Parallel Deep Research
PassAudited by VirusTotal on May 12, 2026.
Overview
Type: OpenClaw Skill Name: parallel-deep-research Version: 1.0.3 The skill bundle provides instructions for an AI agent to interface with the 'parallel-cli' tool for conducting deep research via the parallel.ai service. The documentation in SKILL.md outlines standard command-line usage, output handling, and integration with OpenClaw's session management without any indicators of malicious intent, data exfiltration, or unauthorized execution.
Findings (0)
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.
Your research question is sent through the Parallel CLI, and commands may create local output files.
The skill instructs the agent to run a local CLI command with user-supplied research questions and options. This is central to the stated purpose, but it is still tool execution that should remain user-directed.
parallel-cli research run "<question>" [options]
Use this only for intended deep-research requests, review the query before running it, and avoid sending confidential information unless you trust the Parallel service and account setup.
Research runs may use your authenticated Parallel account and any associated API quota or billing policy.
The skill expects use of an authenticated Parallel CLI account. That is appropriate for a Parallel API research skill, but it means the agent may consume account/API privileges when invoked.
Requires `parallel-cli` (installed and authenticated).
Confirm the CLI is authenticated to the intended account and understand any usage limits or costs before using high-depth processor tiers.
Installing or using the wrong CLI binary could expose account credentials or produce untrusted behavior outside this skill's reviewed artifacts.
The runnable dependency is an external CLI that is not included in the skill package. This is disclosed and purpose-aligned, but users must obtain it from a trusted source.
If `parallel-cli --version` fails, or if a later command fails with an authentication error, tell the user to see https://docs.parallel.ai/integrations/cli and stop.
Install parallel-cli only from the official Parallel documentation and verify it before authenticating.
Saved research results, which may include sensitive research topics, could be processed in another agent session.
The skill documents an optional handoff where a spawned sub-agent reads a saved local research file. The scope is narrow, but it introduces an inter-agent data flow.
"tool": "sessions_spawn", "task": "Read /tmp/research-<topic>.json and present the executive summary and key findings with sources."
Use the sub-agent workflow only when needed, keep saved reports in appropriate locations, and avoid using it for confidential material unless the user approves.
A research job may continue running after the initial command returns, potentially consuming provider resources until it completes.
The skill can start non-blocking research tasks that continue outside the immediate command flow. This is disclosed and aligned with long research, but users should track task IDs and outputs.
parallel-cli research run "research question" --no-wait # Check status later parallel-cli research status <task-id> # Poll until complete
Use `--no-wait` intentionally, record the task ID, and avoid starting duplicate high-depth research jobs unnecessarily.