Back to skill

Security audit

MyLister - Organizer

Security checks for vulnerabilities and agentic risk

Overview

The skill is a disclosed MyLister task manager, but it gives an agent broad data-sharing, credential, endpoint-switching, and local-file upload powers that need review before installation.

Install only if you trust the MyLister API integration and are comfortable giving the agent authority to read and change your task data, email/export it, share lists, manage collaborator permissions, and upload local files. Avoid using chat commands to enter API keys, do not switch the API base URL to untrusted hosts, and require explicit confirmation for deletes, shares, emails, permission changes, file URLs, and any upload path.

Vulnerability Patterns
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • 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
Findings (2)

T09 · Insecure Skill Coding Practices

Error
Location
src/index.ts:1716
Finding

API Credential Exfiltration Through Arbitrary Runtime Endpoint Switching

Content
View full analysis
{ return { 'X-API-Key': this.apiKey, 'Accept': 'application/json', 'Content-Type': 'application/json', }; } private getMultipartHeader(): Record { // Let fetch add the multipart boundary; setting Content-Type manually breaks uploads. return { 'X-API-Key': this.apiKey, 'Accept': 'application/json', }; } ``` ```typescript let runtimeBaseUrl = normalizeBaseUrl(CONFIG.baseUrl); const runtimeApiKeys: Record = { production: DEFAULT_PRODUCTION_API_KEY, staging: DEFAULT_STAGING_API_KEY, custom: CONFIG.apiKey, }; let runtimeApiEnvironment = resolveRuntimeApiEnvironment(runtimeBaseUrl); let runtimeApiKey = runtimeApiKeys[runtimeApiEnvironment]; let client = new ListerClient(runtimeBaseUrl, runtimeApiKey); function getConfiguredApiBaseUrl(targetEnvironment: ApiEnvironment): string | undefined { if (targetEnvironment === 'production' ...[truncated 4078 chars]
Remediation
View remediation

T05 · Unauthorized Access and Privilege Escalation

Error
Location
src/index.ts:1560
Finding

Arbitrary Local File Disclosure Through Unrestricted Upload Paths

Content
View full analysis
): Promise { try { const form = new FormData(); form.append('file', await fileFrom(filePath)); for (const [name, value] of Object.entries(fields)) { if (value !== undefined) form.append(name, String ...[truncated 3982 chars]
Remediation
View remediation
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Findings (36)

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 · .gitignore (reported line 3)May include surrounding context.

text
node_modules/
dist/
.env
*.tsbuildinfo

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The skill advertises natural language task management but also supports sharing lists, managing user permissions, exporting/emailing records, file upload, presigned URL retrieval, and runtime API key/base-URL changes. In security terms, that is a material expansion of authority and data-movement capability beyond what a user may reasonably expect from the description.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The skill advertises natural language task management but also supports sharing lists, managing user permissions, exporting/emailing records, file upload, presigned URL retrieval, and runtime API key/base-URL changes. In security terms, that is a material expansion of authority and data-movement capability beyond what a user may reasonably expect from the description.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill supports emailing list/item data, sharing content with other users, and uploading local files or media to an external service, but the documentation does not present clear user warnings or consent requirements for these transfers. That materially increases the risk of unintended data exfiltration from the user's workspace or local machine.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The command documentation says completed items can be moved from one list to another, while the API reference later states the operation only reorders completed items within the same list. Contradictory semantics for a state-changing command are dangerous because agents or users may believe data is being moved or archived when it is not, leading to accidental disclosure, retention, or workflow failures.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
95% confidence
Finding

Allowing natural-language commands to set API keys and override the API environment or base URL is dangerous because an attacker can socially engineer the agent into redirecting requests to attacker-controlled infrastructure and then disclose or misuse secrets. In the context of a task/list skill, these controls are not necessary for normal user outcomes and substantially raise the risk of credential exfiltration and unauthorized backend access.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The skill exposes runtime API environment switching and API key reconfiguration through natural-language commands, even though the manifest frames it as ordinary task/list management. This expands the trust boundary significantly: a user can be induced to redirect requests to an attacker-controlled endpoint and then transmit credentials or task data there, creating an SSRF/exfiltration-style channel and enabling credential misuse.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The skill can read arbitrary local file paths via upload commands and send file contents to the remote API, which is far beyond simple list management. Because file paths come from natural language and there is no path restriction or consent checkpoint, an attacker could socially engineer uploads of sensitive local files such as SSH keys, environment files, or documents.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

High
Category
Not specified by scanner
Confidence
99% confidence
Finding

The upload helpers accept arbitrary local file paths and immediately read them with fileFrom() before POSTing contents to the API. This creates a direct local-file exfiltration primitive under natural-language control, especially dangerous because the same skill also supports changing the API base URL to a custom endpoint.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The README tells users to 'ask naturally' for task actions, which encourages broad implicit invocation of a skill that can perform state-changing operations. In agent hosts, vague trigger guidance can increase the chance of accidental activation from ordinary conversation, leading to unintended task creation, deletion, emailing, or environment changes.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The Hermes instructions similarly allow normal-language use without constraints, despite the skill supporting destructive and data-transmitting commands. That broad trigger model raises the risk that unrelated user text could be interpreted as an instruction to modify lists or send content externally.

Content

No source excerpt is available for this finding.

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
85% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

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

npm ci

text

If `CODEX_HOME` is not set, use the Codex skills directory shown by your installation, commonly `~/.codex/skills/lister`. Restart Codex or start a new session, set `LISTER_API_KEY` in the environment, and ask for a MyLister action. Codex can then load the root `SKILL.md` and run the bundled CLI.

### Claude Code

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
85% confidence
Finding

Skill enumerates or reads other installed skills. Access to other skills' SKILL.md files or the skills directory reveals prompt instructions, capabilities, and secrets that should be invisible to peer skills.

Content

Scanner excerpt · README.md (reported line 84)May include surrounding context.

Claude Code

Claude Code supports project skills in .claude/skills/ and personal skills in ~/.claude/skills/. Install it for all projects with:

bash
git clone https://github.com/mgoettel13/MyLister-Skill.git ~/.claude/skills/lister

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The Claude Code section says to 'ask naturally' or invoke the skill, which normalizes ambiguous triggering for a capability that can alter data and transmit it by email/API. In environments where skills may be auto-selected from user intent, that ambiguity increases the chance of unintended execution.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The supported commands include deletion, updates, emailing lists/items, file attachment/upload, API base URL switching, and API key setting, but the README does not warn users that these actions are destructive, privacy-impacting, or security-sensitive. Without clear warnings and confirmation guidance, users and host agents may execute risky operations without appreciating the consequences.

Content

No source excerpt is available for this finding.

External Transmission

Medium
Category
Data Exfiltration
Confidence
50% confidence
Finding

Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.

Content

Scanner excerpt · README.md (reported line 146)May include surrounding context.

md
## API

- **API documentation:** [MyLister API docs](https://api.mylister.dev/docs)
- **Base URL:** `https://api.mylister.dev` (or override with `LISTER_BASE_URL`, `LISTER_PRODUCTION_BASE_URL`, `LISTER_STAGING_BASE_URL`)
- **Production/staging keys:** Set `LISTER_PRODUCTION_API_KEY` / `LISTER_STAGING_API_KEY` when envs use different credentials.
- **Auth:** API key via `X-API-Key`

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
96% confidence
Finding

The skill declares network and environment-backed behavior but does not define an explicit tool/permission scope. That makes it easier for an agent framework to invoke sensitive capabilities such as external API access and secret usage without clear policy boundaries or user awareness.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

Broad trigger guidance such as 'whenever the user wants to manage tasks' and 'not limited to' creates an overbroad invocation surface. In agent settings, vague routing criteria can cause this skill to activate on loosely related prompts and perform unintended external actions against the API.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

L595 states that if a list does not exist, the skill reports an error and does not auto-create lists. However, the documented API reference includes POST /v1/lists/defaults at L505, which is a list-creation capability beyond that stated behavior. This is an active contradiction between the notes and the documented behavior/capability surface.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The command trigger list is unusually broad and includes many generic words such as "add," "get," "list," "send," "set," "api," and "key" that are likely to appear in normal conversation. In an agent skill context, this increases the chance of accidental invocation or misrouting of user intent, which could trigger task management actions or expose integrations unexpectedly when the user did not intend to use this skill.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The tests confirm the skill supports commands for API-key setting and environment/base-URL switching, which extend beyond the declared natural-language list-management purpose. Capability expansion into infrastructure/configuration control increases the attack surface and creates opportunities for redirection, credential misuse, or misuse of the skill as a transport for administrative actions.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
80% confidence
Finding

The tests document commands that accept API keys in natural language without any indication of secret-handling safeguards. This encourages unsafe operational patterns where secrets may appear in chat logs, telemetry, transcripts, or prompt context, making accidental exposure more likely.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The tests show support for local file, image, and voice upload flows. Even if intended for attachments, agent-accessible file upload can expose local files or sensitive user content if prompts or indirect instructions cause the agent to select unintended files for upload.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

File-URL retrieval plus API health/version inspection are non-core capabilities that can aid reconnaissance and broaden data exposure. Presigned or direct file URLs may enable access-sharing outside intended boundaries, while health/version endpoints can help fingerprint the backend for follow-on attacks.

Content

No source excerpt is available for this finding.

Ssd 3

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill supports exporting and emailing full lists or items to arbitrary recipients, creating a straightforward data exfiltration path across trust boundaries. In the context of a task manager, list contents may include sensitive business plans, personal reminders, contacts, or attachments, so unrestricted outbound sharing materially increases risk.

Content

No source excerpt is available for this finding.