Skillv0.1.0

ClawScan security

Emergence PPT Orchestra · ClawHub's context-aware review of the artifact, metadata, and declared behavior.

Scanner verdict

BenignApr 16, 2026, 2:05 PM

Verdict: benign
Confidence: high
Model: gpt-5-mini
Summary: The skill's requirements and instructions are consistent with its stated purpose: it needs an EMERGENCE_API_KEY to call the Emergence Render API and otherwise operates locally using Marp; nothing requested appears unrelated or excessive.
Guidance: This skill appears coherent and does what it says: it drafts Marp markdown, calls the Emergence Render API for diagrams using your EMERGENCE_API_KEY, saves returned images locally, and compiles with Marp. Before installing, consider: 1) only provide an EMERGENCE_API_KEY that you trust and ideally scope or rotate it; 2) prefer installing a pinned marp-cli locally (or pin the version) instead of running `npx ...@latest` to avoid unexpected package changes; 3) confirm how any CSV or private data will be supplied to the agent so you don't unintentionally upload sensitive files to the render API; 4) review Emergence's privacy/security docs and the OpenAPI endpoint (https://api.emergence.science/tools/render/openapi.json) to verify what data is logged/retained; and 5) monitor logs or outputs for accidental leakage of the API key or other sensitive content. If you need stronger guarantees, run the rendering step in an isolated environment or use a short-lived, minimally scoped API key.

Review Dimensions

Purpose & Capability: okThe name/description (Marp + Emergence Render API) aligns with the actual requirements and instructions. The only required environment variable is EMERGENCE_API_KEY, which is exactly what the rendering endpoint needs. No unrelated credentials, binaries, or config paths are requested.
Instruction Scope: noteSKILL.md instructs the agent to: 1) interactively build a Marp markdown file, 2) POST diagram code to https://api.emergence.science/tools/render with an Authorization: Bearer $EMERGENCE_API_KEY, 3) save returned base64 images to assets/, and 4) run the Marp CLI (npx) to compile outputs. These steps stay within the presentation-generation workflow. Minor concerns: the guidance that EMERGENCE_API_KEY is "securely transmitted only to the rendering endpoint" is a claim the skill cannot itself enforce — the agent/runtime could accidentally log or transmit the key elsewhere. The workflow references data visualizations from CSVs but does not specify how CSVs are supplied or accessed; ensure the agent is not instructed (or given permission) to read arbitrary local data you don't want shared.
Install Mechanism: noteThis is an instruction-only skill (no install spec, no code written to disk by the skill). That is low-risk. One caveat: the compilation step uses `npx @marp-team/marp-cli@latest`, which will fetch and execute remote package code at runtime; this is reasonable for converting Marp markdown but carries typical supply-chain/runtime execution risk. If you prefer, install a pinned marp-cli version locally instead of using npx with @latest.
Credentials: okOnly EMERGENCE_API_KEY is declared as required and as the primary credential, which is appropriate for a tool that POSTs diagram code to the Emergence rendering API. No other secrets or unrelated env vars are requested.
Persistence & Privilege: okThe skill does not request always:true or other elevated persistence. It does not ask to modify agent-wide configuration or other skills. It writes outputs to presentation.md and an assets/ directory as part of normal operation, which is expected.