ClawHub

Guided Conversation

v1.0.0

Minimize user friction by asking up to 4 clarifying questions before executing actions.

⭐ 0· 46·1 current·1 all-time
by@szpili
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
Name/description (guided clarification) match the SKILL.md: no extra binaries, env vars, or installs are requested and nothing outside conversational guidance is required.
ℹ
Instruction Scope
Instructions are limited to pre-action clarification and summarization, which is in-scope. Note: the policy 'if still ambiguous after 4 questions, make best-effort interpretation and proceed' increases risk of unintended actions; this is a design choice (usability vs safety) rather than an incoherence. Recommend enforcing requireConfirmationHighRisk and explicit safeguards for destructive or external-facing operations.
✓
Install Mechanism
No install spec and no code files — lowest-risk form (instruction-only). Nothing will be written to disk or downloaded by the skill itself.
✓
Credentials
The skill requests no environment variables, credentials, or config paths. Declared configuration options (maxQuestions, autoProceedAfterClarity, requireConfirmationHighRisk) are appropriate and proportional to the purpose.
✓
Persistence & Privilege
always:false (default) and no persistent modifications or system-wide config changes are requested. Autonomous invocation is allowed by platform default but is not unusually privileged for this skill.
Assessment
This skill is coherent and low-risk because it only provides conversational instructions and asks for no secrets or installs. Before enabling: (1) enable requireConfirmationHighRisk (or set autoProceedAfterClarity=false) so high-impact actions always require explicit user consent; (2) consider lowering auto-proceed behavior or increasing maxQuestions if you prefer safety over convenience; (3) test the skill in a safe environment to ensure its 'best-effort' behavior doesn't perform unwanted actions; (4) verify it does not prompt for or transmit sensitive info (passwords, keys) and that it respects existing context/memory as promised; (5) review how it interacts with other skills (e.g., tools that perform external actions) to ensure confirmations are honored.

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

latestvk97e6dmzjs8bmfaxfcczejscax83zgkd

License

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

Runtime requirements

🗣️ Clawdis

Comments