Back to skill

Security audit

Sender Immersive Web Department

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious with medium confidence, mainly because it flags public sharing of Lighthouse reports and user videos without privacy safeguards, while some referenced files were unavailable for analysis.

Consider the privacy of any reports or videos shared through the suggested workflows, and whether you are comfortable with instructions that can deploy changes to a live site. The scan is incomplete: three referenced paths could not be matched to bundled files. The AIG scan completed without findings, but that does not resolve the missing-reference coverage.

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
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • Tool MisuseTool Parameter Abuse, Chaining Abuse, Unsafe Defaults
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (7)

Chaining Abuse

High
Category
Tool Misuse
Confidence
75% confidence
Finding

Tool calls are chained to bypass individual safety checks or escalate capabilities beyond what any single tool call would allow.

Content

Scanner excerpt · WORKFLOW.md (reported line 11)May include surrounding context.

md
| Repo | Rol | Reglas |
|---|---|---|
| `sender-site` (main) | Producción + Pages auto-deploy | Único que toca el dominio público. Cada push = deploy + audit CI. Commits semánticos `feat/fix/docs(vNN)`. |
| `sender-site-lab` | Sandbox de dirección creativa | Prueba conceptos rompedores sin riesgo; `compara.html` con iframes A/B lado a lado. Se sincroniza DESPUÉS de cada round aprobado (nunca al revés; su `vendor/` es copia plana, no submodule). |
| `sender-web` | REDISEÑO NUEVO (repo independiente, estático sin bundler): web "Editorial Film" — cristianoleamiranda-dotcom.github.io/sender-web |
| `sender-fx-lab` | Repo EXCLUSIVO de efectos scroll/3D/film: demos aislados P01/P14/P15/P07/P06 con Pages propia (cristianoleamiranda-dotcom.github.io/sender-fx-lab). Todo efecto nuevo se prototipa aqui antes de tocar produccion. |
| `sender-motion-kit` | Cerebro del departamento | `SKILL.md` (skill completa), `CHECKLIST.md` (gates), `TRACEABILITY.md` (rendición de cuentas), `patterns.md` P01-P14

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
85% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · SKILL.md (reported line 35)May include surrounding context.

md
3. Sistema: paleta oficial + tipografías self-hosted + grid hairline + lenguaje de motion.
4. Construir en repo LAB separado si el cambio es de dirección; iterar en main si es refinamiento.
5. Auditar: LHCI en CI + capturas móvil / móvil-modo-escritorio / desktop.
6. Ship: Pages auto-deploy + marcador `?v=N` + entrada en STATE.md + scorecard.
7. Loop: el feedback del usuario define la siguiente ronda (capturas > opiniones).

## S1 · Backbone inmersivo de scroll (la firma del sitio)

The CI workflow uploads Lighthouse reports to public temporary storage without a privacy warning.

Medium
Category
Output Handling
Confidence
94% confidence
Finding

The skill prescribes uploading Lighthouse reports to temporary public storage without checking whether the audited content is suitable for disclosure. Reports can contain screenshots, URLs, and page details, potentially exposing unpublished client material or sensitive URL parameters; temporary retention does not provide access control.

Content

No source excerpt is available for this finding.

The workflow requests third-party uploads of user videos without warning about privacy implications.

Medium
Category
Output Handling
Confidence
97% confidence
Finding

The skill instructs the agent to request that users re-upload videos to third-party file hosts without assessing confidentiality or explaining the disclosure. In a client-production workflow, those videos may contain private footage, personal information, or unreleased assets, and possession of a sharing link is not equivalent to authenticated access control.

Content

No source excerpt is available for this finding.

Autonomous Decision Making

Medium
Category
Excessive Agency
Confidence
85% confidence
Finding

Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.

Content

Scanner excerpt · WORKFLOW.md (reported line 10)May include surrounding context.

md
| Repo | Rol | Reglas |
|---|---|---|
| `sender-site` (main) | Producción + Pages auto-deploy | Único que toca el dominio público. Cada push = deploy + audit CI. Commits semánticos `feat/fix/docs(vNN)`. |
| `sender-site-lab` | Sandbox de dirección creativa | Prueba conceptos rompedores sin riesgo; `compara.html` con iframes A/B lado a lado. Se sincroniza DESPUÉS de cada round aprobado (nunca al revés; su `vendor/` es copia plana, no submodule). |
| `sender-web` | REDISEÑO NUEVO (repo independiente, estático sin bundler): web "Editorial Film" — cristianoleamiranda-dotcom.github.io/sender-web |
| `sender-fx-lab` | Repo EXCLUSIVO de efectos scroll/3D/film: demos aislados P01/P14/P15/P07/P06 con Pages propia (cristianoleamiranda-dotcom.github.io/sender-fx-lab). Todo efecto nuevo se prototipa aqui antes de tocar produccion. |

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

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.

The workflow directs production deployment without warning about changes to the live site.

Medium
Category
Output Handling
Confidence
88% confidence
Finding

Line 9 instructs the agent to commit, push, and deploy, while line 17 directs implementation followed by production verification. Neither workflow warns that following a supplied video can publish changes affecting the live site and its users.

Content

No source excerpt is available for this finding.