@slopus/rig@0.0.24

Opinionated coding-agent harness with strong defaults for provider-aligned workflows.

AI Security Review

scanned 3mo ago · by lpm-firewall-ai

Source inspection shows a user-invoked coding-agent CLI with explicit permission and MCP-trust controls. No lifecycle execution, stealth persistence, credential harvesting, or unconsented exfiltration chain was found.

Static reason

No blocking static signals were detected.

Trigger

User runs rig, submits an agent task, or authorizes a tool action.

Impact

Intended agent capabilities may access approved workspace data or network resources; no covert collection, exfiltration, install-time execution, or foreign agent-control mutation was found.

Mechanism

Interactive coding-agent execution with guarded filesystem, shell, MCP, and web tools.

AI rationale

Evidence

package.jsondist/main.jsdist/agent/context/createSensitiveReadPaths.jsdist/agent/context/assertCanReadPath.jsdist/agent/context/assertCanWritePath.jsdist/mcp/McpClientManager.jsdist/tools/claude/WebFetch.jsdist/providers/requestAnthropicMessage.js

Network endpoints2

api.anthropic.comchatgpt.com/backend-api

Decision evidence

public snapshot

AI called this Clean at 94.0% confidence as Benign with low false-positive risk.

Evidence for AI clean decision

Evidence against

package.json has no preinstall/install/postinstall lifecycle hook.
dist/main.js only starts after the user invokes the rig CLI.
dist/agent/context/assertCanReadPath.js blocks listed credential paths outside the workspace unless full_access is selected.
dist/tools/claude/WebFetch.js requires auto/full access and marks network fetches for review.
dist/mcp/McpClientManager.js records MCP trust only after a trust decision.

Behavioral surface

Source

ChildProcessCryptoEnvironmentVarsFilesystemNetworkShell

Supply chain

HighEntropyStringsMinifiedUrlStrings

ManifestNo manifest risk signals triggered.

scanned 700 file(s), 1.65 MB of source, external domains: api.anthropic.com, api.kimi.com, api.openai.com, auth.kimi.com, auth.x.ai, chatgpt.com, cli-chat-proxy.grok.com, rig.dev, support.anthropic.com

Source & flagged code

1 flagged

dist/images/prepareImageForClaude.jsView file

Lines 1-25javascript

1import { createHash } from "node:crypto";

2import { LRUCache } from "lru-cache";

3import { getImageProcessor } from "./getImageProcessor.js";

4import { ImageProcessingError } from "./ImageProcessingError.js";

5import { MAX_PROMPT_IMAGE_INPUT_BYTES, } from "./prepareImageForPrompt.js";

6export const CLAUDE_IMAGE_MAX_BASE64_BYTES = 5 * 1024 * 1024;

7export const CLAUDE_IMAGE_MAX_RAW_BYTES = (CLAUDE_IMAGE_MAX_BASE64_BYTES * 3) / 4;

Low

Weak Crypto

Package source references weak cryptographic algorithms.

dist/images/prepareImageForClaude.jsView on unpkg · L5

8export const CLAUDE_IMAGE_MAX_DIMENSION = 2000;

9const JPEG_QUALITIES = [80, 60, 40, 20];

10const IMAGE_CACHE = new LRUCache({

11 max: 32,

12 maxSize: 64 * 1024 * 1024,

13 sizeCalculation: (image) => image.bytes.length,

14});

15/** Prepare an image using the same size and dimension policy as Claude Code. */

16export async function prepareImageForClaude(bytes) {

17 const input = Buffer.from(bytes.buffer, bytes.byteOffset, bytes.byteLength);

18 if (input.length === 0) {

19 throw new ImageProcessingError("Image file is empty.");

20 }

21 if (input.length > MAX_PROMPT_IMAGE_INPUT_BYTES) {

22 throw new ImageProcessingError("Image exceeds the supported size limit.");

23 }

24 const cacheKey = createHash("sha1").update(input).update("claude").digest("hex");

25 const cached = IMAGE_CACHE.get(cacheKey);

Source & flagged code

1 flagged · loading source

dist/images/prepareImageForClaude.jsView file

5import { MAX_PROMPT_IMAGE_INPUT_BYTES, } from "./prepareImageForPrompt.js"; L6: export const CLAUDE_IMAGE_MAX_BASE64_BYTES = 5 * 1024 * 1024; L7: export const CLAUDE_IMAGE_MAX_RAW_BYTES = (CLAUDE_IMAGE_MAX_BASE64_BYTES * 3) / 4;

Low

Weak Crypto

Package source references weak cryptographic algorithms.

dist/images/prepareImageForClaude.jsView on unpkg · L5

Findings

2 Medium5 Low

MediumNetwork

MediumEnvironment Vars

LowScripts Present

LowWeak Cryptodist/images/prepareImageForClaude.js

LowFilesystem

LowHigh Entropy Strings

LowUrl Strings

Affected versions and remediation

This report applies to @slopus/rig@0.0.24.

See all versions of @slopus/rig for other recorded verdicts.

Review the evidence and your use of @slopus/rig@0.0.24 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: 2026-08-02 09:14:53.028Z (UTC).

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