Back to skill

Security audit

LaunchDarkly

Security checks for vulnerabilities and agentic risk

Overview

The LaunchDarkly workflow has clear approval safeguards, but its setup instructions run downloaded software without a separate approval or verification step.

Review and approve CLI installation separately, including its bundled-skill changes, rather than letting a LaunchDarkly request trigger the download-and-run fallback. Use a verified installation and narrowly scoped LaunchDarkly credentials, and retain the required write and deletion approvals. OOMOL mediates account access; its hosted connector and remote installer implementations were not verified in this review.

Vulnerability Patterns
  • 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
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
Vulnerability Patterns
  • 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
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (11)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The supplied command installs software from Oomol rather than performing LaunchDarkly operations. Nothing in the code establishes this installation as a supporting dependency for the declared purpose. The downloaded script's contents and further effects are not available in the supplied chunk.

Content

No source excerpt is available for this finding.

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

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

Content

Scanner excerpt · SKILL.md (reported line 46)May include surrounding context.

md
- `create_project` — Create a LaunchDarkly project using either common fields or a full official request body. [write]
- `create_segment` — Create a LaunchDarkly segment using either common fields or a full official request body. [write]
- `create_team` — Create a LaunchDarkly team using either common fields or a full official request body. [write]
- `create_token` — Create a LaunchDarkly access token using either common fields or a full official request body. [write]
- `delete_environment` — Delete a LaunchDarkly environment by project key and environment key. [destructive]
- `delete_feature_flag` — Delete a LaunchDarkly feature flag by project key and feature flag key. [destructive]
- `delete_project` — Delete a LaunchDarkly project by project key. [destructive]

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

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

Content

Scanner excerpt · SKILL.md (reported line 76)May include surrounding context.

md
- `patch_project` — Patch a LaunchDarkly project with standard JSON Patch operations. [write]
- `patch_segment` — Patch a LaunchDarkly segment with JSON Patch, JSON Merge Patch, or semantic patch instructions. [write]
- `patch_team` — Patch a LaunchDarkly team with semantic patch instructions. [write]
- `patch_token` — Patch a LaunchDarkly access token with standard JSON Patch operations. [write]
- `reset_token` — Reset a LaunchDarkly access token value and optionally control when the old value expires. [destructive]
- `search_context_instances` — Search LaunchDarkly context instances in a project environment with filtering, sorting, and pagination.
- `search_contexts` — Search LaunchDarkly contexts in a project environment with filtering, sorting, and pagination.

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

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

Content

Scanner excerpt · SKILL.md (reported line 77)May include surrounding context.

md
- `patch_segment` — Patch a LaunchDarkly segment with JSON Patch, JSON Merge Patch, or semantic patch instructions. [write]
- `patch_team` — Patch a LaunchDarkly team with semantic patch instructions. [write]
- `patch_token` — Patch a LaunchDarkly access token with standard JSON Patch operations. [write]
- `reset_token` — Reset a LaunchDarkly access token value and optionally control when the old value expires. [destructive]
- `search_context_instances` — Search LaunchDarkly context instances in a project environment with filtering, sorting, and pagination.
- `search_contexts` — Search LaunchDarkly contexts in a project environment with filtering, sorting, and pagination.

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

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

Content

Scanner excerpt · SKILL.md (reported line 52)May include surrounding context.

md
- `delete_project` — Delete a LaunchDarkly project by project key. [destructive]
- `delete_segment` — Delete a LaunchDarkly segment by project key, environment key, and segment key. [destructive]
- `delete_team` — Delete a LaunchDarkly team by team key. [destructive]
- `delete_token` — Delete a LaunchDarkly access token by token identifier. [destructive]
- `get_caller_identity` — Get the LaunchDarkly caller identity for the current access token.
- `get_context_instances` — Get a LaunchDarkly context instance by project key, environment key, and context instance identifier.
- `get_contexts` — Get a LaunchDarkly context by context kind and key, with optional paging over related results.

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

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

Content

Scanner excerpt · SKILL.md (reported line 53)May include surrounding context.

md
- `delete_segment` — Delete a LaunchDarkly segment by project key, environment key, and segment key. [destructive]
- `delete_team` — Delete a LaunchDarkly team by team key. [destructive]
- `delete_token` — Delete a LaunchDarkly access token by token identifier. [destructive]
- `get_caller_identity` — Get the LaunchDarkly caller identity for the current access token.
- `get_context_instances` — Get a LaunchDarkly context instance by project key, environment key, and context instance identifier.
- `get_contexts` — Get a LaunchDarkly context by context kind and key, with optional paging over related results.
- `get_environment` — Get a LaunchDarkly environment by project key and environment key.

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

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

Content

Scanner excerpt · SKILL.md (reported line 67)May include surrounding context.

md
- `get_segments` — List LaunchDarkly segments in a project environment with optional filtering and pagination.
- `get_tags` — List LaunchDarkly tags with optional prefix and resource-kind filters.
- `get_team` — Get a LaunchDarkly team by team key.
- `get_token` — Get a LaunchDarkly access token by token identifier.
- `get_tokens` — List LaunchDarkly access tokens with optional pagination and visibility scope.
- `list_projects` — List LaunchDarkly projects with optional filtering, sorting, pagination, and expansion.
- `list_teams` — List LaunchDarkly teams with optional filtering, pagination, and expansion controls.

Credential Access

High
Category
Privilege Escalation
Confidence
70% confidence
Finding

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

Content

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

md
- `get_tags` — List LaunchDarkly tags with optional prefix and resource-kind filters.
- `get_team` — Get a LaunchDarkly team by team key.
- `get_token` — Get a LaunchDarkly access token by token identifier.
- `get_tokens` — List LaunchDarkly access tokens with optional pagination and visibility scope.
- `list_projects` — List LaunchDarkly projects with optional filtering, sorting, pagination, and expansion.
- `list_teams` — List LaunchDarkly teams with optional filtering, pagination, and expansion controls.
- `patch_environment` — Patch a LaunchDarkly environment with standard JSON Patch operations. [write]

External Script Fetching

High
Category
Supply Chain
Confidence
98% confidence
Finding

The command pipes a mutable remote response directly into Bash without version pinning, independent integrity verification, or review. Compromise of the installer distribution path could execute attacker-controlled code with the agent's local privileges; HTTPS and the legitimate setup purpose do not eliminate that supply-chain risk.

Content

Scanner excerpt · SKILL.md (reported line 94)May include surrounding context.

The setup workflow executes a downloaded installer without warning about remote code execution or system changes.

Medium
Category
Output Handling
Confidence
96% confidence
Finding

The missing-CLI fallback directs execution of downloaded shell or PowerShell code without a separate approval step or explanation of potential system changes. Although conditional and relevant to setup, it can turn an ordinary LaunchDarkly request into local code execution with the agent's privileges.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
80% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 115)May include surrounding context.

md
- **`OOMOL_INSUFFICIENT_CREDIT`** — OOMOL billing stop. Recharge at `https://console.oomol.com/billing/token-recharge` before retrying.

- **HTTP 402 without `OOMOL_INSUFFICIENT_CREDIT`** — the third-party account itself has no active plan or credit. Report the provider error message as is; do not ask the user to recharge OOMOL.

## Resources