debate-research
v1.0.0Multi-perspective structured debate for complex topics. Spawn parallel subagents with opposing stances, cross-inject arguments for rebuttal, then synthesize...
⭐ 1· 34·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description match the runtime instructions: the SKILL.md describes spawning parallel subagents, cross-injecting outputs, and synthesizing a report. Required capabilities (ability to spawn sessions, yield results, and a web_search tool for Phase 1) are consistent with that purpose.
Instruction Scope
Instructions stay within the stated debate/research scope. They explicitly enable web_search for Phase 1 and disable it for later phases (intentional to prevent information drift). The skill passes subagent outputs between roles and can optionally write the assembled report to an output_path; both behaviors are expected but mean the skill will fetch external web content (when web_search is available) and may write to filesystem if you provide a path. Confirm-plan default is true, which helps limit unexpected runs.
Install Mechanism
No install spec or code is included (instruction-only). No downloads, packages, or binaries are required by the skill itself — lowest install risk.
Credentials
The skill requests no environment variables, credentials, or config paths. It does rely on configured LLM providers and a web_search tool (mentioned in README and SKILL.md) but the registry metadata does not explicitly list 'web_search' as a required tool — minor metadata omission but not a security inconsistency with the skill's purpose.
Persistence & Privilege
No elevated persistence requested (always: false). The skill does not modify other skills or system-wide settings. Autonomous invocation is allowed (platform default) but this is expected for a skill that spawns subagents.
Assessment
This skill is internally consistent: it orchestrates multiple subagents, uses web_search during initial evidence collection, then disables web_search for rebuttals and judgment to prevent information drift. Before using it: (1) ensure you trust the configured LLM provider(s) because the skill will spawn sessions on those models; (2) be aware that Phase 1 can fetch external web content via the platform's web_search tool — avoid feeding sensitive or private topics if you don't want external queries; (3) if you plan to write the report to disk, supply a safe output_path and verify the agent has only the intended file permissions; (4) enable confirm_plan (default true) so you can review the planned roles/models/count of subagent calls before execution. The only minor inconsistency is that the README/SKILL.md require a web_search tool but the registry metadata doesn't explicitly declare that dependency — confirm your environment provides web_search before running.Like a lobster shell, security has layers — review code before you run it.
latestvk97415dzvw7sqtsdgj17xxsgmh841z53
License
MIT-0
Free to use, modify, and redistribute. No attribution required.