registry / AI-agent capability abuse / @kryptosai/mcp-observatory / 1.47.0

@kryptosai/mcp-observatory@1.47.0

MCP security scanner and CI gate. Test, secure, and monitor MCP servers with attack simulation, schema drift detection, health scoring, and SARIF before agents depend on them.

AI Security Review

scanned 14d ago · by lpm-firewall-ai

Source confirms an explicit command that modifies MCP client configuration and wraps server startup, so this merits a capability warning. The install hook’s workflow creation is separately gated by an explicit opt-in and its other early-exit guards.

Static reason

One or more suspicious static signals were detected.; source matched previously finalized malicious package; routed for review; previous stored version diff introduced dangerous source

Trigger

A user runs `mcp-observatory protect`, with no file argument to target discovered configs or a file path to target one config.

Impact

MCP server launch behavior can change until the configuration is restored or edited; no attack or credential exposure was confirmed.

Mechanism

The command wraps configured server entries and saves an `.observatory.bak` copy; the wrapper blocks startup without a passing local receipt unless `--allow-fail` is used.

AI rationale

Evidence

dist/src/handshake.jsdist/src/commands/handshake.js

Decision evidence

public snapshot

AI called this Suspicious at 91.0% confidence as Dangerous Capability with medium false-positive risk.

Evidence for warning

The postinstall hook runs a CI setup helper, with disable, self-install, and CI guards.
Automatic workflow creation requires an explicit environment or project configuration opt-in; the file is created only if it does not already exist.
The explicit `protect` command can discover MCP client configs and rewrite server entries to start through the package wrapper, keeping a backup.
The wrapper checks for a passing local receipt before starting the server unless its caller uses `--allow-fail`.
The `protect` command is user-invoked and supports restoring its backup with `--undo`.

Evidence against

No unconsented install-time mutation of agent configuration or confirmed credential exfiltration was established.

Behavioral surface

Source

ChildProcessCryptoDynamicRequireEnvironmentVarsFilesystemNetworkShell

Supply chain

HighEntropyStringsUrlStrings

Source & flagged code

10 flagged

package.jsonView file

Lines 85-125json

85 "docs/target-registry.md",

86 "docs/tool-call-receipts.md",

87 "examples/demo-mcp-server.mjs",

88 "examples/insecure-mcp-server",

89 "scripts/postinstall.mjs",

90 "schemas",

91 "docs/usage-learning.md",

92 "examples/usage-learning.mjs"

93 ],

94 "scripts": {

95 "build": "tsc -p tsconfig.json",

96 "cli": "tsx src/cli.ts",

97 "dashboard:build": "tsx scripts/build-dashboard.ts",

98 "dashboard:validate": "node scripts/validate-dashboard.mjs",

99 "integration:real": "tsx scripts/run-real-server-matrix.ts",

100 "evaluate:permission-delta": "tsx scripts/evaluate-permission-delta-conformance.ts",

101 "evaluate:permission-delta:mcp-diff": "tsx scripts/run-mcp-diff-baseline.ts",

102 "safety-index": "tsx scripts/run-safety-index.ts",

103 "lint": "eslint .",

104 "prepack": "npm run build",

105 "postinstall": "node scripts/postinstall.mjs",

High

Install Time Lifecycle Scripts

Package defines install-time lifecycle scripts.

package.jsonView on unpkg · L105

Medium

Ambiguous Install Lifecycle Script

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

package.jsonView on unpkg · L105

106 "proof:refresh": "tsx scripts/refresh-proof-artifacts.ts",

107 "release:prep": "node scripts/release.mjs",

108 "certification:pr-body": "tsx scripts/print-certification-pr-body.ts",

109 "typecheck": "tsc --noEmit -p tsconfig.json",

110 "test": "vitest run",

111 "test:permission-delta-model": "vitest run tests/permission-delta-model.test.ts",

112 "test:coverage": "vitest run --coverage",

113 "validate:artifacts": "tsx scripts/validate-artifacts.ts",

114 "verify:packed-install": "node scripts/verify-packed-install.mjs",

115 "smoke": "npm run cli -- run --target tests/fixtures/sample-target-config.json && npm run cli -- diff tests/fixtures/sample-run-a.json tests/fixtures/sample-run-b.json",

116 "boundary:check": "node scripts/check-repository-boundary.mjs"

117 },

118 "keywords": [

119 "mcp",

120 "mcp-server",

121 "model-context-protocol",

122 "ai-agent",

123 "agent-security",

124 "ai-supply-chain",

125 "ai-tools",

dist/src/ci-issue.jsView file

Lines 1-21javascript

1import { execFile } from "node:child_process";

2import { mkdtemp, rm, writeFile } from "node:fs/promises";

High

Child Process

Package source references child process execution.

dist/src/ci-issue.jsView on unpkg · L1

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

dist/src/commands/demo.jsView file

Lines 1-31javascript

1import { createRequire } from "node:module";

2import path from "node:path";

3import { fileURLToPath } from "node:url";

4import { runTarget } from "../index.js";

dist/src/commands/test.jsView file

Lines 222-262javascript

222 }

223 else {

224 await maybeConvertPassingCheckToCi({

225 artifact,

226 target,

227 targetPath: options.target,

228 setupCi: conversionFlags.setupCi,

229 yes: conversionFlags.yes,

230

dist/src/checks/skill-scan.js#virtual:normalized:round1View file

•stage = ast_semantic_analysis; reason = ast_parse_error; limitedFiles = 6

High

Semantic Analysis Limited

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

dist/src/checks/skill-scan.js#virtual:normalized:round1View on unpkg

dist/src/commands/helpers.jsView file

•matchType = normalized_sha256 matchedPackage = @kryptosai/mcp-observatory@1.36.4 matchedPath = dist/src/commands/helpers.js matchedIdentity = npm:[redacted]:1.36.4 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.

dist/src/commands/helpers.jsView on unpkg

dist/src/risk-graph.jsView file

•matchType = normalized_sha256 matchedPackage = @kryptosai/mcp-observatory@1.36.4 matchedPath = dist/src/risk-graph.js matchedIdentity = npm:[redacted]:1.36.4 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.

dist/src/risk-graph.jsView on unpkg

dist/src/utils/failure-diagnosis.jsView file

•matchType = normalized_sha256 matchedPackage = @kryptosai/mcp-observatory@1.36.4 matchedPath = dist/src/utils/failure-diagnosis.js matchedIdentity = npm:[redacted]:1.36.4 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.

dist/src/utils/failure-diagnosis.jsView on unpkg

dist/src/usage/review.jsView file

•matchType = previous_version_dangerous_delta matchedPackage = @kryptosai/mcp-observatory@1.45.5 matchedIdentity = npm:[redacted]:1.45.5 similarity = 0.908 summary = stored previous version shares package body but lacks this dangerous source file

Critical

Previous Version Dangerous Delta

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

dist/src/usage/review.jsView on unpkg

Source & flagged code

10 flagged · loading source

package.jsonView file

•scripts.postinstall = node scripts/postinstall.mjs

High

Install Time Lifecycle Scripts

Package defines install-time lifecycle scripts.

package.jsonView on unpkg

•scripts.postinstall = node scripts/postinstall.mjs

Medium

Ambiguous Install Lifecycle Script

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

package.jsonView on unpkg

dist/src/ci-issue.jsView file

1import { execFile } from "node:child_process"; L2: import { mkdtemp, rm, writeFile } from "node:fs/promises";

High

Child Process

Package source references child process execution.

dist/src/ci-issue.jsView on unpkg · L1

Findings

1 Critical8 High5 Medium4 Low

CriticalPrevious Version Dangerous Deltadist/src/usage/review.js

HighInstall Time Lifecycle Scriptspackage.json

HighChild Processdist/src/ci-issue.js

HighShell

HighRuntime Package Installdist/src/commands/test.js

HighSemantic Analysis Limiteddist/src/checks/skill-scan.js#virtual:normalized:round1

HighKnown Malware Source Similaritydist/src/commands/helpers.js

HighKnown Malware Source Similaritydist/src/risk-graph.js

HighKnown Malware Source Similaritydist/src/utils/failure-diagnosis.js

MediumAmbiguous Install Lifecycle Scriptpackage.json

MediumDynamic Requiredist/src/commands/demo.js

Affected versions and remediation

This report applies to @kryptosai/mcp-observatory@1.47.0.

See all versions of @kryptosai/mcp-observatory for other recorded verdicts.

Review the evidence and your use of @kryptosai/mcp-observatory@1.47.0 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 16:54:17.963Z (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