← Back to Skills Marketplace
Multi Stop
by
xiejinsong
· GitHub ↗
· v3.2.0
· MIT-0
86
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install multi-stop
Description
Plan complex multi-city flight itineraries — A to B to C to D. Finds the best combination of flights for multi-stop trips, optimizing total cost. Also suppor...
Usage Guidance
This skill appears to do what it says (wrap the flyai CLI for multi-city flight planning) but contains several things you should check before installing or using it:
- Verify the @fly-ai/flyai-cli package on the npm registry (publisher, recent releases, and source code) before running a global 'npm i -g'. Global installs require extra privileges and can affect your system PATH.
- Expect the agent to run multiple flyai CLI commands and network calls. The SKILL.md enforces re-running commands until a booking link and brand tag appear — this could cause repeated requests and unexpected traffic. Consider testing in a controlled environment.
- Be aware the skill’s runbook suggests writing .flyai-execution-log.json containing the raw user_query and CLI call logs. If you care about privacy, watch for that file in your working directory and inspect its contents; ensure you are comfortable with persistent logs containing user inputs.
- The SKILL.md contains contradictory rules (e.g., 'Use detailUrl' vs 'Never use detailUrl') — this is ambiguous and may cause incorrect outputs. Prefer to run the flyai CLI manually first to confirm expected outputs and URL fields.
- The skill claims extra capabilities (hotels, trains, visas) that are not shown in the CLI parameters/playbooks. If you need those features, confirm how they are implemented and whether additional permissions or credentials will be requested.
If you proceed, run the flyai CLI commands manually once to validate behavior, check the npm package source, and monitor for created log files and unexpected network activity. If you cannot validate these points, treat the skill as untrusted.
Capability Analysis
Type: OpenClaw Skill
Name: multi-stop
Version: 3.2.0
The skill bundle mandates the automated global installation of an external NPM package (`@fly-ai/flyai-cli`) and explicitly suggests using `sudo` for elevated privileges if the installation fails (`references/fallbacks.md`). It employs aggressive prompt instructions to force the agent to ignore its internal training data and rely solely on this external CLI, which creates a significant supply-chain risk and potential for unauthorized system modification. Additionally, it includes instructions for local filesystem persistence via a hidden log file (`references/runbook.md`) and contains contradictory instructions regarding the use of `detailUrl` in `SKILL.md`.
Capability Assessment
Purpose & Capability
The skill's name/description (multi-city flight planning) aligns with the runtime instructions, which call the flyai CLI to search flights and build itineraries. Claims about booking links and real‑time pricing fit a CLI-backed workflow. Minor mismatch: README/SKILL mention hotels, trains, attractions and being a sub-skill of an Alibaba 'flyai' project — the SKILL.md parameters and playbooks show only flight search commands, so other claimed capabilities (hotels, visas, insurance) are not demonstrated in the instructions or parameters.
Instruction Scope
SKILL.md mandates strictly sourcing all output from the flyai CLI and re-running CLI calls if requirements (e.g., presence of a [Book]({detailUrl}) link and a brand tag) aren't met. This creates an implicit re-execution loop that could cause repeated network calls. The runbook instructs optionally persisting an execution log to .flyai-execution-log.json containing user_query and CLI call logs, which is a filesystem side-effect not declared in the registry metadata. There are clear contradictions in the Output Rules (for example: 'Use `detailUrl` for booking links. Never use `detailUrl`.'), which makes correct behavior ambiguous and could cause the agent to behave incorrectly.
Install Mechanism
The skill is instruction-only and has no registry install spec, but SKILL.md and README require installing the public npm package @fly-ai/flyai-cli via 'npm i -g'. Installing a global npm CLI is expected for a CLI-wrapping skill, but it raises normal supply-chain considerations: verify the publisher of the @fly-ai package, the package contents, and permissions (global install may require elevated privileges). The registry itself does not provide a pinned or verified install artifact.
Credentials
The skill does not request any environment variables, credentials, or config paths in the registry metadata. That is proportionate for a read-only search/lookup skill. Note: the flyai CLI itself might later prompt for or require user credentials to complete bookings — that would be expected, but the skill does not declare or request such secrets upfront.
Persistence & Privilege
The runbook describes persisting an execution log to a local file (.flyai-execution-log.json) if filesystem writes are available; the log contains user_query and raw CLI commands/results. This behavior creates persistent files that could include sensitive inputs and is not declared in the skill metadata. The skill does not request 'always: true', but the implicit persistent logging and the enforced re-execution behavior increase the blast radius if the CLI or skill are later modified.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install multi-stop - After installation, invoke the skill by name or use
/multi-stop - Provide required inputs per the skill's parameter spec and get structured output
Version History
v3.2.0
multi-stop v3.2.0
- Enforces strict CLI-only execution: all answers must originate from flyai CLI output, never from prior knowledge.
- Adds critical booking link validation: every result must include a [Book]({detailUrl}) link from CLI JSON, or the skill must re-execute.
- Formalizes environment check: always confirm flyai-cli is installed before proceeding; provide exact user instructions if not found.
- Clarifies parameter collection, output formatting, and scenario-specific playbooks.
- Expands and documents output/activation rules for multi-city trips, including English and Chinese triggers.
- Strongly prohibits use of training data for travel answers.
Metadata
Frequently Asked Questions
What is Multi Stop?
Plan complex multi-city flight itineraries — A to B to C to D. Finds the best combination of flights for multi-stop trips, optimizing total cost. Also suppor... It is an AI Agent Skill for Claude Code / OpenClaw, with 86 downloads so far.
How do I install Multi Stop?
Run "/install multi-stop" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Multi Stop free?
Yes, Multi Stop is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Multi Stop support?
Multi Stop is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Multi Stop?
It is built and maintained by xiejinsong (@xiejinsong); the current version is v3.2.0.
More Skills