Low Altitude Guardian
ReviewAudited by ClawScan on May 1, 2026.
Overview
This looks like a local advisory drone-emergency analysis skill with no evidenced network, credential, or flight-control access, but users should not connect its recommendations to real aircraft automation without certified safeguards.
Install only if you understand it is an advisory analysis tool. Keep it separate from real flight-control systems unless reviewed by qualified safety engineers, and periodically audit any local .guardian incident data or learned templates before using them for operational decisions.
Findings (3)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
If treated as more than advice, the skill could encourage automatic emergency actions in a safety-critical drone/eVTOL context.
The human-in-the-loop policy encodes no-confirmation modes for L5/L4 crises and timeout behavior for L3. The artifacts do not show actual flight-control APIs, but this logic matters if an integrator wires the output to real control tools.
"mode": "autonomous_execute", "wait_for_confirm": False, "timeout_s": 0
Use it as advisory analysis only unless a certified flight-safety system and explicit integration controls are in place.
Bad incident data or an unverified learned template could influence later emergency recommendations.
The skill intends to persist and reuse learned incident-derived templates, which can improve future recommendations but can also carry forward incorrect or unreviewed data.
知识库更新：第一性原理推导出的有效方案，自动封装为新模板。每次事件都让系统更强。
Review and approve learned templates, define retention rules for .guardian data, and keep operational incident records access-controlled.
That example command may fail, or a user might be tempted to obtain an unreviewed replacement script elsewhere.
SKILL.md references a helper script that is not included in the provided file manifest, making this a documentation/package mismatch rather than evidence of hidden execution.
python3 scripts/knowledge_updater.py --learn --incident-id <id>
Use only the files included in the reviewed package, or verify any missing helper from a trusted source before running it.