Back to skill

Security audit

Sales Pro

Security checks for vulnerabilities and agentic risk

Overview

The skill is mostly a sales-assistant package, but it asks to persist and synchronize customer deal data without enough scoping or consent details.

Install only if you are comfortable with a sales skill that may store lead and deal details in agent memory and synchronize or reuse them across other workflows. Before using it with real customer data, require explicit opt-in storage, define the Supabase tenant and access rules, limit fields shared with other skills, and confirm how records can be deleted or sync disabled.

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

T05 · Unauthorized Access and Privilege Escalation

Error
Location
SKILL.md:710
Finding

Unspecified External Synchronization and Cross-Skill Propagation of Customer Data

Content
View full analysis

Vulnerability Details

File Location: SKILL.md:710-715
Vulnerability Type: Excessive data access, external synchronization, and cross-skill data propagation
Risk Level: High

Vulnerable Code

text
## 17. Интеграция с OpenClaw
- Сделки и лиды хранятся в memory/ агента, синхронизируются через Supabase
- Работает с AI-Поддержка: клиент из поддержки → триггер лида на допродажу
- Работает с AI-Аналитик: данные воронки → автоматический дашборд и прогнозы
- Работает с Контент-мастер: кейсы → контент → новые лиды
- Follow-up ставится на таймер через AI-Офис
- Win/loss данные → автоматически в отчёт для OpenClaw

Technical Analysis

The skill instructs the agent to persist lead and deal records in agent memory, synchronize those records through an unspecified Supabase service, and propagate related data to support, analytics, content, office, and reporting workflows.

Lead and deal records may contain personal information, customer communications, commercial terms, pipeline details, and other confidential business information. These instructions do not define:

  • The Supabase endpoint, tenant, or data controller
  • The exact fields transmitted or retained
  • User or workspace consent requirements
  • Authentication and authorization boundaries
  • Data minimization and redaction rules
  • Retention and deletion policies
  • Access restrictions for downstream skills
  • Restrictions on secondary use, such as converting customer cases into marketing content

This exceeds the minimum permissions required for a text-based sales-advice skill. It also broadens the number of components able to access customer data, increasing the consequences of a compromised or incorrectly configured integration.

No executable Supabase client, endpoint, or synchronization implementation was found in the audited project. Therefore, the audit does not confirm that data is currently transmitted by the included shell scripts. The risk arises from operational instructions that ...[truncated 1557 chars]

Remediation
View remediation

Remediation Suggestions

  1. Default to session-only, local processing and disable persistence or synchronization unless the user explicitly opts in.
  2. Require separate, informed confirmation before storing lead or deal data, synchronizing it externally, or transferring it to another skill.
  3. Specify and validate an allowlisted Supabase endpoint, tenant, region, and data controller.
  4. Document the exact fields collected, their processing purpose, retention period, and deletion procedure.
  5. Minimize stored fields and redact personal data, credentials, payment information, contract contents, and confidential communications.
  6. Use tenant-isolated storage, row-level security, scoped service credentials, encryption in transit, and encryption at rest.
  7. Grant each downstream skill access only to the minimum fields required for its declared function.
  8. Prohibit support-to-sales and customer-case-to-content reuse unless the user separately authorizes that purpose.
  9. Add auditable controls to inspect, export, delete, and disable synchronization of stored records.
  10. Clearly distinguish implemented integrations from aspirational functionality so an agent does not interpret unsupported claims as authorization to transmit data.
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Findings (90)

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The declared purpose describes end-user sales functionality, but the supplied code does not implement any sales workflows, qualification logic, objection handling, pipeline management, or offer generation. Instead, it is an infrastructure/build script whose primary purpose is assembling and validating a distributable ZIP package. This is a materially different purpose from the declared sales system, so it should be flagged as a mismatch.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
95% confidence
Finding

The declared description presents the skill as an operational sales system with business-facing capabilities such as scripting, objection handling, qualification, pipeline management, and offer creation. The supplied code chunk does not implement those sales functions. Instead, it is an installation script whose purpose is to deploy the skill files locally, initialize environment/config files, and guide setup. While an installer can be a supporting component of such a skill, this specific code chunk’s actual behavior is materially different from the declared purpose and includes undeclared filesystem and local environment interaction capabilities.

Content

No source excerpt is available for this finding.

Tp4

High
Category
MCP Tool Poisoning
Confidence
98% confidence
Finding

The supplied code does not implement sales workflows, qualification logic, objection handling, follow-up generation, pipeline management, or offer creation. Instead, it is a test utility that inspects the skill's file layout and content using shell commands like grep, find, and head, then reports pass/fail status. While some checked keywords relate to sales, the script's actual purpose is QA/compliance testing for the skill package, which is materially different from the declared end-user sales system description. Therefore this is a clear description-behavior mismatch.

Content

No source excerpt is available for this finding.

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 · install.sh (reported line 31)May include surrounding context.

sh
cp -r "$SRC_DIR/docs" "$SRC_DIR/examples" "$SRC_DIR/proof" "$SRC_DIR/test" "$TARGET/"
chmod +x "$TARGET/install.sh" "$TARGET/test/smoke-test.sh" 2>/dev/null || true # Опциональные папки (marketing, competitors)
for sub in marketing competitors; do if [[ -d "$SRC_DIR/$sub" ]]; then cp -r "$SRC_DIR/$sub" "$TARGET/$sub" fi
done echo "[✓] Файлы скопированы в $TARGET" # 3. Создать .env из шаблона если нет
if [[ ! -f "$TARGET/.env" ]]; then cp "$SRC_DIR/.env.example" "$TARGET/.env" echo "[✓] Создан $TARGET/.env (заполнять только если нужны интеграции)"
else echo "[i] $TARGET/.env уже существует — не перезаписываю"
fi # 4. Напомнить про OBLIGATORY-поля

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 · install.sh (reported line 32)May include surrounding context.

sh
cp -r "$SRC_DIR/docs" "$SRC_DIR/examples" "$SRC_DIR/proof" "$SRC_DIR/test" "$TARGET/"
chmod +x "$TARGET/install.sh" "$TARGET/test/smoke-test.sh" 2>/dev/null || true # Опциональные папки (marketing, competitors)
for sub in marketing competitors; do if [[ -d "$SRC_DIR/$sub" ]]; then cp -r "$SRC_DIR/$sub" "$TARGET/$sub" fi
done echo "[✓] Файлы скопированы в $TARGET" # 3. Создать .env из шаблона если нет
if [[ ! -f "$TARGET/.env" ]]; then cp "$SRC_DIR/.env.example" "$TARGET/.env" echo "[✓] Создан $TARGET/.env (заполнять только если нужны интеграции)"
else echo "[i] $TARGET/.env уже существует — не перезаписываю"
fi # 4. Напомнить про OBLIGATORY-поля

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 · install.sh (reported line 33)May include surrounding context.

sh
cp -r "$SRC_DIR/docs" "$SRC_DIR/examples" "$SRC_DIR/proof" "$SRC_DIR/test" "$TARGET/"
chmod +x "$TARGET/install.sh" "$TARGET/test/smoke-test.sh" 2>/dev/null || true # Опциональные папки (marketing, competitors)
for sub in marketing competitors; do if [[ -d "$SRC_DIR/$sub" ]]; then cp -r "$SRC_DIR/$sub" "$TARGET/$sub" fi
done echo "[✓] Файлы скопированы в $TARGET" # 3. Создать .env из шаблона если нет
if [[ ! -f "$TARGET/.env" ]]; then cp "$SRC_DIR/.env.example" "$TARGET/.env" echo "[✓] Создан $TARGET/.env (заполнять только если нужны интеграции)"
else echo "[i] $TARGET/.env уже существует — не перезаписываю"
fi # 4. Напомнить про OBLIGATORY-поля

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 · install.sh (reported line 32)May include surrounding context.

sh
chmod +x "$TARGET/install.sh" "$TARGET/test/smoke-test.sh" 2>/dev/null || true # Опциональные папки (marketing, competitors)
for sub in marketing competitors; do if [[ -d "$SRC_DIR/$sub" ]]; then cp -r "$SRC_DIR/$sub" "$TARGET/$sub" fi
done echo "[✓] Файлы скопированы в $TARGET" # 3. Создать .env из шаблона если нет
if [[ ! -f "$TARGET/.env" ]]; then cp "$SRC_DIR/.env.example" "$TARGET/.env" echo "[✓] Создан $TARGET/.env (заполнять только если нужны интеграции)"
else echo "[i] $TARGET/.env уже существует — не перезаписываю"
fi # 4. Напомнить про OBLIGATORY-поля
echo ""

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

The changelog content is written in Russian mixed with English terms, with no indication that language choice is optional or that the file is intentionally region-specific. This can violate language/locale policy when users are not given an opt-in or alternative language.

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The Windows guidance tells users to open SKILL.md and send any trigger from the 'modes of operation' section, which encourages broad activation from natural-language phrases. That increases the chance the skill will activate unintentionally during ordinary sales discussion and steer the model into skill behavior without clear user intent.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

Many listed triggers are highly generic words or short phrases such as 'pipeline', 'воронка', 'квалификация', 'разбор', and 'шаблоны'. In a sales-focused assistant, these are likely to appear in normal user conversation, causing accidental mode switching, prompt collisions, or unintended execution of the skill's internal workflows.

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

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.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The trigger set includes broad phrases such as variants of 'how to call,' which can match ordinary conversation rather than an intentional request to invoke this skill. In an agent environment, overbroad activation can cause the sales skill to engage unexpectedly, steering unrelated chats, processing unintended user content, or initiating downstream workflows based on false activations.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The trigger phrase for correspondence/help-writing is too generic and can overlap with common user requests outside sales operations. This increases the chance of accidental skill activation and unintended handling of arbitrary user text, which is especially risky because the skill encourages follow-up and CRM-oriented outputs.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

A generic objection-handling trigger can match many non-sales requests, making unintended invocation plausible. In multi-skill systems, this may cause the agent to reframe unrelated dialogue as persuasion or negotiation content, producing incorrect or manipulative outputs in contexts where they were not requested.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The Challenger-related trigger aliases include broad educational phrasing that could match general advisory requests. This can make the skill intrude into unrelated domains and apply aggressive sales framing inappropriately, which is a behavior-control issue even if it is not code execution.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The single-word trigger 'pipeline' is especially broad and likely to collide with software, DevOps, analytics, hiring, and general workflow discussions. Such collisions create a high risk of accidental invocation and context confusion in shared agent environments.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The commercial-proposal trigger is broad enough to match ordinary document drafting requests that are not related to this skill. Unintended activation could lead the agent to generate sales-oriented proposals or insert persuasive structure where the user wanted neutral business writing.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
85% confidence
Finding

The dialogue-analysis trigger is ambiguous and may match general requests to analyze a conversation, including support, HR, legal, or personal exchanges. This can route sensitive non-sales conversations into a sales-optimization framework, causing privacy and appropriateness issues.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill explicitly states that deals and leads are stored in agent memory and synchronized via Supabase, but it provides no user-facing warning, consent flow, retention details, or handling guidance for potentially sensitive business and personal data. In a sales context, this may involve names, contact details, pipeline status, budgets, and internal notes, so silent persistence materially raises privacy and data-governance risk.

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The build script automatically unpacks the newly created archive and executes test/smoke-test.sh from that unpacked content. If any file included in the archive is malicious or tampered with, running the build implicitly triggers arbitrary shell execution, which is especially risky because this is a packaging script that may be run by maintainers or CI with elevated trust.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The trigger phrases are broad, natural-language requests such as "Review my pipeline" and "Forecast this quarter," which can easily overlap with ordinary user prompts and cause ambiguous or unintended skill activation. In an agent environment, this can misroute user requests into this sales skill when the user did not explicitly intend it, potentially causing irrelevant guidance, context leakage across skills, or unsafe autonomous actions tied to sales workflows.

Content

No source excerpt is available for this finding.