Parallel Enrichment
PassAudited by ClawScan on May 7, 2026.
Overview
The skill is a coherent Parallel API enrichment guide, but it can send CSV/list data to an external provider and optionally hand results to another agent, so users should review data sensitivity before use.
Before using this skill, confirm that the Parallel CLI is official and authenticated as intended, and only enrich files or inline data you are allowed to share with Parallel. Avoid writing confidential enriched outputs to shared temporary locations, and do not spawn a sub-agent or preview rows in chat if the results contain sensitive personal, lead, customer, or business data.
Findings (3)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
The skill may fail unless the Parallel CLI is already installed, and users need to ensure they are using the legitimate CLI.
The skill relies on an external CLI, while the registry metadata reports no install spec or required binaries. That makes CLI installation, version, and provenance outside the reviewed artifact.
parallel-cli enrich run [options]
Install `parallel-cli` only from Parallel's official source and verify the command/version before using this skill.
Input CSV rows, inline records, and requested enrichment fields may leave the local environment and be processed by Parallel.
The workflow sends user-provided records and enrichment requests to the Parallel API, and the examples include lead/contact enrichment that may involve personal or sensitive business data.
Bulk data enrichment via Parallel API. Adds web-sourced fields... "look up contact info for..."
Use this only with data you are allowed to share with Parallel, and check Parallel's privacy, retention, and compliance terms for regulated or confidential datasets.
Another agent/session may read and summarize the enriched CSV, which could expose sensitive rows or contact information in the conversation or logs.
For large enrichments, the skill suggests handing the enriched output file to a spawned sub-agent. This is bounded to summarization, but it expands access to the enriched data.
"tool": "sessions_spawn", "task": "Read /tmp/enriched-<topic>.csv and summarize the results."
Only spawn a sub-agent for non-sensitive or approved datasets, avoid shared temporary paths for confidential outputs, and clean up temporary files after use.