T08 · Insecure Dependencies
- Location
SKILL_en.md:382- Finding
Unpinned Third-Party Dependency Installation
- Content
View full analysis
- Remediation
View remediation
Security audit
Security checks for vulnerabilities and agentic risk
This is a localized Xiamen tide and beachcombing helper with disclosed scripts and no hidden persistence or data access, but users should verify safety-critical tide advice independently.
Install only if you are comfortable using a local Python script and the `zhdate` dependency. Treat all tide times and beachcombing recommendations as informational; check official tide tables, weather, local warnings, and on-site conditions before going out.
SKILL_en.md:382Unpinned Third-Party Dependency Installation
该 README 全文以中文编写,未见任何语言选择、双语说明或仅限中文用户的明确约束。根据规则,若技能在自然语言层面强制特定语言而没有用户选择或合理、清晰的地域性说明,属于语言/locale 政策风险。
The trigger description includes broad phrases such as '今天适合赶海吗' and '赶海最佳时间' that could match ordinary conversation without sufficiently constraining the request to Xiamen tide/beachcombing intent. Over-broad activation can cause the wrong skill to run, leading to misleading safety-sensitive recommendations about tides and coastal activity.
The file’s natural-language description and all user-facing output are written entirely in Chinese and scoped specifically to Xiamen, with no indication that users may choose another language or locale. Under the policy, forcing a specific language without opt-in is a natural-language policy violation unless the locale restriction is clearly documented and justified.
Earlier sections present the skill as having enhanced features and usage examples consistent with a mature beachcombing assistant, while the notes later state solar-to-lunar conversion must be handled manually and that auto-detection/conversion are future improvements. This is an active contradiction within the file's own documentation about what the skill currently does.
The trigger phrases for beachcombing advice are broad and generic, which could cause the skill to activate on loosely related requests instead of only clear Xiamen tide/beachcombing queries. In an agent environment, overbroad invocation can route user requests to the wrong skill, producing irrelevant advice or suppressing more appropriate tools, which is a prompt-scope and control issue even if not a code-execution flaw.
The manifest claims the skill supports automatic Gregorian-to-lunar conversion, current date auto-detection, leap-month handling, and presents these as available functionality. However, the usage guide says the current version requires lunar date input, manual solar-to-lunar conversion, and lists automatic conversion and current-date detection as future improvements, indicating the documented behavior does not match the manifest's claimed capabilities.
The documentation makes conflicting claims about key features: earlier sections and metadata say Gregorian-to-lunar conversion, automatic current-date detection, and leap-month handling are supported, while the later '注意事项' and '后续改进计划' state these capabilities are not implemented yet. In a safety-relevant tide and shore-foraging skill, users may rely on unsupported automation and receive advice for the wrong date, which can lead to unsafe trip timing around rising tides.
The guide assigns '赶海评分', recommendation levels, time windows, and location/equipment suggestions in a real-world coastal setting, but the referenced section does not clearly state that results are only advisory and must not be relied on for personal safety. Users may over-trust the output for trip planning around tides, terrain, or weather, increasing the risk of injury, stranding, or other safety incidents if the information is inaccurate, incomplete, or stale.
The workflow says the skill runs 'when the user asks about Xiamen tides' but does not define boundaries for ambiguous or partial matches, and later includes mode selection for broad advice requests. This can cause unintended activation on loosely related queries, which is risky because the skill provides action-oriented outdoor safety guidance.
Skill's behavior or capabilities extend beyond its stated purpose. Scope creep allows an agent to perform actions unrelated to its documented functionality, increasing the attack surface.
copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
Skill's behavior or capabilities extend beyond its stated purpose. Scope creep allows an agent to perform actions unrelated to its documented functionality, increasing the attack surface.
copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
The file presents all guidance and output examples in English for a locale-specific tide tool tied to Xiamen and lunar-calendar concepts, without indicating that users may choose another language or that English is an explicit opt-in. Under the policy rule for language or locale constraints, this can be considered a natural-language locale choice omission.
The natural-language content in the module docstring, CLI description, help text, and output strings is Chinese-only. Under the policy, forcing a specific language without user opt-in can be a locale-policy violation unless the restriction is explicitly justified; this file does not provide such opt-in or justification.
The safety reminder line contains mixed-language output ("Watch out for reef areas,小心 slippery surfaces") in an otherwise English usage guide. This introduces a language/locale inconsistency without documenting user opt-in or explaining that multilingual output is intentional.
The multi-day recommendation examples are somewhat vague about geographic and functional scope, which can increase the chance of unintended invocation for generic planning requests. The impact is limited because the skill is informational, but accidental activation can still degrade agent reliability and cause misleading localized recommendations to be applied outside Xiamen.
The file presents the skill as an enhanced, newer-version capability set, but the changelog only documents up to v2.0 and the future-work section still lists headline features as not yet implemented. This kind of version and capability misrepresentation undermines trust and can cause users or integrators to assume the skill is more reliable and complete than it really is.