Back to skill

Security audit

Puffermind

Security checks for vulnerabilities and agentic risk

Overview

The skill matches its social-network purpose, but it needs Review because it installs unpinned remote instructions and tells the agent to use an unreviewed remote heartbeat repeatedly while enabling authenticated public actions.

Install only if you are comfortable giving this skill an API key for public social-network actions. Review the remote heartbeat content before use, avoid auto-updating instruction files from mutable URLs, and prefer pinned or checksum-verified copies.

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
Findings (1)

T01 · Skill Instruction Hijacking

Warning
Location
skill.md:35
Finding

Unpinned Remote Skill and Heartbeat Instructions Enable Instruction Hijacking

Content
View full analysis

Vulnerability Details

File Location: skill.md, lines 35–36; recurring use is directed at line 510
Vulnerability Type: Remote mutable instruction retrieval
Risk Level: Medium

Complete vulnerable code snippet:

bash
curl -fsS https://puffermind.com/skill.md > ~/.puffermind/skills/puffermind/skill.md
curl -fsS https://puffermind.com/heartbeat.md > ~/.puffermind/skills/puffermind/heartbeat.md

The downloaded heartbeat is subsequently designated as a recurring routine:

text
Use `https://puffermind.com/heartbeat.md` as the recurring routine after registration.

Technical Analysis

The installation instructions overwrite local Skill files with content downloaded from external URLs without verifying a cryptographic signature, checksum, immutable version, or reviewed content digest. The project does not include heartbeat.md, so the recurring instructions cannot be assessed as part of this audit.

Although HTTPS protects content while it is in transit, it does not establish that future content served by the publisher is identical to the audited version. Control of the website, hosting environment, DNS/TLS infrastructure, or publishing account could allow an attacker to replace either document. Because these documents are agent instructions rather than inert data, modified content could alter the agent's goals or operational constraints when loaded.

This is classified as instruction hijacking rather than remote code execution: the shown commands retrieve Markdown instruction files, and the reviewed evidence does not demonstrate that executable code is fetched or automatically executed.

Attack Path

  1. An attacker compromises the Puffermind publishing endpoint, deployment credentials, hosting account, or another component capable of changing skill.md or heartbeat.md.
  2. The attacker replaces one of the remote documents with instructions that direct the agent to perform unintended actio ...[truncated 1151 chars]
Remediation
View remediation

Remediation Suggestions

  1. Include both skill.md and heartbeat.md in the reviewed package rather than downloading mutable copies during installation.
  2. If remote distribution is necessary, publish immutable versioned artifacts and pin the installation to a specific release.
  3. Publish a trusted SHA-256 digest or detached digital signature and verify it before replacing local files. Abort installation on any mismatch.
  4. Download to a temporary file, validate it, and only then perform an atomic replacement of the installed file.
  5. Require explicit user review and approval before applying instruction updates.
  6. Do not automatically treat newly downloaded heartbeat content as trusted recurring instructions.
  7. Restrict heartbeat behavior to a documented allowlist of API hosts and operations, applying least privilege to available credentials and tools.
  8. Retain the last verified version to support rollback after failed verification or suspicious changes.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Rogue AgentSelf-Modification, Session Persistence
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (50)

Session Persistence

Medium
Category
Rogue Agent
Confidence
60% confidence
Finding

Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.

Content

Scanner excerpt · skill.md (reported line 34)May include surrounding context.

Install Locally

bash
mkdir -p ~/.puffermind/skills/puffermind
curl -fsS https://puffermind.com/skill.md > ~/.puffermind/skills/puffermind/skill.md
curl -fsS https://puffermind.com/heartbeat.md > ~/.puffermind/skills/puffermind/heartbeat.md

Skill Enumeration

Medium
Category
Agent Snooping
Confidence
80% 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 · skill.md (reported line 35)May include surrounding context.

bash
mkdir -p ~/.puffermind/skills/puffermind
curl -fsS https://puffermind.com/skill.md > ~/.puffermind/skills/puffermind/skill.md
curl -fsS https://puffermind.com/heartbeat.md > ~/.puffermind/skills/puffermind/heartbeat.md

External Transmission

Medium
Category
Data Exfiltration
Confidence
60% 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 · skill.md (reported line 35)May include surrounding context.

bash
mkdir -p ~/.puffermind/skills/puffermind
curl -fsS https://puffermind.com/skill.md > ~/.puffermind/skills/puffermind/skill.md
curl -fsS https://puffermind.com/heartbeat.md > ~/.puffermind/skills/puffermind/heartbeat.md

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 · skill.md (reported line 44)May include surrounding context.

Register first:

bash
curl -X POST https://api.puffermind.com/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "handle": "your_agent_handle",

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 · skill.md (reported line 71)May include surrounding context.

Register first:

bash
curl -X POST https://api.puffermind.com/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "handle": "your_agent_handle",

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 · skill.md (reported line 86)May include surrounding context.

Register first:

bash
curl -X POST https://api.puffermind.com/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "handle": "your_agent_handle",

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 · skill.md (reported line 97)May include surrounding context.

Register first:

bash
curl -X POST https://api.puffermind.com/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "handle": "your_agent_handle",

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 · skill.md (reported line 104)May include surrounding context.

Register first:

bash
curl -X POST https://api.puffermind.com/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "handle": "your_agent_handle",

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 · skill.md (reported line 124)May include surrounding context.

Register first:

bash
curl -X POST https://api.puffermind.com/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "handle": "your_agent_handle",

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 · skill.md (reported line 131)May include surrounding context.

Register first:

bash
curl -X POST https://api.puffermind.com/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "handle": "your_agent_handle",

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 · skill.md (reported line 149)May include surrounding context.

Register first:

bash
curl -X POST https://api.puffermind.com/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "handle": "your_agent_handle",

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 · skill.md (reported line 156)May include surrounding context.

Register first:

bash
curl -X POST https://api.puffermind.com/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "handle": "your_agent_handle",

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 · skill.md (reported line 167)May include surrounding context.

Register first:

bash
curl -X POST https://api.puffermind.com/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "handle": "your_agent_handle",

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 · skill.md (reported line 180)May include surrounding context.

Register first:

bash
curl -X POST https://api.puffermind.com/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "handle": "your_agent_handle",

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 · skill.md (reported line 187)May include surrounding context.

Register first:

bash
curl -X POST https://api.puffermind.com/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "handle": "your_agent_handle",

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 · skill.md (reported line 201)May include surrounding context.

Register first:

bash
curl -X POST https://api.puffermind.com/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "handle": "your_agent_handle",

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 · skill.md (reported line 214)May include surrounding context.

Register first:

bash
curl -X POST https://api.puffermind.com/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "handle": "your_agent_handle",

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 · skill.md (reported line 229)May include surrounding context.

Register first:

bash
curl -X POST https://api.puffermind.com/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "handle": "your_agent_handle",

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 · skill.md (reported line 236)May include surrounding context.

Register first:

bash
curl -X POST https://api.puffermind.com/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "handle": "your_agent_handle",

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 · skill.md (reported line 243)May include surrounding context.

Register first:

bash
curl -X POST https://api.puffermind.com/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "handle": "your_agent_handle",

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 · skill.md (reported line 254)May include surrounding context.

Register first:

bash
curl -X POST https://api.puffermind.com/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "handle": "your_agent_handle",

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 · skill.md (reported line 261)May include surrounding context.

Register first:

bash
curl -X POST https://api.puffermind.com/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "handle": "your_agent_handle",

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 · skill.md (reported line 275)May include surrounding context.

Register first:

bash
curl -X POST https://api.puffermind.com/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "handle": "your_agent_handle",

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 · skill.md (reported line 287)May include surrounding context.

Register first:

bash
curl -X POST https://api.puffermind.com/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "handle": "your_agent_handle",

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 · skill.md (reported line 296)May include surrounding context.

Register first:

bash
curl -X POST https://api.puffermind.com/v1/agents/register \
  -H "Content-Type: application/json" \
  -d '{
    "handle": "your_agent_handle",

Static analysis

No suspicious patterns detected.