registry / @raditia/craftkit / 1.58.0

@raditia/craftkit@1.58.0

AI coding skills that auto-sync across Claude Code, Cursor, Gemini CLI, and Codex CLI

AI Security Review

scanned 14d ago · by lpm-firewall-ai

The automatic postinstall can set up CraftKit’s own Codex hooks and alter Codex configuration. Under the agent integration policy, classify this package-owned setup as a warning rather than a confirmed attack.

Static reason

One or more suspicious static signals were detected.

Trigger

Installing the package through npm, which runs its postinstall script.

Impact

This changes Codex behavior by enabling CraftKit hooks. The inspected evidence supports a package-owned agent integration lifecycle risk, not a confirmed attack.

Mechanism

The sync path copies hook scripts into the Codex hooks directory and adds hook commands to the Codex configuration for several events.

AI rationale

Evidence

package.jsonscripts/postinstall.shadapters/codex.shhooks/craftkit-codex.js

Decision evidence

public snapshot

AI called this Suspicious at 85.0% confidence as Dangerous Capability with medium false-positive risk.

Evidence for warning

The manifest runs a postinstall script that invokes the sync installer.
The installer copies CraftKit hooks into Codex’s hooks directory and registers them for multiple Codex events in the user’s hooks configuration.
The Codex hook contains guarded behavior that checks whether an agent profile is CraftKit-managed and read-only.

Evidence against

The Codex installer describes review and trust of changed hook definitions and preserves unrelated hooks.
The context source restricts repository remotes to GitHub and rejects credential-bearing URLs.

Behavioral surface

Source

ChildProcessCryptoDynamicRequireEnvironmentVarsFilesystemNetwork

Supply chain

HighEntropyStringsUrlStrings

ManifestNo manifest risk signals triggered.

scanned 17 file(s), 121 KB of source, external domains: github.com, registry.npmjs.org

Source & flagged code

6 flagged

package.jsonView file

Lines 1-26json

2 "name": "@raditia/craftkit",

3 "version": "1.58.0",

4 "description": "AI coding skills that auto-sync across Claude Code, Cursor, Gemini CLI, and Codex CLI",

5 "scripts": {

6 "postinstall": "bash scripts/postinstall.sh"

High

Install Time Lifecycle Scripts

Package defines install-time lifecycle scripts.

package.jsonView on unpkg · L6

Medium

Ambiguous Install Lifecycle Script

Install-time lifecycle script is not statically allowlisted and needs review.

package.jsonView on unpkg · L6

7 },

8 "engines": {

9 "os": ["darwin", "linux"]

10 },

11 "files": [

12 "rules/",

13 "skills/",

14 "commands/",

15 "agents/",

16 "partials/",

17 "adapters/",

18 "hooks/",

19 "scripts/",

20 "sync.sh",

21 "install.sh",

22 "CHANGELOG.md"

23 ],

24 "keywords": [

25 "claude-code",

26 "cursor",

hooks/craftkit-read-cap.jsView file

Lines 1-33javascript

1#!/usr/bin/env node

2// CraftKit PreToolUse rewrite: an uncapped whole-file read gets a line cap.

3// `rtk hook claude` already rewrites `cat F` to `rtk read F`, but passes no flags at any

4// size, so a 2000-line file still lands whole and is re-sent every turn after. This adds

5// the cap rtk already supports. It rewrites, it never refuses: nothing here can block a

6// call, and a skipped rewrite leaves today's behavior.

7// `-m` and not `-l`: the `-l` filters strip comments and nothing else, measured 10577 to

8// 5506 bytes on craftkit-routing.js and a no-op on markdown, so they delete the `why`

9// comments plus every ponytail:/flag: marker that karpathy-guidelines makes contract.

hooks/craftkit-transcript.jsView file

Lines 4-44javascript

4// once: a disagreement between the two would make one of them fire on the wrong turn,

5// which is indistinguishable from a broken gate.

6// Installed alongside the gates by adapters/claude.sh; required by relative path.

8const crypto = require('crypto');

9const fs =

install.shView file

•path = install.sh kind = build_helper sizeBytes = 1104 magicHex = [redacted]

Medium

Ships Build Helper

Package ships non-JavaScript build or shell helper files.

install.shView on unpkg

hooks/craftkit-codex.js#virtual:normalized:round1View file

•stage = ast_semantic_analysis; reason = ast_parse_error; limitedFiles = 4

High

Semantic Analysis Limited

A bounded semantic-analysis stage reached its safety limit; remaining detectors completed, but this package requires AI review.

hooks/craftkit-codex.js#virtual:normalized:round1View on unpkg

Source & flagged code

6 flagged · loading source

package.jsonView file

•scripts.postinstall = bash scripts/postinstall.sh

High

Install Time Lifecycle Scripts

Package defines install-time lifecycle scripts.

package.jsonView on unpkg

•scripts.postinstall = bash scripts/postinstall.sh

Medium

Ambiguous Install Lifecycle Script

Install-time lifecycle script is not statically allowlisted and needs review.

package.jsonView on unpkg

hooks/craftkit-read-cap.jsView file

13L14: const path = require('path'); L15: const { BIG_LINES, overLines } = require(path.join(__dirname, 'craftkit-filesize.js'));

Medium

Dynamic Require

Package source references dynamic require/import behavior.

hooks/craftkit-read-cap.jsView on unpkg · L13

Findings

2 High6 Medium5 Low

HighInstall Time Lifecycle Scriptspackage.json

HighSemantic Analysis Limitedhooks/craftkit-codex.js#virtual:normalized:round1

MediumAmbiguous Install Lifecycle Scriptpackage.json

MediumDynamic Requirehooks/craftkit-read-cap.js

MediumNetwork

MediumEnvironment Vars

MediumShips Build Helperinstall.sh

MediumStructural Risk Force Deep Review

LowScripts Present

LowWeak Cryptohooks/craftkit-transcript.js

LowFilesystem

LowHigh Entropy Strings

LowUrl Strings

Affected versions and remediation

This report applies to @raditia/craftkit@1.58.0.

See all versions of @raditia/craftkit for other recorded verdicts.

Review the evidence and your use of @raditia/craftkit@1.58.0 before allowing it. Restrict the permissions described in this report.
Choose an independently verified alternative or release. This report does not establish that other versions are safe.

Evidence last updated: 2026-10-09 10:07:19.134Z (UTC).

Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Affected versions and remediation

Runtime surface

Behavioral surface

Source & flagged code