Back to skill

Security audit

pontx-sendbird-chat-platform

Security checks for vulnerabilities and agentic risk

Overview

This skill is a disclosed Sendbird API helper that uses caller-owned credentials and requires preview plus confirmation before mutations, with a supply-chain caution around unpinned Pontx tooling.

Before installing, use a least-privilege Sendbird token where possible, keep the token out of logs and examples, pin and verify Pontx package versions, review dry-run output carefully, and require explicit confirmation for any mutation such as sending messages, deleting channels, blocking users, or scheduling announcements.

Vulnerability Patterns
  • Insecure DependenciesIntroduces malicious components through unsafe dependency sources
  • 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)

T08 · Insecure Dependencies

Warning
Location
SKILL.md:11
Finding

Unpinned Third-Party Tooling Handles Privileged Sendbird Credentials and Operations

Content
View full analysis
pontx-hub sdk sendbird-chat-platform ``` Use `@pontx/sendbird-chat-platform` in application code and `pontx-sendbird-chat-platform` for an optional local script. ``` ```markdown - The generated SDK attaches the Application API Token as the `api-token` request header. Provide it via the `SENDBIRD_API_TOKEN` environment variable; never log, print, or commit it, and never put it in a request example. ``` ```markdown ```bash pnpm exec pontx-sendbird-chat-platform list apis pnpm exec pontx-sendbird-chat-platform call --dry-run ``` ``` ### Technical Analysis The skill instructs users to discover API contracts and execute the `pontx-hub` and `pontx-sendbird-chat-platform` tooling without specifying exact package versions, integrity hashes, a lockfile, a verified source repository, or publisher-validation requirements. The generated SDK is trusted with `SENDBIRD_API_TOKEN` and can issue administrative requests involving users, channels, messages, moderation, bots, and announcements. Because dependency identity and contents are not pinned, a compromised publisher account, malicious package replacement, dependency-confusion event, or unexpectedly unsafe update could introduce attacker-controlled code. Such code would execute in the local package-manager context and could read environment variables, change request destinations or bodies, or make additional requests outside the documented preview workflow. The use of live Pontx contract discovery further makes operational behavior dependent on mutable external content rather than a locally reviewed and immutable API contract. The skill's preview-and-confirmation ...[truncated 1841 chars]
Remediation
View remediation
Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep