context-mode 1.0.157 → 1.0.158

package/.claude-plugin/marketplace.json

@@ -6,14 +6,14 @@
   },
   "metadata": {
     "description": "Claude Code plugins by Mert Koseoğlu",
-    "version": "1.0.157"
+    "version": "1.0.158"
   },
   "plugins": [
     {
       "name": "context-mode",
       "source": "./",
       "description": "Claude Code MCP plugin that saves 98% of your context window. Sandboxed code execution in 11 languages, FTS5 knowledge base with BM25 ranking, and intent-driven search.",
-      "version": "1.0.157",
+      "version": "1.0.158",
       "author": {
         "name": "Mert Koseoğlu"
       },

package/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "context-mode",
-  "version": "1.0.157",
+  "version": "1.0.158",
   "description": "MCP server that saves 98% of your context window with session continuity. Sandboxed code execution in 11 languages, FTS5 knowledge base with BM25 ranking, and automatic state restore across compactions.",
   "author": {
     "name": "Mert Koseoğlu",

package/.codex-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "context-mode",
-  "version": "1.0.157",
+  "version": "1.0.158",
   "description": "MCP server that saves 98% of your context window with session continuity. Sandboxed code execution in 11 languages, FTS5 knowledge base with BM25 ranking, and automatic state restore across compactions.",
   "author": {
     "name": "Mert Koseoğlu",

package/.openclaw-plugin/openclaw.plugin.json

@@ -3,7 +3,7 @@
   "name": "Context Mode",
   "kind": "tool",
   "description": "OpenClaw plugin that saves 98% of your context window. Sandboxed code execution in 11 languages, FTS5 knowledge base with BM25 ranking, and intent-driven search.",
-  "version": "1.0.157",
+  "version": "1.0.158",
   "sandbox": {
     "mode": "permissive",
     "filesystem_access": "full",

package/.openclaw-plugin/package.json

@@ -1,6 +1,6 @@
 {
   "name": "context-mode",
-  "version": "1.0.157",
+  "version": "1.0.158",
   "description": "OpenClaw plugin that saves 98% of your context window. Sandboxed code execution in 11 languages, FTS5 knowledge base with BM25 ranking, and intent-driven search.",
   "author": {
     "name": "Mert Koseoğlu",

package/build/session/db.d.ts

@@ -164,6 +164,28 @@ export interface SessionMeta {
     event_count: number;
     compact_count: number;
 }
+/**
+ * Session rollup snapshot (seed-parity aggregate).
+ *
+ * 12 fields that mirror the platform's `session_summary` + `session_metadata`
+ * stamps from src/routes/seed.ts. Each outgoing canonical event carries
+ * this snapshot computed at the moment of forward so the analytics engine
+ * can run its SUM/AVG/MAX rollups across per-event rows.
+ */
+export interface SessionRollup {
+    tool_calls: number;
+    errors: number;
+    unique_tools: number;
+    unique_files: number;
+    max_file_edits: number;
+    has_commit: 0 | 1;
+    edit_test_cycles: number;
+    duration_min: number;
+    compact_count: number;
+    sources_indexed: number;
+    total_chunks: number;
+    search_queries: number;
+}
 /** Resume snapshot row from the session_resume table. */
 export interface ResumeRow {
     snapshot: string;
@@ -340,6 +362,19 @@ export declare class SessionDB extends SQLiteBase {
      * Get session statistics/metadata.
      */
     getSessionStats(sessionId: string): SessionMeta | null;
+    /**
+     * Session rollup snapshot — 12 aggregate fields the analytics platform
+     * stamps onto every outgoing event row (seed.ts shape parity).
+     *
+     * Called from session-loaders BEFORE `maybeForward`; the snapshot is
+     * computed against the LOCAL SessionDB and threaded into the canonical
+     * event so the platform-side Zod schema receives the rich shape without
+     * the bridge ever hand-mapping fields (PRD §5.4 ABI passthrough).
+     *
+     * Returns zeroed defaults for unknown sessions — callers MUST tolerate
+     * a snapshot from an empty session (first event into a fresh DB).
+     */
+    getSessionRollup(sessionId: string): SessionRollup;
     /**
      * Increment the compact_count for a session (tracks snapshot rebuilds).
      */

package/build/session/db.js

@@ -477,6 +477,8 @@ const S = {
     updateMetaLastEvent: "updateMetaLastEvent",
     ensureSession: "ensureSession",
     getSessionStats: "getSessionStats",
+    getSessionRollup: "getSessionRollup",
+    getMaxFileEdits: "getMaxFileEdits",
     incrementCompactCount: "incrementCompactCount",
     upsertResume: "upsertResume",
     getResume: "getResume",
@@ -716,6 +718,35 @@ export class SessionDB extends SQLiteBase {
         p(S.ensureSession, `INSERT OR IGNORE INTO session_meta (session_id, project_dir) VALUES (?, ?)`);
         p(S.getSessionStats, `SELECT session_id, project_dir, started_at, last_event_at, event_count, compact_count
        FROM session_meta WHERE session_id = ?`);
+        // ── Session rollup (seed-parity aggregator) ────────────────────────
+        // Single query producing 9 of the 12 platform-side session_summary +
+        // session_metadata fields. Computed against the local SessionDB
+        // session_events table at forward time so every outgoing canonical
+        // event carries a session-wide snapshot at that moment — matches the
+        // seed.ts shape where each event row has tool_calls/errors/etc. stamped.
+        // max_file_edits and edit_test_cycles need separate GROUP BY queries
+        // (below). compact_count is read from session_meta (already in getSessionStats).
+        p(S.getSessionRollup, `SELECT
+         COUNT(*) AS tool_calls,
+         COALESCE(SUM(CASE WHEN category = 'error' THEN 1 ELSE 0 END), 0) AS errors,
+         COUNT(DISTINCT type) AS unique_tools,
+         COUNT(DISTINCT CASE WHEN category = 'file' THEN data END) AS unique_files,
+         CASE WHEN SUM(CASE WHEN category = 'git' THEN 1 ELSE 0 END) > 0 THEN 1 ELSE 0 END AS has_commit,
+         CAST(COALESCE((MAX(strftime('%s', created_at)) - MIN(strftime('%s', created_at))) / 60.0, 0) AS INTEGER) AS duration_min,
+         COALESCE(SUM(CASE WHEN type = 'external_ref' THEN 1 ELSE 0 END), 0) AS sources_indexed,
+         CAST(COALESCE(SUM(bytes_avoided) / 1024.0, 0) AS INTEGER) AS total_chunks,
+         COALESCE(SUM(CASE WHEN type IN ('file_search', 'file_glob') THEN 1 ELSE 0 END), 0) AS search_queries
+       FROM session_events
+       WHERE session_id = ?`);
+        // max_file_edits: max edits on any single file path in the session.
+        // Two-level aggregation — GROUP BY data first, then MAX of those counts.
+        p(S.getMaxFileEdits, `SELECT COALESCE(MAX(c), 0) AS max_file_edits
+       FROM (
+         SELECT COUNT(*) AS c
+         FROM session_events
+         WHERE session_id = ? AND category = 'file' AND type IN ('file_edit', 'file_write')
+         GROUP BY data
+       )`);
         p(S.incrementCompactCount, `UPDATE session_meta SET compact_count = compact_count + 1 WHERE session_id = ?`);
         // ── Resume ──
         p(S.upsertResume, `INSERT INTO session_resume (session_id, snapshot, event_count)
@@ -1017,6 +1048,46 @@ export class SessionDB extends SQLiteBase {
         const row = this.stmt(S.getSessionStats).get(sessionId);
         return row ?? null;
     }
+    /**
+     * Session rollup snapshot — 12 aggregate fields the analytics platform
+     * stamps onto every outgoing event row (seed.ts shape parity).
+     *
+     * Called from session-loaders BEFORE `maybeForward`; the snapshot is
+     * computed against the LOCAL SessionDB and threaded into the canonical
+     * event so the platform-side Zod schema receives the rich shape without
+     * the bridge ever hand-mapping fields (PRD §5.4 ABI passthrough).
+     *
+     * Returns zeroed defaults for unknown sessions — callers MUST tolerate
+     * a snapshot from an empty session (first event into a fresh DB).
+     */
+    getSessionRollup(sessionId) {
+        const main = this.stmt(S.getSessionRollup).get(sessionId);
+        const maxRow = this.stmt(S.getMaxFileEdits).get(sessionId);
+        const meta = this.getSessionStats(sessionId);
+        // edit_test_cycles: heuristic — min(file edits, errors) approximates
+        // the number of edit-then-test attempts in a session. Exact pattern
+        // detection (consecutive file_edit followed by error_tool) would need
+        // a windowed query; this scalar pair under-counts but never overshoots.
+        const fileEdits = (main?.tool_calls ?? 0) > 0
+            ? (main?.unique_files ?? 0)
+            : 0;
+        const errors = main?.errors ?? 0;
+        const editTestCycles = Math.min(fileEdits, errors);
+        return {
+            tool_calls: main?.tool_calls ?? 0,
+            errors: main?.errors ?? 0,
+            unique_tools: main?.unique_tools ?? 0,
+            unique_files: main?.unique_files ?? 0,
+            max_file_edits: maxRow?.max_file_edits ?? 0,
+            has_commit: main?.has_commit ?? 0,
+            edit_test_cycles: editTestCycles,
+            duration_min: main?.duration_min ?? 0,
+            compact_count: meta?.compact_count ?? 0,
+            sources_indexed: main?.sources_indexed ?? 0,
+            total_chunks: main?.total_chunks ?? 0,
+            search_queries: main?.search_queries ?? 0,
+        };
+    }
     /**
      * Increment the compact_count for a session (tracks snapshot rebuilds).
      */

package/build/session/error-classifier.d.ts

@@ -0,0 +1,87 @@
+/**
+ * error-classifier — PRD-context-as-a-service §5 ABI parity
+ *
+ * Pure classifiers that derive per-event metadata from PostToolUse hook
+ * stdin, mirroring the seed shape produced by context-mode-platform's
+ * `seed.ts` so the bridge can ship full seed-shape parity.
+ *
+ * The dashboard reads three derived columns that are populated by the
+ * platform seeder but, per the OSS handoff (ANOMALY #3), are NEVER READ
+ * by the engine when ingesting live events:
+ *
+ *   • error_category   — one of 10 fixed buckets
+ *   • error_tool       — tool name that produced the error
+ *   • command_type     — git / test / build / lint / install / format / run / other
+ *   • command_tool     — first token of the command (npm, git, pytest, …)
+ *   • duration_bucket  — fast / medium / slow / timeout
+ *
+ * We populate them anyway, with sane defaults derived from message text
+ * and tool name, so the dashboard renders symmetrically with seed.
+ *
+ * Hard constraints:
+ *   1. Pure functions — no I/O, no globals, no module state.
+ *   2. `error_category` MUST match seed's `pickErrorClassification`
+ *      output exactly. The literal union below is the ABI.
+ *   3. No external dependencies; TypeScript stdlib only.
+ *   4. Robust to null / undefined / empty / malformed input.
+ */
+export type ErrorCategory = "file_not_found" | "command_not_found" | "edit_match_failed" | "test_failed" | "syntax_error" | "runtime_error" | "permission_denied" | "git_conflict" | "timeout" | "unknown";
+export type CommandType = "test" | "build" | "lint" | "git" | "install" | "format" | "run" | "other";
+export type DurationBucket = "fast" | "medium" | "slow" | "timeout";
+export interface ErrorClassification {
+    error_category: ErrorCategory;
+    error_tool: string;
+}
+export interface CommandClassification {
+    command_type: CommandType;
+    command_tool: string;
+}
+/**
+ * Classify an error message + tool name into one of seed's 10 categories.
+ *
+ * Patterns (ordered most-specific → least-specific so the first match
+ * wins). Rationale for each pattern is inline.
+ *
+ *  • file_not_found   — Node's ENOENT, Python's FileNotFoundError, and
+ *                       Claude's "Cannot find module" all signal a
+ *                       missing file/path.
+ *  • command_not_found — POSIX shells emit "command not found" (exit 127)
+ *                       and "bash: <bin>: not found"; npx prints similar.
+ *  • edit_match_failed — Claude's Edit tool prints "old_string not found"
+ *                       or "matches multiple locations" when the patch
+ *                       context is stale.
+ *  • test_failed      — vitest/jest/pytest format `FAIL ` prefix; npm
+ *                       prints "Test failed".
+ *  • syntax_error     — SyntaxError (JS/TS), tsc TS-codes, or Python's
+ *                       SyntaxError header.
+ *  • runtime_error    — TypeError/ReferenceError/RangeError (uncaught).
+ *  • permission_denied — EACCES / "permission denied" from fs or sudo.
+ *  • git_conflict     — git's "CONFLICT" header on rebase/merge, plus
+ *                       "Merge conflict in".
+ *  • timeout          — explicit "timeout"/"timed out"/"ETIMEDOUT".
+ *  • unknown          — fallthrough; never throws.
+ */
+export declare function classifyError(message: unknown, toolName: unknown): ErrorClassification;
+/**
+ * Classify a Bash command into a workflow bucket. The bucket is derived
+ * from the (tool, sub-verb) pair: `npm test` → test, `npm run build` →
+ * build, `git commit` → git, `pip install` → install, etc.
+ *
+ * Heuristic ordering — most-specific verb checks first; falls through to
+ * `run` for generic execution and `other` for anything we can't classify.
+ */
+export declare function classifyCommand(command: unknown): CommandClassification;
+/**
+ * Bucketise a latency_ms reading into the four fixed buckets the
+ * dashboard renders. Bucket edges (in ms):
+ *
+ *    fast    : [0,     1_000)
+ *    medium  : [1_000, 10_000)
+ *    slow    : [10_000, 60_000)
+ *    timeout : [60_000, ∞)
+ *
+ * Defensive: negative and non-numeric inputs collapse to `fast` (the
+ * neutral bucket) rather than throwing, because hooks have been seen
+ * to emit `null` for very short calls.
+ */
+export declare function bucketizeDuration(latencyMs: unknown): DurationBucket;

package/build/session/error-classifier.js

@@ -0,0 +1,303 @@
+/**
+ * error-classifier — PRD-context-as-a-service §5 ABI parity
+ *
+ * Pure classifiers that derive per-event metadata from PostToolUse hook
+ * stdin, mirroring the seed shape produced by context-mode-platform's
+ * `seed.ts` so the bridge can ship full seed-shape parity.
+ *
+ * The dashboard reads three derived columns that are populated by the
+ * platform seeder but, per the OSS handoff (ANOMALY #3), are NEVER READ
+ * by the engine when ingesting live events:
+ *
+ *   • error_category   — one of 10 fixed buckets
+ *   • error_tool       — tool name that produced the error
+ *   • command_type     — git / test / build / lint / install / format / run / other
+ *   • command_tool     — first token of the command (npm, git, pytest, …)
+ *   • duration_bucket  — fast / medium / slow / timeout
+ *
+ * We populate them anyway, with sane defaults derived from message text
+ * and tool name, so the dashboard renders symmetrically with seed.
+ *
+ * Hard constraints:
+ *   1. Pure functions — no I/O, no globals, no module state.
+ *   2. `error_category` MUST match seed's `pickErrorClassification`
+ *      output exactly. The literal union below is the ABI.
+ *   3. No external dependencies; TypeScript stdlib only.
+ *   4. Robust to null / undefined / empty / malformed input.
+ */
+// ── Internal helpers ────────────────────────────────────────────────────
+/**
+ * Normalise message for case-insensitive substring matching. Coerces
+ * non-string input (null, undefined, numbers from errno objects) to
+ * empty string so downstream `.includes()` is always safe.
+ */
+function normalise(s) {
+    if (typeof s !== "string")
+        return "";
+    return s.toLowerCase();
+}
+/**
+ * Best-effort `error_tool` derivation. Prefers the PostToolUse-supplied
+ * `toolName`; falls back to scanning the message for a canonical tool
+ * name (Read/Edit/Write/Bash/Grep/Glob) so isolated error strings still
+ * resolve to a useful value rather than the literal "unknown".
+ */
+function deriveErrorTool(toolName, message) {
+    if (typeof toolName === "string" && toolName.trim().length > 0) {
+        return toolName.trim();
+    }
+    // Canonical Claude Code tool names that may appear in error strings.
+    const known = ["Edit", "Read", "Write", "Bash", "Grep", "Glob", "MultiEdit"];
+    for (const t of known) {
+        if (message.toLowerCase().includes(t.toLowerCase()))
+            return t;
+    }
+    return "unknown";
+}
+// ── Error classifier ────────────────────────────────────────────────────
+/**
+ * Classify an error message + tool name into one of seed's 10 categories.
+ *
+ * Patterns (ordered most-specific → least-specific so the first match
+ * wins). Rationale for each pattern is inline.
+ *
+ *  • file_not_found   — Node's ENOENT, Python's FileNotFoundError, and
+ *                       Claude's "Cannot find module" all signal a
+ *                       missing file/path.
+ *  • command_not_found — POSIX shells emit "command not found" (exit 127)
+ *                       and "bash: <bin>: not found"; npx prints similar.
+ *  • edit_match_failed — Claude's Edit tool prints "old_string not found"
+ *                       or "matches multiple locations" when the patch
+ *                       context is stale.
+ *  • test_failed      — vitest/jest/pytest format `FAIL ` prefix; npm
+ *                       prints "Test failed".
+ *  • syntax_error     — SyntaxError (JS/TS), tsc TS-codes, or Python's
+ *                       SyntaxError header.
+ *  • runtime_error    — TypeError/ReferenceError/RangeError (uncaught).
+ *  • permission_denied — EACCES / "permission denied" from fs or sudo.
+ *  • git_conflict     — git's "CONFLICT" header on rebase/merge, plus
+ *                       "Merge conflict in".
+ *  • timeout          — explicit "timeout"/"timed out"/"ETIMEDOUT".
+ *  • unknown          — fallthrough; never throws.
+ */
+export function classifyError(message, toolName) {
+    const msg = normalise(message);
+    const tool = typeof toolName === "string" ? toolName : "";
+    const error_tool = deriveErrorTool(toolName, typeof message === "string" ? message : "");
+    // Empty / malformed input → unknown bucket but still return a valid tool.
+    if (msg.length === 0) {
+        return { error_category: "unknown", error_tool };
+    }
+    // ── file_not_found ────────────────────────────────────────────────
+    // Node's ENOENT carries "no such file or directory"; Claude's Read
+    // tool surfaces the raw errno line. "cannot find module" is the
+    // require-resolution variant.
+    if (msg.includes("enoent") ||
+        msg.includes("no such file") ||
+        msg.includes("cannot find module") ||
+        msg.includes("filenotfounderror")) {
+        return { error_category: "file_not_found", error_tool };
+    }
+    // ── command_not_found ────────────────────────────────────────────
+    // POSIX shells: "<shell>: <bin>: command not found" → exit 127.
+    // Some hooks surface the bare exit code without the message.
+    if (msg.includes("command not found") ||
+        msg.includes(": not found") ||
+        /\bexit(?:\s+code)?\s*[:=]?\s*127\b/.test(msg)) {
+        return { error_category: "command_not_found", error_tool };
+    }
+    // ── edit_match_failed ────────────────────────────────────────────
+    // Claude Code's Edit tool prints these exact phrases when the
+    // `old_string` parameter no longer matches the file contents.
+    // Gate on the Edit tool to avoid false positives when a Bash script
+    // happens to contain the literal text "old_string".
+    if ((tool === "Edit" || tool === "MultiEdit") &&
+        (msg.includes("old_string") ||
+            msg.includes("could not find string") ||
+            msg.includes("string to replace not found") ||
+            msg.includes("matches multiple"))) {
+        return { error_category: "edit_match_failed", error_tool };
+    }
+    // Edit-style failure can also leak without tool name — match the
+    // distinctive Claude phrasing alone.
+    if (msg.includes("old_string not found") || msg.includes("string to replace not found")) {
+        return { error_category: "edit_match_failed", error_tool };
+    }
+    // ── git_conflict ─────────────────────────────────────────────────
+    // Check BEFORE test_failed because "CONFLICT" can co-occur with the
+    // word "fail" in some merge outputs.
+    if (msg.includes("conflict") &&
+        (msg.includes("merge") || msg.includes("rebase") || msg.includes("git"))) {
+        return { error_category: "git_conflict", error_tool };
+    }
+    if (msg.startsWith("conflict") || msg.includes("merge conflict")) {
+        return { error_category: "git_conflict", error_tool };
+    }
+    // ── timeout ──────────────────────────────────────────────────────
+    // Check BEFORE test_failed; "test timed out" should bucket as timeout.
+    if (msg.includes("etimedout") ||
+        msg.includes("timed out") ||
+        msg.includes("timeout") ||
+        msg.includes("deadline exceeded")) {
+        return { error_category: "timeout", error_tool };
+    }
+    // ── permission_denied ────────────────────────────────────────────
+    if (msg.includes("eacces") ||
+        msg.includes("permission denied") ||
+        msg.includes("operation not permitted") ||
+        msg.includes("eperm")) {
+        return { error_category: "permission_denied", error_tool };
+    }
+    // ── syntax_error ─────────────────────────────────────────────────
+    // tsc errors look like "error TS2322"; JS SyntaxError header is exact.
+    if (msg.includes("syntaxerror") ||
+        /\berror\s+ts\d{3,5}\b/.test(msg) ||
+        msg.includes("unexpected token") ||
+        msg.includes("unexpected end of") ||
+        msg.includes("parse error")) {
+        return { error_category: "syntax_error", error_tool };
+    }
+    // ── test_failed ──────────────────────────────────────────────────
+    // vitest/jest print `FAIL `; npm test prints "Test failed".
+    // pytest prints "FAILED" capital. Bias to Bash tool but accept
+    // unscoped messages because hooks sometimes drop tool name.
+    if (msg.includes("test failed") ||
+        msg.includes("tests failed") ||
+        /\bfail(?:ed)?\b.*\btest\b/.test(msg) ||
+        /\b\d+\s+tests?\s+failed\b/.test(msg) ||
+        msg.includes("assertion") ||
+        /^fail\s/.test(msg) ||
+        /\nfail\s/.test(msg)) {
+        return { error_category: "test_failed", error_tool };
+    }
+    // ── runtime_error ────────────────────────────────────────────────
+    // Catch-all for uncaught JS exceptions and the common Python ones.
+    // Placed near the end so more specific buckets (syntax_error) win
+    // when both could match.
+    if (msg.includes("typeerror") ||
+        msg.includes("referenceerror") ||
+        msg.includes("rangeerror") ||
+        msg.includes("uncaught exception") ||
+        msg.includes("traceback (most recent call last)") ||
+        msg.includes("nullpointerexception")) {
+        return { error_category: "runtime_error", error_tool };
+    }
+    // Fallthrough — categorisation is best-effort; the dashboard renders
+    // "unknown" as a neutral bucket rather than dropping the event.
+    return { error_category: "unknown", error_tool };
+}
+// ── Command classifier ──────────────────────────────────────────────────
+/**
+ * Tokenise a shell command, stripping leading env-var prefixes
+ * (`FOO=bar npm run build`) and `sudo`/`time` wrappers so the first
+ * meaningful token is returned.
+ */
+function firstToken(command) {
+    const parts = command.trim().split(/\s+/);
+    let i = 0;
+    // Skip env-var assignments: KEY=value
+    while (i < parts.length && /^[A-Za-z_][A-Za-z0-9_]*=.*/.test(parts[i]))
+        i++;
+    // Skip common wrappers that aren't the "real" tool.
+    while (i < parts.length && (parts[i] === "sudo" || parts[i] === "time" || parts[i] === "nice"))
+        i++;
+    return parts[i] ?? "";
+}
+/**
+ * Classify a Bash command into a workflow bucket. The bucket is derived
+ * from the (tool, sub-verb) pair: `npm test` → test, `npm run build` →
+ * build, `git commit` → git, `pip install` → install, etc.
+ *
+ * Heuristic ordering — most-specific verb checks first; falls through to
+ * `run` for generic execution and `other` for anything we can't classify.
+ */
+export function classifyCommand(command) {
+    if (typeof command !== "string" || command.trim().length === 0) {
+        return { command_type: "other", command_tool: "" };
+    }
+    const raw = command.trim();
+    const tool = firstToken(raw).toLowerCase();
+    const lower = raw.toLowerCase();
+    // ── git ──────────────────────────────────────────────────────────
+    if (tool === "git") {
+        return { command_type: "git", command_tool: "git" };
+    }
+    // ── install ──────────────────────────────────────────────────────
+    // npm/pnpm/yarn/pip install variants come before generic test/build
+    // because `npm install --save-dev jest` would otherwise match test.
+    if (/\b(install|add|ci)\b/.test(lower) &&
+        (tool === "npm" || tool === "pnpm" || tool === "yarn" || tool === "bun" || tool === "pip" || tool === "pip3")) {
+        return { command_type: "install", command_tool: tool };
+    }
+    if (tool === "brew" && /\binstall\b/.test(lower)) {
+        return { command_type: "install", command_tool: "brew" };
+    }
+    // ── test ─────────────────────────────────────────────────────────
+    // Direct test runners, plus `npm test` / `pnpm test` / `yarn test`.
+    if (tool === "vitest" || tool === "jest" || tool === "pytest" ||
+        tool === "mocha" || tool === "playwright" || tool === "cypress") {
+        return { command_type: "test", command_tool: tool };
+    }
+    if ((tool === "npm" || tool === "pnpm" || tool === "yarn" || tool === "bun") &&
+        /\btest\b/.test(lower)) {
+        return { command_type: "test", command_tool: tool };
+    }
+    // ── lint ─────────────────────────────────────────────────────────
+    if (tool === "eslint" || tool === "tslint" || tool === "ruff" || tool === "flake8" || tool === "pylint") {
+        return { command_type: "lint", command_tool: tool };
+    }
+    if (/\blint\b/.test(lower) && (tool === "npm" || tool === "pnpm" || tool === "yarn" || tool === "bun")) {
+        return { command_type: "lint", command_tool: tool };
+    }
+    // ── format ───────────────────────────────────────────────────────
+    if (tool === "prettier" || tool === "black" || tool === "gofmt" || tool === "rustfmt") {
+        return { command_type: "format", command_tool: tool };
+    }
+    if (/\b(format|fmt)\b/.test(lower) && (tool === "npm" || tool === "pnpm" || tool === "yarn" || tool === "bun")) {
+        return { command_type: "format", command_tool: tool };
+    }
+    // ── build ────────────────────────────────────────────────────────
+    if (tool === "tsc" || tool === "webpack" || tool === "rollup" || tool === "vite" || tool === "esbuild" || tool === "make") {
+        return { command_type: "build", command_tool: tool };
+    }
+    if (/\bbuild\b/.test(lower) && (tool === "npm" || tool === "pnpm" || tool === "yarn" || tool === "bun" || tool === "cargo" || tool === "go")) {
+        return { command_type: "build", command_tool: tool };
+    }
+    // ── run ──────────────────────────────────────────────────────────
+    // Generic execution — `npm run …` (with no recognised sub-verb),
+    // `node script.js`, `python -m foo`, etc.
+    if ((tool === "npm" || tool === "pnpm" || tool === "yarn" || tool === "bun") &&
+        /\brun\b/.test(lower)) {
+        return { command_type: "run", command_tool: tool };
+    }
+    if (tool === "node" || tool === "python" || tool === "python3" || tool === "deno" || tool === "bun") {
+        return { command_type: "run", command_tool: tool };
+    }
+    // Fallthrough — preserves the tool name so the dashboard can still
+    // chart "other" commands by their underlying binary.
+    return { command_type: "other", command_tool: tool || "" };
+}
+// ── Duration bucketiser ────────────────────────────────────────────────
+/**
+ * Bucketise a latency_ms reading into the four fixed buckets the
+ * dashboard renders. Bucket edges (in ms):
+ *
+ *    fast    : [0,     1_000)
+ *    medium  : [1_000, 10_000)
+ *    slow    : [10_000, 60_000)
+ *    timeout : [60_000, ∞)
+ *
+ * Defensive: negative and non-numeric inputs collapse to `fast` (the
+ * neutral bucket) rather than throwing, because hooks have been seen
+ * to emit `null` for very short calls.
+ */
+export function bucketizeDuration(latencyMs) {
+    const n = typeof latencyMs === "number" && Number.isFinite(latencyMs) ? latencyMs : 0;
+    if (n < 1_000)
+        return "fast";
+    if (n < 10_000)
+        return "medium";
+    if (n < 60_000)
+        return "slow";
+    return "timeout";
+}