← Back to Skills Marketplace
704
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install q-kdb-code-review
Description
AI-powered code review for Q/kdb+ — catch bugs in the most terse language in finance
Usage Guidance
This skill appears to do what it says (send Q code to an external service for review), but be aware of two actionable risks before installing:
1) External submission of code and keys: The plugin sends your Q code to an external server (Astrai) for analysis. The required ASTRAI_API_KEY and any optional BYOK provider keys you set will be transmitted (the code places them in request headers). If your code or keys are sensitive, do not use this skill against production secrets unless you trust the Astrai service and its privacy claims.
2) Undocumented endpoint override (ASTRAI_BASE_URL): The plugin reads ASTRAI_BASE_URL (default https://as-trai.com/v1) but this env var is not documented in SKILL.md or config.example.toml. If an attacker or misconfiguration sets ASTRAI_BASE_URL to a malicious server, your ASTRAI_API_KEY, BYOK provider keys, and file contents could be sent to that endpoint. Treat ASTRAI_BASE_URL as a high-impact configuration and only set it to trusted hosts.
Recommendations:
- Only provide ASTRAI_API_KEY and any BYOK keys you are willing to send to the external service. Consider creating scoped/ephemeral keys for testing.
- If you require local-only reviews, do not set any BYOK provider keys (the plugin notes local_only when provider keys are absent), and verify that 'local-only' behavior meets your privacy needs (but note the plugin still contacts the Astrai endpoint by default unless ASTRAI_BASE_URL is changed).
- Verify the skill source (the homepage link exists, but owner is unknown). Review plugin.py yourself (it's included) or run it in an isolated environment before granting access to real secrets.
- If you need guarantees that code never leaves your network, do not use this skill; instead use an offline tool or run the plugin pointed at a trusted internal inference endpoint.
I have medium/high confidence in these findings based on the provided files. If you want, I can scan plugin.py line-by-line for any additional suspicious code paths or provide suggested changes to harden the plugin (for example, remove ASTRAI_BASE_URL override or validate it against an allowlist).
Capability Analysis
Type: OpenClaw Skill
Name: q-kdb-code-review
Version: 1.0.0
The skill's primary function is benign: AI-powered Q/kdb+ code review. However, it is classified as suspicious due to its 'Bring Your Own Keys' (BYOK) feature. The `plugin.py` code collects all configured AI provider API keys (e.g., ANTHROPIC_API_KEY, OPENAI_API_KEY) from environment variables and transmits them to the `as-trai.com` endpoint via HTTP headers (`X-Astrai-Provider-Keys`). While this behavior is explicitly documented in `SKILL.md` as part of the inference routing mechanism, centralizing the transmission of multiple sensitive API keys to a single third-party service, even for a stated purpose, introduces a significant trust assumption and a potential single point of compromise for these credentials.
Capability Assessment
Purpose & Capability
Name/description match the implementation: the plugin collects an ASTRAI API key and optional provider keys and sends Q code to an Astrai router for analysis. The declared optional BYOK keys correspond to providers in the code and are reasonable for a routing feature.
Instruction Scope
The SKILL.md and plugin send user Q code to the external Astrai endpoint for analysis (this is expected for a hosted LLM review), but the plugin also reads optional provider keys from many environment variables and includes them in a header. SKILL.md claims 'local processing' for some steps, but the core review sends code externally. Additionally the plugin uses an override env var ASTRAI_BASE_URL (defaults to https://as-trai.com/v1) which is not documented in SKILL.md or config.example.toml; that allows redirecting where code and keys are sent.
Install Mechanism
No install spec or downloads; it's instruction-only plus a single plugin.py file. Nothing is written to disk by an installer and no external archives/third-party packages are pulled during install.
Credentials
The required primary credential (ASTRAI_API_KEY) is proportional to the declared purpose. Optional BYOK provider keys are appropriate for a routing feature, but the plugin collects and forwards them in a header (X-Astrai-Provider-Keys). The undocumented ASTRAI_BASE_URL env var is problematic because it allows redirecting both the ASTRAI_API_KEY and any provider keys and code to an arbitrary endpoint; ASTRAI_BASE_URL is not listed in SKILL.md's Environment Variables table or config.example.toml.
Persistence & Privilege
The skill does not request always:true, does not modify other skills or system configs, and runs only when invoked. It does not install persistent agents or escalate privileges.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install q-kdb-code-review - After installation, invoke the skill by name or use
/q-kdb-code-review - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Initial release: AI-powered Q/kdb+ code review — catch type errors, rank errors, performance issues, and security vulnerabilities in the most terse language in finance
Metadata
Frequently Asked Questions
What is Q Kdb Code Review?
AI-powered code review for Q/kdb+ — catch bugs in the most terse language in finance. It is an AI Agent Skill for Claude Code / OpenClaw, with 704 downloads so far.
How do I install Q Kdb Code Review?
Run "/install q-kdb-code-review" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Q Kdb Code Review free?
Yes, Q Kdb Code Review is completely free (open-source). You can download, install and use it at no cost.
Which platforms does Q Kdb Code Review support?
Q Kdb Code Review is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Q Kdb Code Review?
It is built and maintained by beee003 (@beee003); the current version is v1.0.0.
More Skills