ClawHub

Deep Research

v0.1.0

Automated deep research that performs comprehensive multi-source investigation and produces detailed reports with citations. Use when user requests research,...

⭐ 0· 218·2 current·2 all-time
by@bird-frank
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
high confidence
✓
Purpose & Capability
The name/description (deep research, plan + execute) match the SKILL.md steps. The skill requests no binaries, env vars, or installs that would be unrelated to research.
ℹ
Instruction Scope
Phase 1 is collaborative and must wait for explicit user confirmation (good). Phase 2 instructs the coordinator to spawn a sub-agent (session_spawn) and let it 'independently decide how to search' and 'dynamically add searches' — this is coherent for autonomous research but grants the spawned agent broad discretion to access web sources and create/collect data. The SKILL.md also directs saving plans to plans/research-plan-{timestamp}.json and references a tool name 'deep-research-executor' that is not included here (availability is unknown).
✓
Install Mechanism
Instruction-only skill with no install spec and no code files. Lowest install risk. It will write research-plan JSON files to a local plans/ path during normal operation.
✓
Credentials
No environment variables, credentials, or config paths are required or requested; that is proportionate to an instruction-only research coordinator.
ℹ
Persistence & Privilege
always:false and user-invocable (normal). However the skill instructs creation of sub-agents via session_spawn which increases the operational blast radius depending on the platform's sub-agent privileges. The skill itself does not request persistent system-wide privileges.
Assessment
This skill appears coherent and doesn't ask for credentials or installs, but before enabling it consider: 1) The skill will spawn a sub-agent that can browse external sources and create files (plans/*.json). Confirm you are comfortable with that agent's web/network/file permissions. 2) The SKILL.md references a tool ('deep-research-executor') and the session_spawn capability — verify those tools exist and what privileges they have on your platform. 3) Keep an eye on the created plans/ files and on outbound queries the sub-agent performs (to avoid accidental data leakage). 4) If you want tighter control, require the coordinator to present the final list of exact sources or deny session_spawn/autonomous browsing before allowing Phase 2.

Like a lobster shell, security has layers — review code before you run it.

latestvk974zfeb1pe3ne9dr3d3tjcbhn836gyw

License

MIT-0
Free to use, modify, and redistribute. No attribution required.
Termshttps://spdx.org/licenses/MIT-0.html

Comments