Tools Litiao

Learns your tool preferences while staying capable of using anything. Adapts to your stack.

Security Scan

Instruction-only skill that adapts tool recommendations based on user-provided preferences; it requests no credentials, has no install steps, and its runtime instructions stay within the stated purpose, though package metadata is inconsistent and you should confirm how/where preferences are stored.

Purpose & Capability

The name and description (learning/adapting tool preferences) match the SKILL.md guidance: it asks the agent to prefer known tools, suggest alternatives, and learn new tools. No unrelated binaries, env vars, or config paths are required. Minor inconsistency: bundled _meta.json uses a different ownerId and slug ("tools") than the registry metadata ("tools-litiao") and the skill's declared owner — this could be an innocuous packaging oversight but is worth checking.

Instruction Scope

SKILL.md only instructs the agent to ask about and record user tool preferences and to consult included criteria/dimensions files when proposing tools. It does not instruct reading arbitrary files, using external endpoints, or accessing environment variables. It does not, however, document where preferences are persisted or whether they're transmitted externally — that omission should be clarified before trusting sensitive info.

Install Mechanism

There is no install spec and no code files; this is instruction-only, so nothing is written to disk or downloaded. That is the lowest-risk install profile.

Credentials

The skill declares no required environment variables, credentials, or config paths. That matches its stated functionality (tracking user preferences and recommending tools).

Persistence & Privilege

always:false (normal) and agent invocation is allowed (platform default). The skill's behavior implies it will 'track' preferences — likely stored in agent memory/state — but SKILL.md does not specify storage location, retention, or external transmission. Confirm storage/retention policies if you care about privacy.

Assessment

This skill is mostly coherent and low-risk: it asks for no credentials and has no install. Before enabling it, do these simple checks: 1) Ask the publisher/owner which account published this skill (the package metadata has a different owner/slug than the registry metadata). 2) Ask where preference data will be stored (agent memory? local only? sent to an external service?) and how long it is retained or how to delete it. 3) Test with non-sensitive example preferences first (do not paste API keys, passwords, or private config into the 'Stack' or 'Preferences' fields). 4) If you have privacy concerns, limit the skill's autonomy or invoke it manually rather than allowing autonomous runs. If the owner or storage answers are unclear, treat the metadata mismatch as a minor red flag and proceed cautiously.

Adaptive Tool Intelligence

You can use ANY tool. Learn new ones instantly. This skill tracks user preferences, not your limits.

Rules:

• Default to user's known tools when they exist
• Propose better alternatives when significantly better for the task
• Learn new tools on demand without friction
• Check dimensions.md for categories, criteria.md for when to suggest new

---

Stack

<!-- Tools user actively uses. Format: "category: tool" -->

Preferences

<!-- When to use what. Format: "context: tool or approach" -->

Open To

<!-- User's appetite for new tools. Format: trait -->

Avoid

<!-- Tools user rejected or dislikes -->

---

Remember: Empty stack ≠ limited. Ask or propose based on task. User preferences guide choices, not restrict capabilities.