Skills/npm/cathelper-elastic-ep

cathelper-elastic-ep@1.0.0

VERIFIED· install only

Publisher claims: "Agent Skill for Elastic-EP fault tolerance on Ascend A3 vLLM: probe fault-tolerance capability, produce fault-tolerant launch parameters, and run retry / scale_down recovery." Observed in sandbox: no network, touches nothing outside the project, writes no memory.

Registry npm (MCP server)Publisher a798347923 · publishing for 12 daysInstalls 386Last detonated 4d ago
$claude mcp add cathelper-elastic-ep -- npx -y @a798347923/cathelper-elastic-ep
WatchReport an issue

Coverage

Nothing executable in the bundle ran; the trace covers the install step only.
Tested
every file read, file written and host contacted by what ran
Not tested
use of the skill by an agent

Is [email protected] safe to install?

KillerSkills ran it in the KillerSkills honeypot sandbox (1 scenario; nothing executable in the bundle ran; the trace covers the install step only) on 2026-11-10 and scored it 82/100 (Verified). In what ran, nothing we could detect went wrong: no undeclared network egress, no credential reads, no persistent instruction writes. What it does with a real credential was not observed; the static findings and the publisher carry the rest of the score. Summary: no network, touches nothing outside the project, writes no memory.

Behavior trace

Honeypot run · 0 s · 1 scenarios · 0 deviations
t+CategoryObserved action
0.0sprocessno install scripts in bundle
0.0sprocessscript backend runs setup only
0.0sprocessscript backend runs setup only
0.0sprocessno install scripts to re-run

Static findings

MEDIUM
Process kill commandSKILL.md:35 · Process kill command — agent can terminate processes**用**:用户想(a)确认某个运行中的 vLLM 服务是否开了容错;(b)知道拉起带容错的 vLLM 服务要传哪些参数;(c)在故障发生后执行重试/缩容——"帮我 retry"、"3 号卡挂了,帮我 scale_down";(d)给服务发一个测试请求,验证推理链路是否正常;(e)kill 掉一个 rank 进程制造故障,演练缩容恢复流程(仅测试环境)。Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Process kill commandscripts/eep.py:1081 · Process kill command — agent can terminate processesp.add_argument("--rank", type=int, default=None, help="specific DP rank to kill (default: random non-dead rank)")Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
npx invocation of non-well-known packagebin/cli.mjs:9 · npx invocation of non-well-known package — may execute untrusted code* npx cathelper-elastic-ep # -> ./.agents/skills/elastic-epFix: Pin versions and sources; vendor anything fetched at install time.
AST06
MEDIUM
Process kill commandscripts/eep.py:806 · Process kill command — agent can terminate processes'echo "__EEP_PIDS__ $pids"; kill -9 $pids && echo "__EEP_KILLED__"'Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
npx invocation of non-well-known packagebin/cli.mjs:11 · npx invocation of non-well-known package — may execute untrusted code* npx cathelper-elastic-ep -d .claude/skills/elastic-epFix: Pin versions and sources; vendor anything fetched at install time.
AST06
MEDIUM
Process kill commandscripts/eep.py:886 · Process kill command — agent can terminate processes"must not run the kill itself without one of them",Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
npx invocation of non-well-known packagebin/cli.mjs:45 · npx invocation of non-well-known package — may execute untrusted codenpx ${pkg.name}Fix: Pin versions and sources; vendor anything fetched at install time.
AST06
MEDIUM
Process kill commandscripts/test_eep.py:489 · Process kill command — agent can terminate processescheck=lambda proc, s: len(s.calls) == 1, # no kill dispatchedFix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
npx invocation of non-well-known packagebin/cli.mjs:10 · npx invocation of non-well-known package — may execute untrusted code* npx cathelper-elastic-ep -g # -> ~/.agents/skills/elastic-epFix: Pin versions and sources; vendor anything fetched at install time.
AST06
MEDIUM
Python subprocess execution detectedscripts/test_eep.py:159 · Python subprocess execution detectedreturn subprocess.run(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Process kill commandscripts/eep.py:1077 · Process kill command — agent can terminate processesp = sub.add_parser("kill-rank", help="kill one DP rank process to trigger fault recovery (testing only)")Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Python subprocess execution detectedscripts/eep.py:818 · Python subprocess execution detectedproc = subprocess.run(Fix: Build commands from fixed arguments, never from interpolated strings; prefer an API client over shelling out.
AST05
MEDIUM
Process kill commandscripts/eep.py:12 · Process kill command — agent can terminate processeslayer exists). Never uses --direct for the kill itself.Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Process kill commandSKILL.md:137 · Process kill command — agent can terminate processes…)执行,工具层二次校验(M8–M10);`--ssh` 模式下由 Agent 远程执行 `pgrep -f 'EngineCore_DP<rank>'` + `kill -9`(默认模式 `EngineCore_DP{rank}([[:space:]]|$)`,可用 `--pattern` 覆盖;找不到时输出 `ps` 候选供排…Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
Process kill commandSKILL.md:131 · Process kill command — agent can terminate processes用户要"kill 一个 rank 制造故障、演练缩容"时,**决策门**(顺序固定,完整表见 [decision-rules.md §9](references/decision-rules.md)):Fix: Build commands from fixed arguments; do not interpolate untrusted strings into a shell.
AST05
MEDIUM
npx invocation of non-well-known packagebin/cli.mjs:30 · npx invocation of non-well-known package — may execute untrusted codenpx ${pkg.name} [target] [options]Fix: Pin versions and sources; vendor anything fetched at install time.
AST06

Cost to the agent

Instructions loaded
≈2,890 tok
1 file · 10 in bundle

Tokens estimated at four characters each. Skills load their instruction text when triggered; the figure is what that costs.

Known vulnerabilities

No advisories in OSV for @a798347923/cathelper-elastic-ep@1.0.0 as of 4d ago. This is the package itself, not its dependencies.