GitHub
Connect AI assistants to GitHub - manage repos, issues, PRs, and workflows through natural language.
Descripción del creador (original en inglés)
The cryptography a JavaScript or TypeScript codebase uses, and which a quantum computer would break.
Estos tres nunca se combinan en una sola puntuación: las opiniones, las mediciones y la popularidad son cosas distintas.
Todavía no escribimos un resumen sencillo para este. Abajo están los datos que leímos del registro.
Elige la app de IA que usas y copia la configuración.
Necesitas
¿Qué app usas?
Elige tu app arriba. La recordamos, así que la próxima vez el botón dirá “Instalar en …”.
Abre Claude Desktop → Settings → Developer → Edit Config y pega esto en claude_desktop_config.json
{
"mcpServers": {
"pqc-scan": {
"command": "npx",
"args": [
"-y",
"@microtoll/pqc-scan",
"mcp"
]
}
}
}Ejecuta esto en tu terminal
claude mcp add pqc-scan -- npx -y @microtoll/pqc-scan mcpPégalo en .cursor/mcp.json de tu proyecto, o en ~/.cursor/mcp.json
{
"mcpServers": {
"pqc-scan": {
"command": "npx",
"args": [
"-y",
"@microtoll/pqc-scan",
"mcp"
]
}
}
}Tu navegador te pide permiso para abrir la app; luego la app muestra la configuración y te pide confirmar. Los valores de las claves quedan vacíos para que los completes tú.
Pégalo en .vscode/mcp.json de tu proyecto
{
"servers": {
"pqc-scan": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@microtoll/pqc-scan",
"mcp"
]
}
}
}Tu navegador te pide permiso para abrir la app; luego la app muestra la configuración y te pide confirmar. Los valores de las claves quedan vacíos para que los completes tú.
Este se ejecuta en tu computadora, y ChatGPT solo puede conectarse a servidores en internet. Elige otra app.
Tu elección se guarda en una cookie en este dispositivo; no necesitas cuenta.
Esta configuración se genera a partir de la ficha del creador en el registro. Si el README dice otra cosa, sigue el README.
Se ejecuta en tu computadora; la lista de herramientas aparece en tu app de IA después de instalarlo.
Revisamos información que cualquiera puede ver; no hemos leído el código. Las revisiones que no aplican no se cuentan. · Revisado el 7 nov 2026
El nombre en el registro se verificó con la cuenta de GitHub microtoll, dueña del repo
Todavía no hemos leído este repo en GitHub
Todavía no hemos leído este repo en GitHub
Este se ejecuta en tu computadora; no hay endpoint en internet
Se ejecuta en tu computadora y usa ajustes locales en su lugar
Se ejecuta en tu computadora; solo podemos leer listas de herramientas de endpoints en internet
Comparado con todos los servidores activos de esta categoría en el registro. Las estrellas solo cuentan cuando quien lo publica es dueño del repo.
Superó 1 de 1 revisiones básicas aplicables. No hemos leído el código. Antes de agregar una clave o un token, revisa qué acceso pide el servidor y usa una clave que tenga solo los permisos que necesitas.
El registro no indica ninguna clave obligatoria, aunque algunos servidores te piden iniciar sesión la primera vez que los usas.
Se ejecuta en tu computadora. Tu app de IA lo inicia cuando hace falta, y los datos no pasan por los servidores del creador (a menos que el propio servidor llame a un servicio en línea).
El repo de GitHub cambió por última vez –, y la última versión (0.3.3) llegó al registro hace 31 días.
Las apps compatibles con MCP, como Claude Desktop, Claude Code, Cursor y VS Code. También necesitas Node.js en tu computadora.
En la misma categoría: GitHub, Desktop Commander, Cloudflare MCP
Servidores de la misma categoría hechos por otros creadores, ordenados por estrellas.
Cómo leer una tarjeta:★ Estrellas de GitHubiTres pilares: reseñas · revisiones · usoiActualizado = fecha en que se publicó la última versión en el registroi
Connect AI assistants to GitHub - manage repos, issues, PRs, and workflows through natural language.
MCP server for terminal commands, file operations, and process management
Cloudflare MCP servers
Copiado tal cual del repo del creador. PopNx MCP no lo ha revisado ni lo respalda. Se quitaron HTML, imágenes y scripts. Los enlaces llevan fuera de este sitio.
An inventory of the cryptography a JavaScript or TypeScript codebase uses, and which of it a large quantum computer would break.
An inventory and pointers, not a compliance certificate.
pqc-scan reads your source files, your package.json and lockfile, and any TLS (Transport Layer Security) configuration written down in the tree. It writes a report of every cryptographic algorithm it finds: through which interface (Web Crypto, node:crypto, a library), with what key size where the code says, where in the code, and a migration priority for each use a quantum computer would break. The report follows the discovery exercise the UK National Cyber Security Centre (NCSC) asks for in its post-quantum migration timeline: discovery and a plan by 2028, the highest-priority migrations by 2031, the rest by 2035.
It only reads. It runs nothing it scans, reads no environment variables, sends nothing anywhere, and has no telemetry. Zero dependencies; Node 20 or later. Free, and the command line stays free: a hosted report (trends over time, an organisation-wide view) may exist later, and would read the same JSON file; nothing of it is built.
pqc-scan is a command-line tool. You need two things on your computer:
node --version in a terminal (Command Prompt, PowerShell or Windows Terminal on Windows; Terminal on macOS or Linux). If the command is not found, or prints a version below 20, install the current "LTS" release from https://nodejs.org. On Windows the installer is a normal .msi; accept the defaults, then open a new terminal window so it picks up the change. npx @microtoll/pqc-scan path/to/your/app
or install it once and call it as pqc-scan from anywhere:
npm install -g @microtoll/pqc-scan
If you would rather read the code first, or work without the registry, take it from this repository in one of two ways:
C:\Tools\pqc-scan on Windows or ~/tools/pqc-scan elsewhere.sh git clone https://github.com/microtoll/pqc-scan.git There is nothing else to install: the scanner has no dependencies, so there is no npm install step inside it, and it needs no configuration, no account and no network connection to run.
Point it at the folder that holds your application's code (the one with its package.json).
On Windows (PowerShell or Command Prompt), with the scanner unzipped at C:\Tools\pqc-scan and your app at C:\Code\my-app:
C:\Tools\pqc-scan\pqc-scan C:\Code\my-app
On macOS or Linux:
node ~/tools/pqc-scan/bin/pqc-scan.mjs ~/code/my-app
It prints a short summary and writes two files into the folder you are standing in: pqc-scan.md, the report to read, and pqc-scan.json, the same for other tools. The summary tells you both paths:
pqc-scan: 6 quantum-vulnerable public-key uses in 5 files; 12 libraries; TLS groups not written down.
272 files read; 1 High, 5 Medium, 0 Low; 2 to check by hand. Report: C:\Users\you\pqc-scan.md (open it; read sections 1 and 3 first) JSON: C:\Users\you\pqc-scan.json
Open pqc-scan.md in any editor or Markdown viewer (Visual Studio Code shows it with *Ctrl+Shift+V*). Start with section 1, the summary, then section 3, the priorities, where each High or Medium says which line of your code it points at and why.
To put the reports somewhere else, name the files: --md C:\Reports\app.md and, if you want it, --json C:\Reports\app.json. If your tests are not in folders called test, tests, spec or __tests__, tell the scanner what their names contain, so the report can say how much of what it found is test code: --test-files self-check, repeatable.
A scan of a medium-sized application takes a few seconds. If you are in the scanner's own folder, . means "this folder", and --help prints every option:
node bin/pqc-scan.mjs --help
node bin/pqc-scan.mjs . --fail-on high # exit 1 if anything is High (for CI)
Things that catch people out:
"C:\My Projects\app"..js, .mjs, .cjs, .ts, .mts, .cts, .jsx, .tsx), lockfiles and TLS configuration files. Code in other languages is not seen (see "What it cannot see").The full set of options:
| Option | Meaning |
|---|---|
dir | The folder to scan (default: the current one). |
--write | Write pqc-scan.md and pqc-scan.json into the current folder: what a run from a terminal does by default. |
--md | Write the human report to this file instead. |
--json | Write the machine-readable report (schema version 1) to this file. |
--cbom | Write a CycloneDX 1.6 cryptographic bill of materials (CBOM) to this file. |
--test-files | Treat a file whose path contains this text as test code. Repeatable. |
--fail-on high | Exit 1 if anything is High. |
--fail-on medium | Exit 1 if anything is High or Medium. |
--exclude | Skip a directory or file name, or a path from dir. Repeatable. node_modules and .git are never read. |
In a script or a pipe, with none of --json, --md or --cbom, the Markdown goes to standard output and nothing is written to disk; at a terminal pqc-scan.md and pqc-scan.json are written instead (the bill of materials only when asked for). Exit codes: 0 done, 1 the --fail-on threshold was met, 2 a usage or read error. The threshold counts everything in the report, test code included; use --exclude to leave a directory out.
permissions:
contents: read steps:
The action runs the scanner on the workspace, adds the Markdown report to the job summary, uploads the two reports and the bill of materials as the pqc-scan artefact, and then, if fail-on is set and met, fails the job. Its inputs: path, fail-on, exclude (one name per line), json, md, cbom, job-summary, upload, artifact-name. Its outputs: high, medium, low, verdict, json, cbom. It needs Node 20 or later on the runner; GitHub's hosted runners have it.
crypto.subtle calls, with the algorithm read from the nearest literal, including the post-quantum and hybrid names (ML-KEM-768, ML-DSA-65, X25519MLKEM768, MLKEM768-X25519).tweetnacl, libsodium, the @noble family, jose, jsonwebtoken, bcrypt, crypto-js, web-push, openpgp, …), from package.json, package-lock.json, yarn.lock and pnpm-lock.yaml, with their calls where the catalogue maps them and JSON Web Token (JWT) algorithm names such as RS256.tls and https options; whether a hybrid post-quantum key-exchange group is enabled.A value it cannot read (an algorithm name held in a variable) is reported as dynamic — check by hand, never guessed.
Only public-key uses get a priority: a large quantum computer breaks RSA, elliptic-curve and Diffie–Hellman cryptography, but only weakens symmetric ciphers and hashes.
The priority is a heuristic from the words near the call (seal, encrypt, store… against session, tls, handshake…). A use with no telling word is High until a person checks. Every entry says which word decided it. A person should review each High.
pqc-scan.json: the inventory, described by schema/pqc-scan.schema.json (version 1). Paths are relative, with forward slashes; the scanned directory appears by name only; two scans of the same tree differ only in scannedAt.pqc-scan.md: a summary, the inventory, the priorities with a replacement to consider, notes on symmetric ciphers and hashes (AES-128, SHA-1, MD5, 3DES, ECB mode, weak PBKDF2 iteration counts), dependencies, TLS, next steps in the NCSC's three milestones, and what the scan cannot see.--cbom : the same inventory as a cryptographic bill of materials in the CycloneDX 1.6 format, the shape a central inventory, a supplier questionnaire or a viewer for these files expects.A CycloneDX 1.6 bill of materials lists each algorithm variant the scan found as a cryptographic-asset component (AES-256-GCM, ECDH P-256, ML-KEM-768…) with CycloneDX's own fields where the name settles them (the primitive, the key size or curve, the mode, the functions used, the classical and NIST post-quantum security levels) and an occurrence for every use: the file, the line, the interface and call, and the source line with its priority. The Web Crypto API, node:crypto and each catalogued package appear as library components that *provide* the assets found through them, with a package URL (pkg:npm/…) and version where the lockfile says. Each TLS configuration is a protocol asset that references its key-exchange groups, so X25519 in a server's group list and X25519 in a Web Crypto call are one asset. What CycloneDX has no field for (the scanner's class and kind, the migration priority and its reason, literal parameters such as an iteration count, the notes) travels in properties named microtoll:pqc-scan:*. The bill of materials says nothing the JSON report does not: it is written from the report alone, its serial number is worked out from its own contents (two scans of one tree give the same bill of materials apart from the timestamp and the serial number), and it keeps the sentence *an inventory and pointers, not a compliance certificate* in its metadata. The mapping is in DESIGN.md §8.14, and every output is checked against the published CycloneDX schema in the tests.
pqc-scan mcp runs the scanner as a Model Context Protocol (MCP) server over standard input and output, so a coding agent in Claude Code, Cursor or any other MCP host can check the cryptography it has just written. Add it to Claude Code, for every project, with:
claude mcp add pqc-scan --scope user -- npx -y @microtoll/pqc-scan mcp
Without --scope user, Claude Code registers the server for the folder the command was run in only. New conversations pick it up; one already open does not.
Any other host takes the same command (npx -y @microtoll/pqc-scan mcp) as a stdio server. It offers one tool:
| Tool | What it does |
|---|
| pqc_scan({ directory, exclude?, testFiles?, detail? }) | Scans the folder
Solo se muestra el principio; lee el resto en el repo