ClawHub

Agent Native Cli

v1.0.1

Use when designing, reviewing, or refactoring a CLI to reliably serve AI agents alongside humans — including evaluating stdout contracts, schema introspectio...

⭐ 0· 52·0 current·0 all-time
byAgents365.ai@agents365-ai
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
Capability signals
CryptoCan make purchases
These labels describe what authority the skill may exercise. They are separate from suspicious or malicious moderation verdicts.
VirusTotalVirusTotal
Benign
View report →
OpenClawOpenClaw
Benign
high confidence
✓
Purpose & Capability
Name/description match the provided SKILL.md, README, and sidecar metadata: the repository contains guidance for designing and reviewing CLIs for agents. There are no unrelated required binaries, env vars, or config paths declared. The small helper script only validates documentation metadata and is consistent with a documentation/packaging check.
✓
Instruction Scope
SKILL.md contains design rules, examples, and suggested command patterns for making CLIs agent-native; it does not instruct the agent to read arbitrary host files, exfiltrate data, or contact external endpoints. Suggested workflows reference human-set env vars and dry-run semantics as policy, not as secret-gathering steps.
✓
Install Mechanism
No install spec is included (instruction-only skill). README suggests cloning the GitHub repo for local installation, which is normal for skills distributed as source. No downloads from untrusted URLs, no extracted archives, and no package-manager installs are present.
✓
Credentials
The skill declares no required environment variables or credentials. SKILL.md discusses trusted environment variables conceptually (e.g., token/profile/safety mode) but does not request or attempt to read specific secrets in the repository. The sidecar metadata likewise lists no prerequisites.
ℹ
Persistence & Privilege
The skill is not marked always:true. It is model-invocable by default (disable-model-invocation: false), and agents/openai.yaml sets allow_implicit_invocation: true — meaning some agent platforms could invoke this skill implicitly. This is expected for a design/review skill but is worth noting if you prefer to tightly control which skills an agent can call autonomously.
Assessment
This skill appears internally consistent and low-risk: it is instruction-only, asks for no credentials, and contains only documentation and a small metadata-check script. Before installing: 1) confirm you trust the GitHub source (clone URL in README points to the repository listed in the skill); 2) if installing globally, be aware the repo will be placed under a home directory path (e.g., ~/.openclaw, ~/.claude, etc.); 3) check whether your agent platform allows implicit invocation — if you prefer manual control, disable implicit/automatic skill invocation or adjust platform policy; 4) optionally inspect the LICENSE file and the small validate-metadata.sh script locally (it only searches files and enforces metadata strings). If you want higher assurance, run the skill in a non-production agent or sandbox first.

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

latestvk97ca8pxyvhjm3n0qfxfds0avs84gv0j

License

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

Runtime requirements

⌨️ Clawdis
OSmacOS · Linux · Windows

Comments