ClawHub

Prompt Engineering Expert

v1.0.0

Advanced expert in prompt engineering, custom instructions design, and prompt optimization for AI agents

⭐ 66· 14.4k·176 current·183 all-time
byTomsTools@tomstools11
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 skill name and description match the provided files (README, BEST_PRACTICES, TECHNIQUES, EXAMPLES, etc.). All declared capabilities (prompt analysis, generation, refinement, templates, testing guidance) are implemented as documentation and examples — nothing requests unrelated access or capabilities.
ℹ
Instruction Scope
SKILL.md and other docs are instruction-only and constrain the skill to analysis/guidance. They do not instruct the agent to read system secrets, exfiltrate data, or call arbitrary external endpoints. The docs include example snippets referencing Claude/Anthropic APIs, the Files API, and the Agent SDK; those are usage examples, not runtime requirements. If you actually integrate the skill with external services (Files API, Agent SDK), those integrations will require separate credentials and may change the risk profile.
✓
Install Mechanism
No install spec and no code files to execute — lowest install risk. The skill is purely markdown documentation and will not download or install binaries when uploaded as-is.
✓
Credentials
The skill declares no required environment variables, credentials, or config paths. Example code shows an Anthropic client call but no credential is required by the skill itself. No disproportionate secret access is requested.
✓
Persistence & Privilege
always:false (default) and model invocation allowed — both are normal. The skill does not request elevated or persistent system privileges and does not modify other skills or system-wide settings.
Assessment
This skill is a documentation-only prompt engineering toolkit and appears internally consistent. Before installing: 1) Understand that the skill itself does not request credentials, but using its examples with the Agent SDK, Files API, or Claude/Anthropic APIs will require you to supply keys — only provide those in trusted environments. 2) Review the example code and templates to ensure they don't embed any sensitive data or hardcoded secrets before uploading. 3) Treat the outputs as expert guidance — test refined prompts in your environment before relying on them for critical tasks. 4) If you extend the skill (add code or install steps), re-evaluate for downloads, network endpoints, or env‑var requirements that could change its risk profile.

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

claudevk9766fy4nq4we8y7gb395czz7x7zx3bmlatestvk9766fy4nq4we8y7gb395czz7x7zx3bmpromptvk9766fy4nq4we8y7gb395czz7x7zx3bmprompt engineeringvk9766fy4nq4we8y7gb395czz7x7zx3bm

License

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

Comments