@levnikolaevich/hex-line-mcp 1.2.0 → 1.3.1

package/README.md

@@ -3,7 +3,9 @@
 Hash-verified file editing MCP + token efficiency hook for AI coding agents.
 
 [![npm](https://img.shields.io/npm/v/@levnikolaevich/hex-line-mcp)](https://www.npmjs.com/package/@levnikolaevich/hex-line-mcp)
-![License](https://img.shields.io/badge/license-MIT-green)
+[![downloads](https://img.shields.io/npm/dm/@levnikolaevich/hex-line-mcp)](https://www.npmjs.com/package/@levnikolaevich/hex-line-mcp)
+[![license](https://img.shields.io/npm/l/@levnikolaevich/hex-line-mcp)](./LICENSE)
+![node](https://img.shields.io/node/v/@levnikolaevich/hex-line-mcp)
 
 Every line carries an FNV-1a content hash. Every edit must present those hashes back -- proving the agent is editing what it thinks it's editing. No stale context, no silent corruption.
 
@@ -16,10 +18,10 @@ Every line carries an FNV-1a content hash. Every edit must present those hashes
 | `read_file` | Read file with hash-annotated lines and range checksums | Partial reads via `offset`/`limit` |
 | `edit_file` | Hash-verified edits with anchor or text replacement | Returns compact diff via `diff` package |
 | `write_file` | Create new file or overwrite, auto-creates parent dirs | Path validation, no hash overhead |
-| `grep_search` | Search with ripgrep, returns hash-annotated matches | Edit-ready results -- search then edit directly |
+| `grep_search` | Search with ripgrep, 3 output modes, per-group checksums | Edit-ready: grep -> edit directly with checksums |
 | `outline` | AST-based structural overview via tree-sitter WASM | 95% token reduction (10 lines instead of 500) |
 | `verify` | Check if held range checksums are still valid | Single-line response avoids full re-read |
-| `directory_tree` | Compact directory tree with .gitignore support | Skips node_modules/.git, shows file sizes |
+| `directory_tree` | Compact directory tree with root .gitignore support | Skips node_modules/.git, shows file sizes |
 | `get_file_info` | File metadata without reading content | Size, lines, mtime, type, binary detection |
 | `setup_hooks` | Configure PreToolUse + PostToolUse hooks for Claude/Gemini/Codex | One call sets up everything, idempotent |
 | `changes` | Compare file against git ref, shows added/removed/modified symbols | AST-level semantic diff |
@@ -106,6 +108,7 @@ Read a file with FNV-1a hash-annotated lines and range checksums. Supports direc
 | Parameter | Type | Required | Description |
 |-----------|------|----------|-------------|
 | `path` | string | yes | File or directory path |
+| `paths` | string[] | no | Array of file paths to read (batch mode) |
 | `offset` | number | no | Start line, 1-indexed (default: 1) |
 | `limit` | number | no | Max lines to return (default: 2000, 0 = all) |
 | `plain` | boolean | no | Omit hashes, output `lineNum\|content` instead |
@@ -121,7 +124,7 @@ checksum: 1-50:f7e2a1b0
 
 ### edit_file
 
-Edit using hash-verified anchors or text replacement. Returns a unified diff.
+Edit using hash-verified anchors or text replacement. Returns diff + post-edit checksums for chaining edits.
 
 | Parameter | Type | Required | Description |
 |-----------|------|----------|-------------|
@@ -137,7 +140,8 @@ Edit operations (JSON array):
   {"set_line": {"anchor": "ab.12", "new_text": "replacement line"}},
   {"replace_lines": {"start_anchor": "ab.10", "end_anchor": "cd.15", "new_text": "..."}},
   {"insert_after": {"anchor": "ab.20", "text": "inserted line"}},
-  {"replace": {"old_text": "find this", "new_text": "replace with", "all": false}}
+  {"replace": {"old_text": "unique text", "new_text": "replacement"}},
+  {"replace": {"old_text": "find all", "new_text": "replace all", "all": true}}
 ]
 ```
 
@@ -152,17 +156,27 @@ Create a new file or overwrite an existing one. Creates parent directories autom
 
 ### grep_search
 
-Search file contents using ripgrep with hash-annotated results.
+Search file contents using ripgrep. Three output modes: `content` (hash-annotated with checksums), `files` (paths only), `count` (match counts).
 
 | Parameter | Type | Required | Description |
 |-----------|------|----------|-------------|
-| `pattern` | string | yes | Regex search pattern |
+| `pattern` | string | yes | Search pattern (regex by default, literal if `literal:true`) |
 | `path` | string | no | Directory or file to search (default: cwd) |
 | `glob` | string | no | Glob filter, e.g. `"*.ts"` |
 | `type` | string | no | File type filter, e.g. `"js"`, `"py"` |
+| `output` | enum | no | Output format: `"content"` (default), `"files"`, `"count"` |
 | `case_insensitive` | boolean | no | Ignore case |
-| `context` | number | no | Context lines around matches |
+| `smart_case` | boolean | no | CI when lowercase, CS when uppercase (`-S`) |
+| `literal` | boolean | no | Literal string search, no regex (`-F`) |
+| `multiline` | boolean | no | Pattern can span multiple lines (`-U`) |
+| `context` | number | no | Symmetric context lines around matches (`-C`) |
+| `context_before` | number | no | Context lines BEFORE match (`-B`) |
+| `context_after` | number | no | Context lines AFTER match (`-A`) |
 | `limit` | number | no | Max matches per file (default: 100) |
+| `total_limit` | number | no | Total match events across all files; multiline matches count as 1 (0 = unlimited) |
+| `plain` | boolean | no | Omit hash tags, return `file:line:content` |
+
+**Content mode** returns per-group checksums enabling direct `replace_lines` from grep results without intermediate `read_file`.
 
 ### outline
 
@@ -189,13 +203,16 @@ Returns a single-line confirmation or lists changed ranges.
 
 ### directory_tree
 
-Compact directory tree with .gitignore support and file sizes.
+Compact directory tree with root .gitignore support (path-based rules, negation, dir-only). Nested .gitignore files are not loaded.
 
 | Parameter | Type | Required | Description |
 |-----------|------|----------|-------------|
 | `path` | string | yes | Directory path |
-| `max_depth` | number | no | Max recursion depth (default: 3) |
-| `gitignore` | boolean | no | Respect .gitignore patterns (default: true) |
+| `pattern` | string | no | Glob filter on names (e.g. `"*-mcp"`, `"*.mjs"`). Returns flat match list instead of tree |
+| `type` | string | no | `"file"`, `"dir"`, or `"all"` (default). Like `find -type f/d` |
+| `max_depth` | number | no | Max recursion depth (default: 3, or 20 in pattern mode) |
+| `gitignore` | boolean | no | Respect root .gitignore patterns (default: true). Nested .gitignore not supported |
+| `format` | string | no | `"compact"` = names only, no sizes, depth 1. `"full"` = default with sizes |
 
 Skips `node_modules`, `.git`, `dist`, `build`, `__pycache__`, `.next`, `coverage` by default.
 
@@ -211,39 +228,43 @@ Returns: size, line count, modification time (absolute + relative), file type, b
 
 ## Hook
 
-The unified PostToolUse hook (`hook.mjs`) handles two concerns:
+The unified hook (`hook.mjs`) handles four events:
+
+### PreToolUse: Tool Redirect
 
-### Hex-line Reminder
+Blocks built-in `Read`, `Edit`, `Write`, `Grep` on text files and redirects to hex-line equivalents. Binary files (images, PDFs, notebooks, archives, executables, fonts, media) are excluded.
 
-Triggers on built-in `Read`, `Edit`, `Write`, `Grep` tool usage for text files. Outputs a short reminder to stderr (exit code 2) nudging the agent to use the corresponding hex-line tool instead.
+### PreToolUse: Bash Redirect + Dangerous Blocker
 
-Binary files (images, PDFs, notebooks, archives, executables, fonts, media) are excluded -- those should use built-in tools.
+Intercepts simple Bash commands (`cat`, `head`, `tail`, `ls`, `find`, `grep`, `sed -i`, etc.) and redirects to hex-line tools. Blocks dangerous commands (`rm -rf /`, `git push --force`, `git reset --hard`, `DROP TABLE`, `chmod 777`, `mkfs`, `dd`).
 
-### RTK Output Filter
+### PostToolUse: RTK Output Filter
 
 Triggers on `Bash` tool output exceeding 50 lines. Pipeline:
 
 1. **Detect command type** -- npm install, test, build, pip install, git verbose, or generic
-2. **Type-specific summary** -- extracts key metrics (e.g., `npm install: 42 added, 3 warnings`)
-3. **Normalize** -- replaces UUIDs, timestamps, IPs, hex values, large numbers with placeholders
-4. **Deduplicate** -- collapses identical normalized lines with `(xN)` counts
-5. **Truncate** -- keeps first 12 + last 12 lines, omits the middle
+2. **Normalize** -- replaces UUIDs, timestamps, IPs, hex values, large numbers with placeholders
+3. **Deduplicate** -- collapses identical normalized lines with `(xN)` counts
+4. **Truncate** -- keeps first 15 + last 15 lines, omits the middle
 
 Configuration constants in `hook.mjs`:
 
 | Constant | Default | Purpose |
-|----------|---------|---------|
+|----------|---------|--------|
 | `LINE_THRESHOLD` | 50 | Minimum lines to trigger filtering |
-| `TRUNCATE_LIMIT` | 30 | Lines below this are kept as-is after dedup |
-| `HEAD_LINES` | 12 | Lines to keep from start |
-| `TAIL_LINES` | 12 | Lines to keep from end |
+| `HEAD_LINES` | 15 | Lines to keep from start |
+| `TAIL_LINES` | 15 | Lines to keep from end |
+
+### SessionStart: Tool Preferences
+
+Injects hex-line tool preference list into agent context at session start.
 
 ## Architecture
 
 ```
 hex-line-mcp/
-  server.mjs          MCP server (stdio transport, 6 tools)
-  hook.mjs            PostToolUse hook (reminder + RTK filter)
+  server.mjs          MCP server (stdio transport, 11 tools)
+  hook.mjs            Unified hook (PreToolUse + PostToolUse + SessionStart)
   package.json
   lib/
     hash.mjs          FNV-1a hashing, 2-char tags, range checksums
@@ -252,6 +273,13 @@ hex-line-mcp/
     search.mjs        ripgrep wrapper with hash-annotated results
     outline.mjs       tree-sitter WASM AST outline
     verify.mjs        Range checksum verification
+    info.mjs          File metadata (size, lines, mtime, type)
+    tree.mjs          Directory tree with .gitignore support
+    changes.mjs       Semantic git diff via AST
+    bulk-replace.mjs  Multi-file search-and-replace
+    setup.mjs         Hook installation for Claude/Gemini/Codex
+    format.mjs        Output formatting utilities
+    coerce.mjs        Parameter alias mapping
     security.mjs      Path validation, binary detection, size limits
     normalize.mjs     Output normalization, deduplication, truncation
 ```
@@ -283,16 +311,6 @@ FNV-1a accumulator over all line hashes in the range (little-endian byte feed).
 - Write path validation (ancestor directory must exist)
 - Directory restrictions delegated to Claude Code sandbox
 
-## Differences from trueline-mcp
-
-| Aspect | hex-line-mcp | trueline-mcp |
-|--------|---------------|--------------|
-| Hash algorithm | FNV-1a (pure JS, zero dependencies) | xxHash (native addon) |
-| Diff output | Compact unified diff via `diff` npm package | Custom diff implementation |
-| Hook | Unified `hook.mjs` (reminder + RTK filter) | Separate hook scripts |
-| Path security | Canonicalization + binary detection, no ALLOWED_DIRS | Explicit ALLOWED_DIRS allowlist |
-| Transport | stdio only | stdio |
-| Outline | tree-sitter WASM (15+ languages) | tree-sitter WASM |
 
 ## FAQ
 
@@ -320,7 +338,7 @@ Outline works on code files only (15+ languages via tree-sitter WASM). For markd
 <details>
 <summary><b>How does the RTK filter reduce tokens?</b></summary>
 
-The PostToolUse hook normalizes Bash output (replaces UUIDs, timestamps, IPs with placeholders), deduplicates identical lines, and truncates to first 12 + last 12 lines. Average savings: 45% (flat) / 52% (weighted) across 18 benchmark scenarios.
+The PostToolUse hook normalizes Bash output (replaces UUIDs, timestamps, IPs with placeholders), deduplicates identical lines, and truncates to first 15 + last 15 lines. Average savings: 45% (flat) / 52% (weighted) across 18 benchmark scenarios.
 
 </details>
 
@@ -331,6 +349,14 @@ Yes. Remove the PreToolUse hook from `.claude/settings.local.json`. The MCP tool
 
 </details>
 
+## Hex Family
+
+| Package | Purpose | npm |
+|---------|---------|-----|
+| [hex-line-mcp](https://www.npmjs.com/package/@levnikolaevich/hex-line-mcp) | Local file editing with hash verification + hooks | [![npm](https://img.shields.io/npm/v/@levnikolaevich/hex-line-mcp)](https://www.npmjs.com/package/@levnikolaevich/hex-line-mcp) |
+| [hex-ssh-mcp](https://www.npmjs.com/package/@levnikolaevich/hex-ssh-mcp) | Remote file editing over SSH | [![npm](https://img.shields.io/npm/v/@levnikolaevich/hex-ssh-mcp)](https://www.npmjs.com/package/@levnikolaevich/hex-ssh-mcp) |
+| [hex-graph-mcp](https://www.npmjs.com/package/@levnikolaevich/hex-graph-mcp) | Code knowledge graph with AST indexing | [![npm](https://img.shields.io/npm/v/@levnikolaevich/hex-graph-mcp)](https://www.npmjs.com/package/@levnikolaevich/hex-graph-mcp) |
+
 ## License
 
 MIT

package/hook.mjs

@@ -23,7 +23,7 @@
  * Exit 2 = block (PreToolUse) or feedback via stderr (PostToolUse)
  */
 
-import { deduplicateLines, smartTruncate } from "./lib/normalize.mjs";
+import { normalizeOutput } from "./lib/normalize.mjs";
 import { readFileSync } from "node:fs";
 import { resolve } from "node:path";
 import { homedir } from "node:os";
@@ -39,17 +39,31 @@ const BINARY_EXT = new Set([
     ".ttf", ".otf", ".woff", ".woff2",
 ]);
 
+const REVERSE_TOOL_HINTS = {
+    "mcp__hex-line__read_file":      "Read (file_path, offset, limit)",
+    "mcp__hex-line__edit_file":      "Edit (file_path, old_string, new_string)",
+    "mcp__hex-line__write_file":     "Write (file_path, content)",
+    "mcp__hex-line__grep_search":    "Grep (pattern, path)",
+    "mcp__hex-line__directory_tree": "Glob (pattern) or Bash(ls)",
+    "mcp__hex-line__get_file_info":  "Bash(stat/wc)",
+    "mcp__hex-line__outline":        "Read with offset/limit",
+    "mcp__hex-line__verify":         "Read the file again with Read",
+    "mcp__hex-line__changes":        "Bash(git diff)",
+    "mcp__hex-line__bulk_replace":   "Edit for each file",
+    "mcp__hex-line__setup_hooks":    "Not available (hex-line disabled)",
+};
+
 const TOOL_HINTS = {
     Read:  "mcp__hex-line__read_file (not Read). For writing: write_file (no prior Read needed)",
     Edit:  "mcp__hex-line__edit_file (not Edit, not sed -i). read_file first for hashes",
     Write: "mcp__hex-line__write_file (not Write). No prior Read needed",
-    Grep:  "mcp__hex-line__grep_search (not Grep). Params: case_insensitive, smart_case",
+    Grep:  "mcp__hex-line__grep_search (not Grep). Params: output, literal, context_before, context_after, multiline",
     cat:   "mcp__hex-line__read_file (not cat/head/tail/less/more)",
     head:  "mcp__hex-line__read_file with limit param (not head)",
     tail:  "mcp__hex-line__read_file with offset param (not tail)",
     ls:    "mcp__hex-line__directory_tree with pattern param (not ls/find/tree). E.g. pattern='*-mcp' type='dir'",
     stat:  "mcp__hex-line__get_file_info (not stat/wc/file)",
-    grep:  "mcp__hex-line__grep_search (not grep/rg). Params: case_insensitive, smart_case",
+    grep:  "mcp__hex-line__grep_search (not grep/rg). Params: output, literal, context_before, context_after, multiline",
     sed:   "mcp__hex-line__edit_file (not sed -i). read_file first for hashes",
     diff:  "mcp__hex-line__changes (not diff). Git-based semantic diff",
     outline: "mcp__hex-line__outline (before reading large code files)",
@@ -138,6 +152,39 @@ function detectCommandType(cmd) {
     return "generic";
 }
 
+/** Cache: null = not computed yet */
+let _hexLineDisabled = null;
+
+/**
+ * Check if hex-line MCP is disabled for the current project.
+ * Reads ~/.claude.json → projects.{cwd}.disabledMcpServers.
+ * Fail-open: returns false on any error.
+ */
+function isHexLineDisabled(configPath) {
+    if (_hexLineDisabled !== null) return _hexLineDisabled;
+    _hexLineDisabled = false;
+    try {
+        const p = configPath || resolve(homedir(), ".claude.json");
+        const claudeJson = JSON.parse(readFileSync(p, "utf-8"));
+        const projects = claudeJson.projects;
+        if (!projects || typeof projects !== "object") return _hexLineDisabled;
+        const cwd = process.cwd().replace(/\\/g, "/").replace(/\/$/, "").toLowerCase();
+        for (const [path, config] of Object.entries(projects)) {
+            if (path.replace(/\\/g, "/").replace(/\/$/, "").toLowerCase() === cwd) {
+                const disabled = config.disabledMcpServers;
+                if (Array.isArray(disabled) && disabled.includes("hex-line")) {
+                    _hexLineDisabled = true;
+                }
+                break;
+            }
+        }
+    } catch { /* fail open */ }
+    return _hexLineDisabled;
+}
+
+/** Reset cache (for testing). */
+function _resetHexLineDisabledCache() { _hexLineDisabled = null; }
+
 function block(reason, context) {
     const output = {
         hookSpecificOutput: {
@@ -237,6 +284,28 @@ function handlePreToolUse(data) {
     process.exit(0);
 }
 
+// ---- PreToolUse REVERSE handler (hex-line disabled) ----
+
+function handlePreToolUseReverse(data) {
+    const toolName = data.tool_name || "";
+
+    // Agent tries hex-line tool that's disabled → redirect to built-in
+    if (toolName.startsWith("mcp__hex-line__")) {
+        const builtIn = REVERSE_TOOL_HINTS[toolName];
+        if (builtIn) {
+            const target = builtIn.split(" ")[0];
+            block(
+                `hex-line is disabled in this project. Use ${target}`,
+                `hex-line disabled. Use built-in: ${builtIn}`
+            );
+        }
+        block("hex-line is disabled in this project", "Disabled via project settings");
+    }
+
+    // All built-in tools — approve silently
+    process.exit(0);
+}
+
 // ---- PostToolUse handler ----
 
 function handlePostToolUse(data) {
@@ -266,10 +335,8 @@ function handlePostToolUse(data) {
 
     const type = detectCommandType(command);
 
-    // Pipeline: deduplicate -> smart truncate
-    const deduped = deduplicateLines(lines);
-    const dedupedText = deduped.join("\n");
-    const filtered = smartTruncate(dedupedText, HEAD_LINES, TAIL_LINES);
+    // Pipeline: normalize -> deduplicate -> smart truncate
+    const filtered = normalizeOutput(lines.join("\n"), { headLines: HEAD_LINES, tailLines: TAIL_LINES });
     const filteredCount = filtered.split("\n").length;
 
     const header = `RTK FILTERED: ${type} (${originalCount} lines -> ${filteredCount} lines)`;
@@ -331,21 +398,36 @@ function handleSessionStart() {
 }
 
 // ---- Main: read stdin, route by hook_event_name ----
+// Guard: only run when executed directly, not when imported for testing
 
-let input = "";
-process.stdin.on("data", (chunk) => {
-    input += chunk;
-});
-process.stdin.on("end", () => {
-    try {
-        const data = JSON.parse(input);
-        const event = data.hook_event_name || "";
-
-        if (event === "SessionStart") handleSessionStart();
-        else if (event === "PreToolUse") handlePreToolUse(data);
-        else if (event === "PostToolUse") handlePostToolUse(data);
-        else process.exit(0);
-    } catch {
-        process.exit(0);
-    }
-});
+import { fileURLToPath } from "node:url";
+
+if (process.argv[1] === fileURLToPath(import.meta.url)) {
+    let input = "";
+    process.stdin.on("data", (chunk) => {
+        input += chunk;
+    });
+    process.stdin.on("end", () => {
+        try {
+            const data = JSON.parse(input);
+            const event = data.hook_event_name || "";
+
+            if (isHexLineDisabled()) {
+                // REVERSE MODE: block hex-line calls, approve everything else
+                if (event === "PreToolUse") handlePreToolUseReverse(data);
+                process.exit(0); // SessionStart, PostToolUse — silent exit
+            }
+
+            // NORMAL MODE
+            if (event === "SessionStart") handleSessionStart();
+            else if (event === "PreToolUse") handlePreToolUse(data);
+            else if (event === "PostToolUse") handlePostToolUse(data);
+            else process.exit(0);
+        } catch {
+            process.exit(0);
+        }
+    });
+}
+
+// ---- Exports for testing ----
+export { isHexLineDisabled, _resetHexLineDisabledCache };

package/lib/bulk-replace.mjs

@@ -1,16 +1,18 @@
-import { readFileSync, writeFileSync } from "node:fs";
-import { execSync } from "node:child_process";
+import { writeFileSync } from "node:fs";
+import { execFileSync } from "node:child_process";
 import { resolve } from "node:path";
 import { simpleDiff } from "./edit.mjs";
+import { normalizePath } from "./security.mjs";
+import { readText, MAX_OUTPUT_CHARS } from "./format.mjs";
 
 export function bulkReplace(rootDir, globPattern, replacements, opts = {}) {
     const { dryRun = false, maxFiles = 100 } = opts;
-    const abs = resolve(rootDir);
+    const abs = resolve(normalizePath(rootDir));
 
     // Find files via ripgrep (respects .gitignore)
     let files;
     try {
-        const rgOut = execSync(`rg --files -g "${globPattern}" "${abs}"`, { encoding: "utf-8", timeout: 10000 });
+        const rgOut = execFileSync("rg", ["--files", "-g", globPattern, abs], { encoding: "utf-8", timeout: 10000 });
         files = rgOut.trim().split("\n").filter(Boolean);
     } catch (e) {
         if (e.status === 1) return "No files matched the glob pattern.";
@@ -23,12 +25,12 @@ export function bulkReplace(rootDir, globPattern, replacements, opts = {}) {
 
     const results = [];
     let changed = 0, skipped = 0, errors = 0;
-    const MAX_OUTPUT = 80000;
+    const MAX_OUTPUT = MAX_OUTPUT_CHARS;
     let totalChars = 0;
 
     for (const file of files) {
         try {
-            const original = readFileSync(file, "utf-8").replace(/\r\n/g, "\n");
+            const original = readText(file);
             let content = original;
 
             for (const { old: oldText, new: newText } of replacements) {

package/lib/changes.mjs

@@ -6,10 +6,11 @@
  * without temp files.
  */
 
-import { execSync } from "node:child_process";
-import { readFileSync, statSync } from "node:fs";
+import { execFileSync } from "node:child_process";
+import { statSync } from "node:fs";
 import { extname } from "node:path";
 import { validatePath } from "./security.mjs";
+import { readText } from "./format.mjs";
 import { outlineFromContent } from "./outline.mjs";
 
 /**
@@ -50,7 +51,7 @@ function toSymbolMap(entries) {
  * Get relative path from git root for `git show`.
  */
 function gitRelativePath(absPath) {
-    const root = execSync("git rev-parse --show-toplevel", {
+    const root = execFileSync("git", ["rev-parse", "--show-toplevel"], {
         cwd: absPath.replace(/[/\\][^/\\]+$/, ""),
         encoding: "utf-8",
         timeout: 5000,
@@ -79,7 +80,7 @@ export async function fileChanges(filePath, compareAgainst = "HEAD") {
     // Directory: return git diff --stat (compact file list, no content reads)
     if (statSync(real).isDirectory()) {
         try {
-            const stat = execSync(`git diff --stat "${compareAgainst}" -- .`, {
+            const stat = execFileSync("git", ["diff", "--stat", compareAgainst, "--", "."], {
                 cwd: real,
                 encoding: "utf-8",
                 timeout: 10000,
@@ -94,7 +95,7 @@ export async function fileChanges(filePath, compareAgainst = "HEAD") {
     const ext = extname(real).toLowerCase();
 
     // Check if outline supports this extension
-    const currentContent = readFileSync(real, "utf-8").replace(/\r\n/g, "\n");
+    const currentContent = readText(real);
     const currentResult = await outlineFromContent(currentContent, ext);
     if (!currentResult) {
         return `Cannot outline ${ext} files. Supported: .js .mjs .ts .py .go .rs .java .c .cpp .cs .rb .php .kt .swift .sh .bash`;
@@ -104,7 +105,7 @@ export async function fileChanges(filePath, compareAgainst = "HEAD") {
     const relPath = gitRelativePath(real);
     let gitContent;
     try {
-        gitContent = execSync(`git show "${compareAgainst}:${relPath}"`, {
+        gitContent = execFileSync("git", ["show", `${compareAgainst}:${relPath}`], {
             cwd: real.replace(/[/\\][^/\\]+$/, ""),
             encoding: "utf-8",
             timeout: 5000,

package/lib/coerce.mjs

@@ -12,12 +12,6 @@ const ALIASES = {
     // grep_search
     query: "pattern",
     search: "pattern",
-    max_results: "limit",
-    maxResults: "limit",
-    maxMatches: "limit",
-    max_matches: "limit",
-    contextLines: "context",
-    context_lines: "context",
     ignoreCase: "case_insensitive",
     ignore_case: "case_insensitive",