registry / novahiz / 0.3.8

novahiz@0.3.8

Enforce AI agent skills before every file edit: prompt classifier, skill catalog, and pre-edit gate.

AI Security Review

scanned 8d ago · by lpm-firewall-ai

Review flagged AI-agent configuration or capability changes. This remains warn-only unless evidence shows foreign-agent hijack through preinstall/install/postinstall, hidden persistence, exfiltration, remote code execution, or other concrete malicious behavior.

Static reason

High-risk behavior combination matched malicious policy.; source matched previously finalized malicious package; routed for review; previous stored version diff introduced dangerous source

Trigger

Running explicit setup without --dry-run, or invoking install/hooks.mjs directly.

Impact

Installed hooks can govern later agent tool use. Interactive setup requests consent; npm postinstall uses dry-run mode and does not activate these writes.

Mechanism

Setup invokes the hook installer, which merges command hooks into agent configuration and calls agent CLIs to register Novahiz MCP tools.

AI rationale

Concrete source shows explicit agent configuration mutation, which the supplied policy classifies as warning-worthy. The dry-run lifecycle gate and interactive confirmation do not support a malicious verdict.

Evidence

install/install.mjsinstall/hooks.mjssettings.jsonhooks.json

Decision evidence

public snapshot

AI called this Suspicious at 94.0% confidence as Dangerous Capability with medium false-positive risk.

Evidence for warning

Explicit setup invokes the agent hook installer only outside dry-run mode.
The hook installer targets Claude settings.json and Codex hooks.json.
The hook installer merges agent hooks and registers the local Novahiz MCP server.
Interactive setup asks for confirmation and exits if declined.
The npm postinstall explicitly runs the installer in dry-run mode.

Evidence against

Inspected configuration writes are guarded during npm postinstall; no automatic install-time control hijack was established.
No concrete credential exfiltration or malicious remote payload execution was identified in the inspected behavior.

Behavioral surface

Source

ChildProcessDynamicRequireEnvironmentVarsFilesystemNetwork

Supply chain

HighEntropyStringsUrlStrings

ManifestNo manifest risk signals triggered.

scanned 87 file(s), 791 KB of source, external domains: github.com, nodejs.org, opencode.ai

Source & flagged code

14 flagged

package.jsonView file

Lines 35-75json

35 "novahiz-uninstall": "install/uninstall.mjs",

36 "novahiz-mcp": "mcp/novahiz-tools/index.mjs"

37 },

38 "engines": {

39 "node": ">=22.18.0"

40 },

41 "scripts": {

42 "check": "node --experimental-strip-types src/cli.ts check",

43 "sync": "node --experimental-strip-types src/cli.ts sync",

44 "classify": "node --experimental-strip-types src/cli.ts classify",

45 "catalog": "node --experimental-strip-types src/cli.ts catalog",

46 "report": "node --experimental-strip-types src/cli.ts report",

47 "setup": "node install/install.mjs",

48 "setup:dry": "node install/install.mjs --dry-run",

49 "bootstrap": "node install/bootstrap.mjs",

50 "bootstrap:dry": "node install/bootstrap.mjs --dry-run",

51 "test": "node --experimental-strip-types --test",

52 "ci:local": "node scripts/ci-local.mjs",

53 "typecheck": "tsc --noEmit",

54 "prepare": "tsc --noEmit && tsc -p tsconfig.build.json",

55 "postinstall": "node install/install.mjs --dry-run || exit 0"

High

Install Time Lifecycle Scripts

Package defines install-time lifecycle scripts.

package.jsonView on unpkg · L55

Medium

Ambiguous Install Lifecycle Script

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

package.jsonView on unpkg · L55

56 },

57 "files": [

58 "src",

59 "dist",

60 "catalog",

61 "adapters",

62 "bin",

63 "install",

64 "mcp",

65 "skills",

66 "docs",

67 "!docs/audit-2026-09-25.md",

68 "README.md",

69 "NOTICE.md",

70 "novahiz.config.example.json",

71 "tsconfig.json",

72 "LICENSE"

73 ],

74 "devDependencies": {

75 "@opencode-ai/plugin": "1.18.30",

bin/novahiz.mjsView file

Lines 14-36javascript

14 );

15 process.exit(1);

16}

18const here = dirname(fileURLToPath(import.meta.url));

19const

install/bootstrap.mjsView file

Lines 1-25javascript

1#!/usr/bin/env node

3import { existsSync, mkdirSync, readFileSync, writeFileSync } from "node:fs";

4import { homedir } from "node:os";

5import { join, resolve } from "node:path";

6import { spawnSync }

install/install.mjsView file

Lines 1-21javascript

1#!/usr/bin/env node

2import { cpSync, existsSync, mkdirSync, readdirSync, statSync, writeFileSync } from "node:fs";

3import { homedir } from "node:os";

Critical

Ai Agent Control Hijack

Source creates an unconsented AI-agent control surface through install-time mutation or a default unauthenticated remote skill channel.

install/install.mjsView on unpkg · L1

4import { join, resolve } from

install/hooks.mjsView file

•Runtime or CLI code links a write operation to an explicit external AI-agent control path: readfilesync(path, "utf8").replace(/^\ufeff/, "")); } catch (error) { process.stderr.write(`refus: ${path} n'est pas un json valide (${error.message}). rien ecrit.\n`); return false; } const backup = `${path}.novahiz-bak`; if (!existssync(backup)) cpsync(path, backup); tracked.backups.push({ path, backup }); } else { tracked.created.push(path); } mkdirsync(dirname(path), { recursive: true }); writefilesync(path, `${json.stringify(mergehooks(base, generated), null, 2)}\n`, "utf8"); return true; } function registermcp(name, args, label) { if (!which(name)) { process.stdout.write(` ${label}: cli '${name}' introuvable, enregistrement mcp ignore\n`);

High

External Ai Agent Control Surface Mutation

Runtime or CLI source writes behavior-bearing configuration into a user or project AI-agent control surface.

install/hooks.mjsView on unpkg

install/install.shView file

•path = install/install.sh kind = build_helper sizeBytes = 124 magicHex = [redacted]

Medium

Ships Build Helper

Package ships non-JavaScript build or shell helper files.

install/install.shView on unpkg

adapters/opencode/novahiz.ts#virtual:normalized:round1View file

•stage = ast_semantic_analysis; reason = ast_parse_error; limitedFiles = 1

High

Semantic Analysis Limited

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

adapters/opencode/novahiz.ts#virtual:normalized:round1View on unpkg

skills/llm-threat-review/scripts/llm_risk_lint.mjsView file

•matchType = normalized_sha256 matchedPackage = novahiz@0.2.5 matchedPath = skills/llm-threat-review/scripts/llm_risk_lint.mjs matchedIdentity = npm:bm92YWhpeg:0.2.5 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

skills/llm-threat-review/scripts/llm_risk_lint.mjsView on unpkg

skills/secrets-hygiene/scripts/secret_scan.mjsView file

•matchType = normalized_sha256 matchedPackage = novahiz@0.2.5 matchedPath = skills/secrets-hygiene/scripts/secret_scan.mjs matchedIdentity = npm:bm92YWhpeg:0.2.5 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

skills/secrets-hygiene/scripts/secret_scan.mjsView on unpkg

skills/skill-authoring/scripts/skill_frontmatter_lint.mjsView file

•matchType = normalized_sha256 matchedPackage = novahiz@0.2.5 matchedPath = skills/sk[redacted]/scripts/sk[redacted] matchedIdentity = npm:bm92YWhpeg:0.2.5 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

skills/skill-authoring/scripts/skill_frontmatter_lint.mjsView on unpkg

skills/skill-eval-loop/scripts/skill_structure_check.mjsView file

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

skills/skill-eval-loop/scripts/skill_structure_check.mjsView on unpkg

skills/ui-slop-remover/scripts/slop_lint.mjsView file

•matchType = normalized_sha256 matchedPackage = novahiz@0.2.5 matchedPath = skills/ui-slop-remover/scripts/slop_lint.mjs matchedIdentity = npm:bm92YWhpeg:0.2.5 similarity = 1.000 summary = normalized source hash matched finalized malicious source

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

skills/ui-slop-remover/scripts/slop_lint.mjsView on unpkg

Source & flagged code

14 flagged · loading source

package.jsonView file

•scripts.postinstall = node install/install.mjs --dry-run || exit 0

High

Install Time Lifecycle Scripts

Package defines install-time lifecycle scripts.

package.jsonView on unpkg

•scripts.postinstall = node install/install.mjs --dry-run || exit 0

Medium

Ambiguous Install Lifecycle Script

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

package.jsonView on unpkg

bin/novahiz.mjsView file

34L35: await import(pathToFileURL(target).href);

Medium

Dynamic Require

Package source references dynamic require/import behavior.

bin/novahiz.mjsView on unpkg · L34

install/bootstrap.mjsView file

5import { join, resolve } from "node:path"; L6: import { spawnSync } from "node:child_process"; L7: import { fileURLToPath } from "node:url"; ... L9: L10: const REPO_URL = "https://github.com/novahiz/novahiz.git"; L11: // Was hardcoded to ~/.config/novahiz, ignoring NOVAHIZ_HOME/NOVAHIZ_HOME. ... L15: function log(msg) { L16: process.stdout.write(`[Novahiz] ${msg}\n`); L17: } ... L24: // ENOENT (post-CVE-2024-* Node refuses to run .cmd via CreateProcess). The L25: // line handed to cmd.exe only contains tokens accepted by unsafeHostToken L26: // (lib.mjs — the install-side mirror of src/exec.ts SAFE_TOKEN, widened by

High

Sandbox Evasion Gated Capability

Source gates dangerous network, credential, or execution behavior behind CI, host, platform, time, or geo fingerprint checks.

install/bootstrap.mjsView on unpkg · L5

•matchType = previous_version_dangerous_delta matchedPackage = novahiz@0.3.7 matchedIdentity = npm:bm92YWhpeg:0.3.7 similarity = 0.802 summary = stored previous version shares package body but lacks this dangerous source file

High

Previous Version Dangerous Delta

This package version adds a dangerous source file absent from the previous stored version; route for source-aware review.

install/bootstrap.mjsView on unpkg

install/install.mjsView file

1#!/usr/bin/env node L2: import { cpSync, existsSync, mkdirSync, readdirSync, statSync, writeFileSync } from "node:fs"; L3: import { homedir } from "node:os"; ... L51: const home = NovahizHome(flags); L52: const configDir = flags.scope === "project" ? resolve(".opencode") : opencodeConfigDir(); L53: const skillsDir = join(configDir, "skills"); ... L72: L73: const claudeDir = process.env.CLAUDE_CONFIG_DIR || join(homedir(), ".claude"); L74: const codexDir = process.env.CODEX_HOME || join(homedir(), ".codex"); L75: const harnessDirs = { opencode: configDir, claude: claudeDir, codex: codexDir }; ... L176: for (const name of configured) { L177: if (!existsSync(harnessDirs[name])) mkdirSync(harnessDirs[name], { recursive: true });

Critical

Ai Agent Control Hijack

Source creates an unconsented AI-agent control surface through install-time mutation or a default unauthenticated remote skill channel.

install/install.mjsView on unpkg · L1

install/hooks.mjsView file

High

External Ai Agent Control Surface Mutation

Runtime or CLI source writes behavior-bearing configuration into a user or project AI-agent control surface.

install/hooks.mjsView on unpkg

install/install.shView file

•path = install/install.sh kind = build_helper sizeBytes = 124 magicHex = [redacted]

Medium

Ships Build Helper

Package ships non-JavaScript build or shell helper files.

install/install.shView on unpkg

adapters/opencode/novahiz.ts#virtual:normalized:round1View file

•stage = ast_semantic_analysis; reason = ast_parse_error; limitedFiles = 1

High

Semantic Analysis Limited

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

adapters/opencode/novahiz.ts#virtual:normalized:round1View on unpkg

skills/llm-threat-review/scripts/llm_risk_lint.mjsView file

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

skills/llm-threat-review/scripts/llm_risk_lint.mjsView on unpkg

skills/secrets-hygiene/scripts/secret_scan.mjsView file

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

skills/secrets-hygiene/scripts/secret_scan.mjsView on unpkg

skills/skill-authoring/scripts/skill_frontmatter_lint.mjsView file

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

skills/skill-authoring/scripts/skill_frontmatter_lint.mjsView on unpkg

skills/skill-eval-loop/scripts/skill_structure_check.mjsView file

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

skills/skill-eval-loop/scripts/skill_structure_check.mjsView on unpkg

skills/ui-slop-remover/scripts/slop_lint.mjsView file

High

Known Malware Source Similarity

Source file is highly similar to a previously finalized malicious package; route for source-aware review.

skills/ui-slop-remover/scripts/slop_lint.mjsView on unpkg

Findings

1 Critical10 High6 Medium5 Low

CriticalAi Agent Control Hijackinstall/install.mjs

HighInstall Time Lifecycle Scriptspackage.json

HighSandbox Evasion Gated Capabilityinstall/bootstrap.mjs

HighExternal Ai Agent Control Surface Mutationinstall/hooks.mjs

HighSemantic Analysis Limitedadapters/opencode/novahiz.ts#virtual:normalized:round1

HighKnown Malware Source Similarityskills/llm-threat-review/scripts/llm_risk_lint.mjs

HighKnown Malware Source Similarityskills/secrets-hygiene/scripts/secret_scan.mjs

HighKnown Malware Source Similarityskills/skill-authoring/scripts/skill_frontmatter_lint.mjs

HighKnown Malware Source Similarityskills/skill-eval-loop/scripts/skill_structure_check.mjs

HighKnown Malware Source Similarityskills/ui-slop-remover/scripts/slop_lint.mjs

HighPrevious Version Dangerous Deltainstall/bootstrap.mjs

MediumAmbiguous Install Lifecycle Scriptpackage.json

MediumDynamic Requirebin/novahiz.mjs

MediumNetwork

MediumEnvironment Vars

MediumShips Build Helperinstall/install.sh

MediumStructural Risk Force Deep Review

LowNon Install Lifecycle Scripts

LowScripts Present

LowFilesystem

LowHigh Entropy Strings

LowUrl Strings

Affected versions and remediation

This report applies to novahiz@0.3.8.

See version security history for other recorded verdicts.

Review the evidence and your use of novahiz@0.3.8 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: 2026-09-30 02:03:37.471Z (UTC).