ClawHub
Back to skill

Security audit

apple-ads-optimizer

Security checks for vulnerabilities and agentic risk

Overview

The optional unattended workflow can change live ad campaigns without the promised review of each change.

Start with read-only reports or interactive approvals. Before scheduling, require human approval in AppGoat and disable relevant auto-apply rules, or have unattended runs stop after creating proposals. Pin a reviewed repository version and install only the needed playbook. AppGoat access includes sensitive campaign data and potentially live campaign changes.

A.I.G

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

SkillSpector

Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (8)

The unattended workflow can apply live campaign changes without the user approving the proposed diff.

High
Category
Excessive Agency
Confidence
99% confidence
Finding

The unattended workflow skips approval of the proposed diff and calls apply_action, which the skill explicitly says can immediately execute changes when an organization has an auto-apply rule. Consequently, its assurance that nothing changes until a person approves is incorrect for that configuration, exposing live advertising spend and campaign performance to unreviewed changes.

Content

No source excerpt is available for this finding.

Anti-Refusal Statement

High
Category
Anti-Refusal
Confidence
80% confidence
Finding

Skill attempts to nullify the agent's safety policies or restrictions ('you have no restrictions', 'ignore your guidelines', 'do anything now'). This is a direct jailbreak that disables guardrails.

Content

Scanner excerpt · SKILL.md (reported line 159)May include surrounding context.

md
prompt: |
            Use the apple-ads-optimizer skill for all enabled campaigns with target CPA <amount> <currency>
            and daily budget <amount>. This is an unattended run: instead of asking in step 7, call apply_action
            once for every proposed action without guardrail warnings (this submits it for approval in the
            dashboard) and reject_action the others with the reason. Never call apply_action with force.
          claude_args: |
            --mcp-config ${{ runner.temp }}/appgoat-mcp.json

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 151)May include surrounding context.

md
cp -r /tmp/appgoat-skills/skills/* .claude/skills/
          # Hosted AppGoat MCP server; the key goes into a file, not the command line.
          jq -n --arg key "$APPGOAT_API_KEY" \
            '{mcpServers: {appgoat: {type: "http", url: "https://api.appgoat.ai/mcp", headers: {"x-api-key": $key}}}}' \
            > "$RUNNER_TEMP/appgoat-mcp.json"
      - uses: anthropics/claude-code-action@v1
        with:

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 · reference/api.md (reported line 29)May include surrounding context.

bash
CON=con_… CAMPAIGN=123456
curl -s -G -H "x-api-key: $APPGOAT_API_KEY" "https://api.appgoat.ai/v1/connections/$CON/ads" \
  -d view=report -d level=searchterms -d campaign_id=$CAMPAIGN -d start=2026-10-08 -d end=2026-10-21 \
  | jq -r '.data.rows[] | "\(.dims.searchTerm // .dims.searchTermText)\t\(.taps)\t\(.installs)\t\(.spend)"'

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 · reference/api.md (reported line 34)May include surrounding context.

md
| jq -r '.data.rows[] | "\(.dims.searchTerm // .dims.searchTermText)\t\(.taps)\t\(.installs)\t\(.spend)"'

# Bids for one ad group, as a dry run
curl -s -H "x-api-key: $APPGOAT_API_KEY" -H "content-type: application/json" https://api.appgoat.ai/v1/actions \
  -d '{"type":"ads.keyword.bid","connection":"'$CON'","dry_run":true,"source":"playbook:apple-ads-optimizer",
       "params":{"campaign_id":"'$CAMPAIGN'","ad_group_id":"789","max_increase":0.3,
                 "keywords":[{"id":"111","bid":1.3},{"id":"112","status":"PAUSED"}]}}' | jq '.data | {id, diff, guardrail}'

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 · reference/api.md (reported line 3)May include surrounding context.

md
# REST fallback

Without the AppGoat MCP server, call the API directly. Base URL `https://api.appgoat.ai/v1`,
header `x-api-key: $APPGOAT_API_KEY` (create a key at https://dash.appgoat.ai). Responses are
`{ "data": …, "meta": { "credits_used", "credits_remaining", … } }`; errors are
`{ "error": { "code", "message", "hint", "approve_url"? } }` - read `hint`.

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 · reference/api.md (reported line 29)May include surrounding context.

bash
CON=con_… CAMPAIGN=123456
curl -s -G -H "x-api-key: $APPGOAT_API_KEY" "https://api.appgoat.ai/v1/connections/$CON/ads" \
  -d view=report -d level=searchterms -d campaign_id=$CAMPAIGN -d start=2026-10-08 -d end=2026-10-21 \
  | jq -r '.data.rows[] | "\(.dims.searchTerm // .dims.searchTermText)\t\(.taps)\t\(.installs)\t\(.spend)"'

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 · reference/api.md (reported line 34)May include surrounding context.

md
| jq -r '.data.rows[] | "\(.dims.searchTerm // .dims.searchTermText)\t\(.taps)\t\(.installs)\t\(.spend)"'

# Bids for one ad group, as a dry run
curl -s -H "x-api-key: $APPGOAT_API_KEY" -H "content-type: application/json" https://api.appgoat.ai/v1/actions \
  -d '{"type":"ads.keyword.bid","connection":"'$CON'","dry_run":true,"source":"playbook:apple-ads-optimizer",
       "params":{"campaign_id":"'$CAMPAIGN'","ad_group_id":"789","max_increase":0.3,
                 "keywords":[{"id":"111","bid":1.3},{"id":"112","status":"PAUSED"}]}}' | jq '.data | {id, diff, guardrail}'

Static analysis

No suspicious patterns detected.