Bragproof

← Reports

@remcp/runtime

Starts with no setup0.2.258npm

Installed, started with no arguments, no environment and no input, and listed its tools. Run 2026-11-16.

Needs to start: nothing
Returns a result when called with no setup: 4 of 5 tools tried
Contacts while installing and starting: nothing

53tools listed
0hosts contacted
1file written
0.3 sto start
18dependencies

Its tool definitions changed since version 0.2.251. If you approved the earlier ones, read them again.

What it contacted

Nothing besides the npm registry it was installed from. Watched from outside its sandbox while it was installed and started with no input. Its tools may still contact things when called.

Starting it

npx -y @remcp/runtime@0.2.258

Installing a package runs its code. Use a sandbox for anything you have not read.

Calling its tools

5 tools that take no input and look read-only were called once with nothing. 4 returned a result.

read_filesReturned a result under 0.1 s · 30 characters
list_searchesReturned a result under 0.1 s · 19 characters
list_sessionsReturned a result under 0.1 s · 28 characters
get_system_infoReturned a result 0.1 s · 700 characters
list_processesNo usable answer tool reported an error: Could not list processes: spawn ps ENOENT

During the calls it contacted nothing it had not already contacted.

Only tools whose names read like looking something up, with no required input, are tried, five at most. Tools that need input, and anything that sounds like it changes or sends something, are not called.

What its tools tell an agent

In the tool descriptions, nothing that talks about the user, other tools or private files, or uses hidden characters.

Found by simple pattern rules in each tool's description (up to 3,000 characters), quoted as written. Parameter descriptions are not scanned. A clean result is not a safety review.

Tools

read_file path, offset, length
Read a plain text or line-oriented file on this computer: source, config, Markdown, scripts or logs. Use read_document for PDF, DOCX, XLSX and other structured documents. Legacy .docx/.pdf extraction remains supported for compatibility. Use offset and length to page through large text files; a negative offset reads from the end.
read_files path, pattern, max_files, max_lines_per_file, include_ignored
Read the contents of files discovered by one glob under a directory, each section prefixed with its path and line count. Use this when the matched file contents are the result you want. Use start_search instead to find filenames or text matches without reading every matched file. If you already know the exact paths, use read_multiple_files; for one known path use read_file.
read_multiple_files paths
Read several explicitly named text files in one call. Use this when the exact paths are already known; each file is returned separately and one failure does not stop the others. Do not use this for glob discovery: use read_files instead. For one path use read_file.
read_image path
Return an image file (PNG, JPEG, GIF, WebP, BMP, AVIF, or SVG) as native MCP image content so screenshots, photos, and diagrams can be inspected or shown. Use read_binary only for byte-for-byte transfer, not merely to display an image. Fails above the inline size limit.
read_binary path, offset_bytes, length_bytes
Read any file as base64 chunks for byte-for-byte transfer of binaries, archives, documents, or oversized media. For a normal image that should be inspected or shown, use read_image instead. Returns size, offset, and nextOffsetBytes; call again with offset_bytes set to nextOffsetBytes until complete is true.
48 more tools
write_binary path, data, mode
Write base64 data to a file byte for byte, creating parent directories. Use mode "append" to send a large file as consecutive chunks. Replaces the file by default.
hash_file path, algorithm
Compute a checksum of a file without reading it into memory. Useful to verify a copy, compare two files, or confirm a download.
list_directory path, depth, pattern
List the files and directories at a path. Entries are prefixed with [DIR], [FILE], [LINK], or [DENIED] when a subdirectory cannot be read. depth controls how many directory levels are included and pattern filters file names by glob.
get_file_info path
Return metadata for a file or directory: type, size, timestamps, permissions, and for small text files the line count.
write_file path, content, mode
Create or replace the complete contents of exactly one text file. Parent directories are created automatically; mode "append" adds to the end. For a small exact-block edit use edit_block, for a known line range use replace_lines, and when you already have a unified diff use apply_patch. Use write_files for two or more independent complete files, and write_binary for binary bytes.
write_files files
Create or replace the complete contents of two or more text files in one batch, each with its own path, content, and optional mode. Use this for scaffolding or coherent multi-file writes; for one path use write_file. For partial edits use edit_block, replace_lines, replace_in_files, or apply_patch instead of rewriting whole files.
apply_patch patch, path, dry_run
Apply an already-prepared unified diff to one file or several files, matching each hunk with a little fuzz so small offsets and whitespace differences still apply. Prefer this when the change is naturally patch-shaped; use edit_block for one exact text block and write_file only when replacing a complete file. Pass dry_run to see the result first.
set_permissions path, mode, recursive, uid, gid
Change the permission mode of a file or directory, optionally recursively and optionally with a numeric owner. Use this to make a script executable after writing it.
edit_block file_path, old_string, new_string, expected_replacements, allow_fuzzy, dry_run
Replace one known text block in one file. Provide enough surrounding context to make old_string unique; the call fails unless the number of matches equals expected_replacements. Use write_file for a whole-file replacement and replace_lines for a known line range. When exact text is not found, a whitespace-tolerant match is attempted and reported. Pass dry_run to preview without writing.
replace_lines path, start_line, end_line, content, dry_run
Replace an inclusive 1-based line range with new text. Use this when line numbers define the target; use edit_block when matching existing text and write_file for a complete rewrite. The rest of the file, including its line endings, is preserved. Pass dry_run to preview without writing.
replace_in_files path, pattern, replacement, filePattern, regex, dry_run, maxFiles
Search and replace the same text or regular expression across multiple text files under a path and report what changed. Use edit_block for one known block in one file and write_file for one complete known file. Applies immediately; pass dry_run true to preview the affected files first.
diff_files left, right, context_lines
Show a unified diff between two local text files, with line counts. Useful to check what changed before reporting or reverting it.
move_to_trash source
Move a file or directory to the system trash instead of deleting it, so the change can be undone. When the trash is outside the device allowed roots, a .remcp-trash folder beside the file is used instead.
create_directory path, paths
Create one directory or many in a single call, including any missing parent directories. Succeeds when a directory already exists.
move_file source, destination, overwrite
Move or rename exactly one file or directory from one source to one destination. Replaces an existing destination file by default; pass overwrite false to refuse instead. Use move_paths for two or more moves.
copy_file source, destination, overwrite
Copy exactly one regular file to a new path. Directories are not copied recursively. Use copy_paths for directories or two or more copy operations; pass overwrite false to refuse an existing destination.
copy_paths paths, overwrite
Copy two or more paths, or copy a directory recursively, in one call. Each item has its own source and destination. For exactly one regular file use copy_file.
move_paths paths, overwrite
Move or rename two or more files/directories in one batch, each with its own source and destination. Falls back to copy-and-delete across filesystems. For one source/destination pair use move_file.
delete_path path, recursive
Delete exactly one file or directory. Directories are removed with their contents unless recursive is false; the filesystem root is refused. Use delete_paths for two or more targets, or move_to_trash when deletion should be reversible.
delete_paths paths, recursive
Delete two or more files/directories in one batch and report each result. For one target use delete_path. Use move_to_trash instead when deletion should be reversible.
create_archive paths, destination, format
Pack files and directories into a tar, tar.gz, or zip archive on the device, so a whole tree can be transferred or backed up in one call.
extract_archive archive, destination
Extract a tar, tar.gz, tar.bz2, tar.xz, or zip archive on the device into a directory, creating it when needed.
take_screenshot directory, keep
Capture the entire desktop only when full-screen pixels are actually needed. Prefer ui_snapshot or browser_snapshot for semantic inspection, and screenshot_region for one window, monitor, or rectangle. This tool uses the platform screenshot portal/utility and returns an image.
start_search path, pattern, searchType, filePattern, ignoreCase, maxResults, includeHidden, includeIgnored, contextLines, literalSearch
Start a filename or content search on this computer and return the first matches without reading every matched file in full. Use read_files instead when a glob is already known and the contents of all matching files are the desired result. Content searches return "path:line: text" rows. Use get_more_search_results to page and stop_search to stop a long search.
get_more_search_results sessionId, offset, length
Read more results from a search started with start_search. Offset is zero-based; a negative offset returns the last N results.
stop_search sessionId
Stop a running search. Results collected so far stay readable until the search is cleaned up.
list_searches
List active and recent searches on this computer with their status and result counts.
start_process command, timeout_ms
Run a shell command, script, build, test suite, or terminal server and return its initial output. Prefer dedicated semantic tools for jobs they already cover: launch_app for GUI apps, service for service-manager operations, power_action for lock/sleep/restart/shutdown, network for routine network inspection/tests, event_log for OS logs, installed_apps for software inventory, environment for environment facts, and open_path/reveal_path for opening or revealing files. The process keeps running so read_process_output or interact_with_process can be used later.
read_process_output pid, offset, length, timeout_ms
Read buffered output from a session started with start_process. Without an offset it returns output produced since the previous read.
interact_with_process pid, input, timeout_ms
Send one line of input to a running session and return the output it produces. Use this for REPLs and other interactive commands.
wait_for_process_output pid, pattern, timeout_ms
Wait until output from a running session matches a regular expression or literal string, then return the output collected since the previous read. Use this instead of polling read_process_output in a loop.
force_terminate pid
Stop a ReMCP terminal session started with start_process, escalating from SIGTERM to SIGKILL when it does not exit. Use kill_process for an arbitrary operating-system PID that is not a ReMCP terminal session.
list_sessions
List only terminal sessions started through ReMCP start_process, with status and runtime. Use list_processes for the computer-wide operating-system process list.
get_system_info
Report machine health and host facts for the paired computer: operating system and kernel, CPU model/load, memory pressure, free disk space, uptime, and default shell. Use environment for PATH/shell/runtime environment variables and get_runtime_info for ReMCP-specific version, policy, roots, and limits.
list_processes limit
List the computer-wide operating-system processes, highest CPU first, with pid, parent pid, CPU/memory usage, and command. Use list_sessions when the question is only about commands started through ReMCP start_process. Values that look like secrets are masked.
kill_process pid
Terminate an arbitrary operating-system process by pid. If the pid belongs to a ReMCP terminal session created by start_process, prefer force_terminate so ReMCP updates the session consistently and handles its process tree. Terminates the process and its children on Windows.
get_runtime_info
Report this computer's local ReMCP runtime: version, allowed roots, command policy, output limits, settable preferences, and usage-metric setting. Use get_system_info for OS/CPU/memory/disk facts and hosted get_configuration for deployment/account configuration. Read-only; use set_config_value only for telemetryEnabled, maxReadLines, maxBufferedLines, or maxOutputBytes. Access roots and command security stay local to the computer.
get_runtime_stats
Report counters from the current local runtime session on this machine: tool calls/failures, blocked commands, active terminal/search sessions, and usage-metric queue state. Use hosted get_usage_statistics for account-level historical usage aggregated across devices and days.
set_config_value key, value
Change one of this local runtime’s own preferences on the selected computer: telemetryEnabled, maxReadLines, maxBufferedLines or maxOutputBytes. This is not account-level device administration; use hosted device_action for ping, rename, enable/disable or removal. Access roots, blocked commands, shell/guardrails and the write limit remain local-only.
computer_snapshot max_ui_nodes, max_ui_depth, include_ui, ui_scope, semantic_tree, ui_browser_dom, include_screenshot, screenshot_region, screenshot_window_id, screenshot_monitor, screenshot_monitor_index, screenshot_padding, grid_columns, grid_rows, include_browser, browser_readiness, browser_max_nodes, include_ocr, ocr_language, ocr_max_words, ocr_psm, ocr_timeout_ms
Use this for the first look at an unfamiliar desktop state when you need one compact bundle of windows, semantic UI, displays, clipboard metadata and optional pixels/browser/OCR. Prefer it over repeated screenshots; narrow later with ui_find, browser_find or screenshot_region.
computer_action action, target, id, label, pid, app, window_title, name, role, automation_id, title, window_id, text, clear, value, x, y, selector, browser_text, delta_x, delta_y, delta, direction, wheel_times, from_x, from_y, to_x, to_y, from_id, to_id, button, hold_ms, duration_ms, ocr_text, ocr_region, ocr_language, ocr_psm, ocr_timeout_ms, endpoint, target_id, url_contains, option, targets, replace_selection, edits, backend
Use this only when one semantic click/type/invoke/select/focus/scroll/drag action must work across an uncertain target backend and may need Accessibility → browser DOM/CDP → OCR → coordinate fallback, or when multi_select/multi_edit avoids many identical native UI calls. When the backend is already known, prefer the specialist tool (ui_action, browser_action, type_text, drag_drop, scroll or window_action) for its full controls. Keep keyboard, clipboard, launch/open/reveal/notification and raw pointer gestures in their dedicated tools.
app_catalog app, category, filter, include_guides, check_installed, include_detected, refresh
Use this to discover curated workflows and, with include_detected=true, new OS software records. Linux prefers visible XDG GUI launchers; macOS .app copies have separate IDs and require PID executable provenance; Windows uses local Start Menu shortcuts with exact PID executable attestation; registry-only rows cannot authorize GUI access. Detection is not proof of an active GUI, permissions or a trusted connector. Use local- ids for inspect/find-visible-controls or explicit checked focus_control and one bounded set_control_text with old-value CAS; other effects are forbidden.
app_task app, intent, query, channel, post_url, expected_text, expected_value_token, replacement_text, commit, dry_run, target_id, window_id, control_id, expected_name, expected_role, pid, verify
Use this to execute one typed intent inside a curated app via trusted local API, scoped Accessibility or CDP without screenshots. New local- IDs support inspect/find-visible-controls and explicitly scoped non-destructive focus_control, after unique PID/window and semantic ID/name/role checks. Unknown apps cannot launch, click, invoke or submit; a single native text value can change only after explicit commit and exact old-value check (raw expected_text or an expiring blind token from inspect/find). Telegram edits require authenticated owner, original text and verified read-back.
app_execute mode, target_id, origin, endpoint, fields, task, commit, dry_run, operation_id, timeout_ms
Use this when you need one verified local task instead of a screenshot-click loop: fill 1–20 exact DOM fields in one origin-locked CDP evaluation with optimistic old-value check and readback, or execute one app_task via the registered API/AX/UIA/AT-SPI policy. Preview by default for mutations. Requires commit=true plus an operation UUID for writes; a private on-disk reservation blocks duplicate UUIDs across process restarts, with uncertain outcomes reported for reinspection. Never implicitly submits posts or forms, never executes scripts or guessed coordinates.
network action, host, port, timeout_ms
Use this for a compact network summary, full interface inventory, DNS/routes/listeners inspection or a bounded TCP connectivity test. Prefer action=summary for orientation and action=interfaces only when the complete adapter/address inventory is required; prefer this tool over shell ip/netstat/nslookup commands for routine diagnostics.
environment include_env
Use this for PATH, shell, Node/runtime environment paths, architecture and optionally sanitized environment variables. Use get_system_info for host health such as CPU/memory/disk/uptime, and get_runtime_info for ReMCP policy, roots and limits. Secret-looking values remain masked.
read_document path, sheet, max_cells
Use this for structured document content, especially PDF, DOCX and XLSX, and for CSV/JSON/XML when document-style parsing is useful. Use read_file for ordinary source code, logs, configuration, Markdown, or other plain line-oriented text. Prefer read_document over shell extraction or launching Office when the goal is document content; PDF reading uses the built-in parser first and may fall back to local pdftotext when embedded fonts prevent direct decoding.
pdf_action action, path, paths, output, output_dir, pattern, pages
Use this for PDF metadata/annotations inspection or structural merge/split/page extraction. For reading PDF text use read_document; supply an output/output_dir for write actions when preserving originals matters.

Descriptions are the package's own words, cut at 3,000 characters. Treat them as data.

Files it wrote

Paths inside its sandbox, where its home folder is /tmp and its own folder cannot be written to.

Install facts

LicenseMIT
Needsnode >=22.5.0
Sourcegithub.com/getremcp/remcp
Install scriptsmacos-temperature-sensor declare them. This runner does not run them.
Tarballsha512-iTGRlsI8m353n3oPslDRSILopTj/z/B0/YjgejDA+G14/qPBD2mrHfXJL4QmNf+fa/tiBlZAH1s1uY18Gxxj7g==

Versions reported

VersionRunResultTools
0.2.2582026-11-16Starts with no setup53
0.2.2512026-11-12Starts with no setup50

Check this report yourself

This site did not write the report. A public job on a GitHub-hosted machine did: the run, from this commit of the open-source runner. GitHub signed a token that names that job and the hash of the report, so any edit since would show.

Show it on your README

Bragproof: Starts with no setup

[![Bragproof](https://bragproof.smaverk.com/badge/@remcp/runtime.svg)](https://bragproof.smaverk.com/r/%40remcp/runtime)

New version out? Ask for a fresh report.