: true, ...options });\n let entries: string[];\n try {\n entries = await readdir(dir);\n } catch (err) {\n const code = (err as NodeJS.ErrnoException).code;\n if (code === \"ENOENT\") return { rows: [], malformed: 0 };\n throw err;\n }\n\n const rows: JudgeTrainingPair[] = [];\n let malformed = 0;\n // Sort so reads are deterministic across platforms.\n entries.sort();\n for (const name of entries) {\n if (!name.endsWith(\".jsonl\")) continue;\n const raw = await readFile(path.join(dir, name), \"utf-8\");\n for (const line of raw.split(\"\\n\")) {\n if (!line.trim()) continue;\n let parsed: unknown;\n try {\n parsed = JSON.parse(line);\n } catch {\n malformed += 1;\n continue;\n }\n if (!isValidTrainingPair(parsed)) {\n malformed += 1;\n continue;\n }\n rows.push(parsed);\n }\n }\n return { rows, malformed };\n}\n\n/**\n * Structural validator matching the persisted schema. Forward-compat: an\n * unknown `verdictKind` string is treated as malformed (strict training\n * signal — we do not want to admit unlabelled gibberish into a trainer).\n */\nexport function isValidTrainingPair(value: unknown): value is JudgeTrainingPair {\n if (typeof value !== \"object\" || value === null || Array.isArray(value)) {\n return false;\n }\n const p = value as Record<string, unknown>;\n if (p.version !== 1) return false;\n if (typeof p.ts !== \"string\") return false;\n if (typeof p.candidateText !== \"string\") return false;\n if (typeof p.candidateCategory !== \"string\") return false;\n if (\n p.verdictKind !== \"accept\" &&\n p.verdictKind !== \"reject\" &&\n p.verdictKind !== \"defer\"\n ) {\n return false;\n }\n if (typeof p.reason !== \"string\") return false;\n if (\n p.candidateConfidence !== undefined &&\n typeof p.candidateConfidence !== \"number\"\n ) {\n return false;\n }\n if (p.priorDeferrals !== undefined && typeof p.priorDeferrals !== \"number\") {\n return false;\n }\n if (\n p.groundTruthLabel !== undefined &&\n p.groundTruthLabel !== \"accept\" &&\n p.groundTruthLabel !== \"reject\" &&\n p.groundTruthLabel !== \"defer\"\n ) {\n return false;\n }\n return true;\n}\n"],"mappings":";;;;;AAsBA,OAAO,UAAU;AACjB,SAAS,eAAe;AACxB,SAAS,YAAY,OAAO,OAAO,UAAU,eAAe;AA4C5D,SAAS,YAAY,GAAmB;AACtC,QAAM,OAAO,QAAQ;AACrB,MAAI,MAAM,OAAO,EAAE,WAAW,IAAI,KAAK,EAAE,WAAW,KAAK,GAAG;AAC1D,WAAO,OAAO,EAAE,MAAM,CAAC;AAAA,EACzB;AACA,MAAI,MAAM,WAAW,EAAE,WAAW,QAAQ,KAAK,EAAE,WAAW,SAAS,GAAG;AACtE,WAAO,OAAO,EAAE,MAAM,CAAC;AAAA,EACzB;AACA,MAAI,MAAM,aAAa,EAAE,WAAW,UAAU,KAAK,EAAE,WAAW,WAAW,GAAG;AAC5E,WAAO,OAAO,EAAE,MAAM,CAAC;AAAA,EACzB;AACA,SAAO;AACT;AAEO,SAAS,mBAAmB,SAAuC;AACxE,MAAI,QAAQ,aAAa,QAAQ,UAAU,SAAS,GAAG;AAGrD,WAAO,YAAY,QAAQ,SAAS;AAAA,EACtC;AACA,SAAO,KAAK,KAAK,QAAQ,GAAG,WAAW,gBAAgB;AACzD;AAEA,SAAS,UAAU,KAAqB;AAItC,QAAM,KAAK,KAAK,MAAM,GAAG;AACzB,QAAM,IAAI,OAAO,SAAS,EAAE,IAAI,IAAI,KAAK,EAAE,IAAI,oBAAI,KAAK;AACxD,QAAM,OAAO,EAAE,eAAe,EAAE,SAAS,EAAE,SAAS,GAAG,GAAG;AAC1D,QAAM,MAAM,EAAE,YAAY,IAAI,GAAG,SAAS,EAAE,SAAS,GAAG,GAAG;AAC3D,QAAM,KAAK,EAAE,WAAW,EAAE,SAAS,EAAE,SAAS,GAAG,GAAG;AACpD,SAAO,GAAG,IAAI,IAAI,EAAE,IAAI,EAAE;AAC5B;AAEO,SAAS,oBACd,WACA,KACQ;AACR,SAAO,KAAK,KAAK,WAAW,GAAG,UAAU,GAAG,CAAC,QAAQ;AACvD;AAOA,eAAsB,wBACpB,KACA,SACe;AACf,MAAI,CAAC,QAAQ,QAAS;AACtB,QAAM,MAAM,mBAAmB,OAAO;AACtC,QAAM,WAAW,oBAAoB,KAAK,IAAI,EAAE;AAChD,MAAI;AACF,UAAM,MAAM,KAAK,EAAE,WAAW,MAAM,MAAM,IAAM,CAAC;AACjD,UAAM,WAAW,UAAU,GAAG,KAAK,UAAU,GAAG,CAAC;AAAA,GAAM;AAAA,MACrD,UAAU;AAAA,MACV,MAAM;AAAA,IACR,CAAC;AACD,UAAM,MAAM,UAAU,GAAK;AAAA,EAC7B,SAAS,KAAK;AACZ,QAAI;AAAA,MACF,yDAAyD,eAAe,QAAQ,IAAI,UAAU,OAAO,GAAG,CAAC;AAAA,IAC3G;AAAA,EACF;AACF;AAOA,eAAsB,uBACpB,SAC2D;AAC3D,QAAM,MAAM,mBAAmB,EAAE,SAAS,MAAM,GAAG,QAAQ,CAAC;AAC5D,MAAI;AACJ,MAAI;AACF,cAAU,MAAM,QAAQ,GAAG;AAAA,EAC7B,SAAS,KAAK;AACZ,UAAM,OAAQ,IAA8B;AAC5C,QAAI,SAAS,SAAU,QAAO,EAAE,MAAM,CAAC,GAAG,WAAW,EAAE;AACvD,UAAM;AAAA,EACR;AAEA,QAAM,OAA4B,CAAC;AACnC,MAAI,YAAY;AAEhB,UAAQ,KAAK;AACb,aAAW,QAAQ,SAAS;AAC1B,QAAI,CAAC,KAAK,SAAS,QAAQ,EAAG;AAC9B,UAAM,MAAM,MAAM,SAAS,KAAK,KAAK,KAAK,IAAI,GAAG,OAAO;AACxD,eAAW,QAAQ,IAAI,MAAM,IAAI,GAAG;AAClC,UAAI,CAAC,KAAK,KAAK,EAAG;AAClB,UAAI;AACJ,UAAI;AACF,iBAAS,KAAK,MAAM,IAAI;AAAA,MAC1B,QAAQ;AACN,qBAAa;AACb;AAAA,MACF;AACA,UAAI,CAAC,oBAAoB,MAAM,GAAG;AAChC,qBAAa;AACb;AAAA,MACF;AACA,WAAK,KAAK,MAAM;AAAA,IAClB;AAAA,EACF;AACA,SAAO,EAAE,MAAM,UAAU;AAC3B;AAOO,SAAS,oBAAoB,OAA4C;AAC9E,MAAI,OAAO,UAAU,YAAY,UAAU,QAAQ,MAAM,QAAQ,KAAK,GAAG;AACvE,WAAO;AAAA,EACT;AACA,QAAM,IAAI;AACV,MAAI,EAAE,YAAY,EAAG,QAAO;AAC5B,MAAI,OAAO,EAAE,OAAO,SAAU,QAAO;AACrC,MAAI,OAAO,EAAE,kBAAkB,SAAU,QAAO;AAChD,MAAI,OAAO,EAAE,sBAAsB,SAAU,QAAO;AACpD,MACE,EAAE,gBAAgB,YAClB,EAAE,gBAAgB,YAClB,EAAE,gBAAgB,SAClB;AACA,WAAO;AAAA,EACT;AACA,MAAI,OAAO,EAAE,WAAW,SAAU,QAAO;AACzC,MACE,EAAE,wBAAwB,UAC1B,OAAO,EAAE,wBAAwB,UACjC;AACA,WAAO;AAAA,EACT;AACA,MAAI,EAAE,mBAAmB,UAAa,OAAO,EAAE,mBAAmB,UAAU;AAC1E,WAAO;AAAA,EACT;AACA,MACE,EAAE,qBAAqB,UACvB,EAAE,qBAAqB,YACvB,EAAE,qBAAqB,YACvB,EAAE,qBAAqB,SACvB;AACA,WAAO;AAAA,EACT;AACA,SAAO;AACT;","names":[]}
|
|
File without changes
|