Back to skill

Security audit

元史 yotta-logs

Security checks for vulnerabilities and agentic risk

Overview

Its local history search can include private memories even when configured to search conversations only.

Review the scope-control defect before installing. Until fixed, restrict searches with an explicit --dir or --source, or use --kind session to exclude memories; do not rely on default_scope to disable sources. Retrieved excerpts enter your agent's context, and redaction cannot guarantee removal of every secret. Use the pinned installer version and a specific destination. The material scanner gaps were independently inspected, but host-agent handling of retrieved text was not verified.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (26)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The supplied chunk is installation tooling, not the described retrieval implementation. Installation can be supporting functionality, but this code materially exceeds the explicit no-modification/no-deletion boundary: installTo() invokes mkdirSync() and copyDir(), which uses copyFileSync() to write or overwrite files; cleanResidue() removes allowlisted artifacts and caches using safeRemove(), unlinkSync(), and rmdirSync(). These actions are limited to installation destinations and include safety checks, but remain undeclared write and deletion capabilities. The local-only/no-upload claim is consistent with this chunk. No conclusion about retrieval functionality elsewhere in the package can be drawn from this file.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

Installation is ordinarily a supporting implementation detail, but this script's filesystem mutations materially exceed the explicit no-modification/no-deletion boundary. install_to and copy_tree create installation directories and copy files into them; clean_residue deletes allowlisted development artifacts and caches when the target contains SKILL.md. The --global option applies installation across known agent directories, while --dir allows a custom target. These are scoped installation operations, not evidence of malicious intent, and no network upload is present. The assessment is limited to this installer; the supplied chunk does not establish whether other files implement the advertised retrieval functionality.

Content

No source excerpt is available for this finding.

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 37)May include surrounding context.

md
python3 scripts/yotta_logs.py locate

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 40)May include surrounding context.

md
python3 scripts/yotta_logs.py search "部署方案"

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 43)May include surrounding context.

md
python3 scripts/yotta_logs.py scan --dir ~/.clawdbot/agents/<agentId>/sessions

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 44)May include surrounding context.

md
python3 scripts/yotta_logs.py search "CI 失败" --regex --date 2026-08-26 --dir /path/to/logs

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 47)May include surrounding context.

md
python3 scripts/yotta_logs.py search "记住" --kind memory

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 48)May include surrounding context.

md
python3 scripts/yotta_logs.py search "XSS" --source opencode-db

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 49)May include surrounding context.

md
python3 scripts/yotta_logs.py search "部署" --format sqlite

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 52)May include surrounding context.

md
python3 scripts/yotta_logs.py search "推送闸门" --kind note

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 55)May include surrounding context.

md
python3 scripts/yotta_logs.py session abc123 --dir /path/to/sessions

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 58)May include surrounding context.

md
python3 scripts/yotta_logs.py stats --dir /path/to/sessions --daily

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 61)May include surrounding context.

md
python3 scripts/yotta_logs.py tools --dir /path/to/logs --format sqlite

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-logs'.

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
70% confidence
Finding

npx commands without a version suffix (e.g. @1.0.0) create a rug-pull risk if the upstream server is compromised and publishes a malicious update.

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 114)May include surrounding context.

text
# Optional China mirror: npm config set registry https://registry.npmmirror.com
npx -y @yottameta/yotta-logs@0.3.5 --agent <agent-name>      # install to the agent's default user-level skills dir
npx -y @yottameta/yotta-logs@0.3.5 --dir <your-skills-dir>   # point to the skills dir itself (e.g. ~/.codex/skills)

Skill declares no tool scope ('permissions' or 'allowed-tools') but code capabilities were detected: env, file_read.

Medium
Category
MCP Least Privilege
Confidence
70% confidence
Finding

Without declared permissions the skill's intent is opaque and cannot be validated.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-logs'.

Medium
Category
MCP Rug Pull
Confidence
91% confidence
Finding

This usage example fetches and executes an unpinned npm release with automatic confirmation. If copied after a malicious or compromised release becomes current, it could execute attacker-controlled code with the user's permissions and write into an agent's skills directory; this is installer supply-chain risk, not an MCP configuration issue.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-logs'.

Medium
Category
MCP Rug Pull
Confidence
91% confidence
Finding

The custom-directory installation example executes whichever npm release is current rather than a reviewed fixed version. A compromised release could run with the user's permissions and tamper with the selected installation, regardless of the legitimate installer's local-only retrieval guarantees.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-logs'.

Medium
Category
MCP Rug Pull
Confidence
92% confidence
Finding

The global-install example combines unpinned package execution with installation across multiple agents' skill directories. A compromised current release could execute arbitrary code and potentially distribute malicious skill content across those agents.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-logs'.

Medium
Category
MCP Rug Pull
Confidence
91% confidence
Finding

The default-install example resolves and executes an unpinned npm package before installing into detected project-level directories. A compromised release could run arbitrary code or alter agent instructions in those projects.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-logs'.

Medium
Category
MCP Rug Pull
Confidence
90% confidence
Finding

Although the legitimate '--list' operation only prints supported directories, npx still resolves and executes the unpinned package first. A malicious current release need not honor that flag and could execute arbitrary code with the user's permissions.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-logs'.

Medium
Category
MCP Rug Pull
Confidence
91% confidence
Finding

The '--dry-run' flag limits writes only in the legitimate installer implementation; it does not prevent npx from executing an unpinned remote package. A compromised release could ignore dry-run behavior and access or modify files with the user's permissions.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-logs'.

Medium
Category
MCP Rug Pull
Confidence
90% confidence
Finding

Even a version query executes package code when issued through this unpinned npx command. A compromised current release could perform arbitrary actions before or instead of printing its version.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx -y @yottameta/yotta-logs'.

Medium
Category
MCP Rug Pull
Confidence
90% confidence
Finding

The help example still downloads and executes whichever package release is current. The harmless intent of '--help' does not constrain a compromised release, which would execute with the user's filesystem permissions.

Content

No source excerpt is available for this finding.