Back to skill

Security audit

Ambient Ride

Security checks for vulnerabilities and agentic risk

Overview

This ride-booking skill needs review because driver messages could steer the agent, and one deposit flow requests unlimited wallet spending permission.

Require host-level approval for bookings, payments and transfers, and keep driver messages from authorizing actions. Review unlimited MVL allowances separately from deposit amounts; use a dedicated wallet with limited funds. Protect ~/.amb because it contains wallet access material, and review notification destinations before sharing ride or chat details. Prefer a release that addresses the default driver-input trust boundary.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
  • Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
  • Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
  • Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Findings (1)

T09 · Insecure Skill Coding Practices

Warning
Location
scripts/ride-relay.js
Finding

Default legacy delivery exposes the agent to driver-chat prompt injection

Content
View full analysis

Impact

A malicious driver can inject apparent follow-up instructions or forged rider requests into the monitoring session, attempting to redirect the agent or induce unauthorized tool use, such as sending information back through driver chat. Success depends on the host model and tool-approval policy; no deterministic code execution, payment bypass, or successful exfiltration was demonstrated.

Vulnerability Patterns
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • System Prompt LeakageDirect Leakage, Indirect Extraction, Tool-Based Exfiltration
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • YARA SignaturesMalware Match, Webshell Match, Cryptominer Match
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (59)

YARA rule 'agent_skill_credential_exfiltration_webhook': AI agent skill credential harvesting followed by webhook or external exfiltration [agent_skills]

Critical
Category
YARA Match
Confidence
85% confidence
Finding

YARA rule matched a known malware signature (reverse shell, backdoor, ransomware, C2 framework, or info stealer).

Content

Scanner excerpt · scripts/ride-relay.js (reported line 19)May include surrounding context.

js
sion: "1.10.0" };

// src/scripts/ride-relay.ts
import path9 from "path";
import fs7 from "fs";
import { createHash as createHash3 } from "crypto";

// src/lib/core/state-paths.ts
import os from "os";
import path from "path";
function resolveDir(raw) {
  if (raw.startsWith("~/")) return path.join(os.homedir(), raw.slice(2));
  return path.resolve(raw);
}
function stateRoot() {
  return resolveDir(process.env["AMB_RIDE_STATE_DIR"] ?? path.join(os.homedir(), ".amb"));
}

// src/lib/core/runtime-env.ts
var SCALAR_KEYS = {
  AMB_RIDE_STATE_DIR: "stateDir",
  AMB_RIDE_PASSPHRASE: "passphrase",
  AMB_RIDE_LOG_LEVEL: "logLevel",
  AMB_RIDE_OPENCLAW_CLI: "openclawCli",
  AMB_RIDE_NOTIFY_WEBHOOK_URL: "notifyWebhookUrl",
  AMB_RIDE_NOTIFY_WEBHOOK_SECRET: "notifyWebhookSecret",
  AMB_RIDE_NOTIFY_TELEGRAM_BOT_TOKEN: "notifyTelegramBotToken",
  AMB_RIDE_NOTIFY_TELEGRAM_CHAT_ID: "notifyTelegramChatId",
  AMB_RIDE_NOTIFY_OWNER_KIND: "notifyOwnerKind",
  AMB_RIDE_NOTIFY_OWNER_REF: "notifyOwnerRef"
};

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
90% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · README.md (reported line 189)May include surrounding context.

md
- If the prebuilt binary cannot be downloaded, build from source: install `apt install build-essential python3` on Linux, Xcode Command Line Tools on macOS, or Visual Studio Build Tools on Windows, then re-run `install.js`.
- Or set an internal mirror: `npm config set better_sqlite3_binary_host_mirror <internal_mirror>` then retry.

For ABI mismatch (`Error: ... NODE_MODULE_VERSION`): make sure a supported Node.js version is active (`node --version`), then `rm -rf ~/.amb/cli && node <SKILL_DIR>/scripts/install.js` (re-clone + re-install). `~/.amb/cli` holds only the CLI itself — your database, keys, and passphrase live under `~/.amb/state/`, which this leaves untouched.

If `npm i -g @ambprotocol/ride-cli` fails with `EACCES`, npm's global prefix is not writable by your user. Do **not** re-run it under `sudo`: the CLI deliberately skips state initialisation when it detects a sudo invocation (it reports `status: "skipped_sudo"`), because root-owned `~/.amb` state would lock you out of y

Tool Parameter Abuse

High
Category
Tool Misuse
Confidence
90% confidence
Finding

Tool parameters are crafted to achieve unintended or unsafe behavior. Parameter abuse can bypass intended safety checks (e.g. shell=True, --force, dangerous glob patterns).

Content

Scanner excerpt · README.md (reported line 189)May include surrounding context.

md
- If the prebuilt binary cannot be downloaded, build from source: install `apt install build-essential python3` on Linux, Xcode Command Line Tools on macOS, or Visual Studio Build Tools on Windows, then re-run `install.js`.
- Or set an internal mirror: `npm config set better_sqlite3_binary_host_mirror <internal_mirror>` then retry.

For ABI mismatch (`Error: ... NODE_MODULE_VERSION`): make sure a supported Node.js version is active (`node --version`), then `rm -rf ~/.amb/cli && node <SKILL_DIR>/scripts/install.js` (re-clone + re-install). `~/.amb/cli` holds only the CLI itself — your database, keys, and passphrase live under `~/.amb/state/`, which this leaves untouched.

If `npm i -g @ambprotocol/ride-cli` fails with `EACCES`, npm's global prefix is not writable by your user. Do **not** re-run it under `sudo`: the CLI deliberately skips state initialisation when it detects a sudo invocation (it reports `status: "skipped_sudo"`), because root-owned `~/.amb` state would lock you out of y

Credential Access

High
Category
Privilege Escalation
Confidence
60% confidence
Finding

Code accesses credential files (SSH keys, AWS credentials, etc.). This could indicate credential theft attempts.

Content

Scanner excerpt · README.md (reported line 304)May include surrounding context.

md
└── state/
    ├── data/
    │   ├── ambient-ride.db   SQLite (wallet metadata, ride state, saved places, ...)
    │   └── .env              AMB_RIDE_PASSPHRASE (created/preserved by install.js)
    └── keys/
        └── <wallet_id>_private.enc, <wallet_id>_public.pem
                              Encrypted Privy keys (decryptable only with the passphrase)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied chunk contains only openclaw gateway restart. It performs gateway service management rather than any declared ride-hailing, wallet, or USDC bridge workflow. No context establishes this restart as a supporting implementation detail of those workflows.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied command performs skill maintenance rather than any stated ride-hailing, wallet, or USDC bridge workflow. Updating all skills has a materially different purpose and scope and is not shown to be a supporting implementation detail of the declared functionality.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The supplied chunk performs only a logout operation rather than any described ride, wallet, or bridge workflow. Session logout is an undeclared account-management action, and the chunk provides no context establishing it as a supporting step for a declared workflow.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
93% confidence
Finding

The supplied chunk only updates a plugin; it contains no implementation of the declared workflows. Although the plugin name matches the service, plugin updating is the chunk's sole behavior rather than a supporting step within a demonstrated declared workflow.

Content

No source excerpt is available for this finding.

Direct Prompt Extraction

High
Category
System Prompt Leakage
Confidence
85% confidence
Finding

Skill contains instructions that could directly expose system prompts, internal rules, or hidden instructions to users or external parties.

Content

Scanner excerpt · SKILL.md (reported line 68)May include surrounding context.

md
> `${SKILL_DIR}` is supplied by the harness at load time (see the "Base directory for this skill: …" line in the system message).
> Do NOT use `${SKILL_DIR}` outside these two cases.

> **⚠️ URL display rule (strictly enforced)**
> When any command returns a URL (e.g. `auth_url`), ALWAYS display it as a markdown hyperlink — NEVER as raw text.
> Format: `[Open link](url)` or `[Authenticate here](url)`
> Raw URLs wrap in the terminal and cannot be copied correctly.

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 80)May include surrounding context.

md
- Crypto ride spend-policy limits and remaining budgets (`amb policy`; see `references/spend-policy.md`)

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 88)May include surrounding context.

md
1. Read `references/spend-policy.md` and refresh the selected wallet's policy. Explain every blocking axis and the observed cap, fare and rolling spent/remainin

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 137)May include surrounding context.

md
mand-level details, keep using the per-feature references below — `wallet.md`, `ride.md`, `tip.md`, `chat.md`.)

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 366)May include surrounding context.

md
ment forms for `ride-search`/`ride-request`/`ride-status`/`ride-cancel` are in `references/ride.md` (`## Ride` → Member mode / Crypto mode). Run `amb whoami` to

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 370)May include surrounding context.

md
**Member coupons:** follow `references/ride.md` → **Member coupons** for the canonical sequence. In brief: proactively find a coupon unless the user supplied a

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 460)May include surrounding context.

md
For the full event schema, reconnect procedure, exit-code semantics, and related details, see the ride-relay section of `references/ride.md`.

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 511)May include surrounding context.

md
ternally. For card details (fields, expiry semantics, is_default meaning), see `references/ride.md` → "Card resolution".

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 526)May include surrounding context.

md
For the full command signatures, field reference, and resolve→ride flow, see `references/ride.md` → "Member place search".

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 610)May include surrounding context.

md
`references/ride.md` → "Ride History".

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 615)May include surrounding context.

md
- Place search, saved places (local catalog + Step 0 lookup), ride booking/status/cancellation/payment, ride history → `references/ride.md`

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 622)May include surrounding context.

md
orrective action. Error code → action table: see the Error Handling section in `references/ride.md`. For an unknown error in wallet mode during a crypto ride, r

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 137)May include surrounding context.

md
details, keep using the per-feature references below — `wallet.md`, `ride.md`, `tip.md`, `chat.md`.)

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 617)May include surrounding context.

md
- Tip configuration and payment → `references/tip.md`

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 137)May include surrounding context.

md
tes. (For command-level details, keep using the per-feature references below — `wallet.md`, `ride.md`, `tip.md`, `chat.md`.)

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 213)May include surrounding context.

md
in [Wallet reference](references/wallet.md), with the Agent running every command. If a MetaMask

Incomplete referenced artifact analysis

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 261)May include surrounding context.

md
For full wallet details (signing strategy, SIWE auth, phone verification, collateral management), read `references/wallet.md`.