How do I install the Fi Zai Probe skill?

Run npx skillmds@latest add ariffazil/fi-zai-probe in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Fi Zai Probe safe to use?

This skill has not completed SkillMD's automated safety review yet. Capability flags: makes network calls, reads secrets. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Fi Zai Probe?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Fi Zai Probe free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Fi Zai Probe?

ariffazil (@ariffazil) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Fi Zai Probe by ariffazil — Agent Skill

Name: Fi Zai Probe
Rating: 4.5 (4 reviews)
Author: ariffazil

Fi Zai Probe

Probe Z.AI GLM coding plan health — models list, anthropic + openai endpoints, credit reality, cross-check 5 wired surfaces. Use when Arif says 'zai probe', 'check Z.AI plan', 'is GLM alive', 'zai health', or when any zai-backed harness (Qwen/Kimi/OpenCode/Claude) behaves oddly.

ariffazil Updated Sep 10, 2026

File contents

Z.AI Plan Probe

Governance: Irreversible infra ops (deploy/destroy/spend/credential changes) route via arifos-governance; irreversible actions require 888_HOLD release. (F1 AMANAH / F13 SOVEREIGN)

Verify the Z.AI GLM Coding Plan is alive and correctly wired. Key lives in env ZAI_API_KEY (vault: /root/.secrets/kunci-root.env; mode-600 configs only — never paste keys into chat, receipts, or configs beyond mode-600).

Probes (all live, F2 OBS)

Plan models: GET https://api.z.ai/api/coding/paas/v4/models (Bearer key) — expect glm-4.5 … glm-5.3 (9 models).
Anthropic-compat (Kimi path): POST https://api.z.ai/api/anthropic/v1/messages with x-api-key + anthropic-version: 2023-06-01, model glm-5.3 — expect HTTP 200.
OpenAI-compat (Qwen path): chat completion at /api/coding/paas/v4/chat/completions, model GLM-5.3 — expect 200.
Cross-check wired surfaces: Qwen settings.json [Z.AI] entries, Kimi config.toml zai-coding-plan provider, OpenCode zai-direct, Claude ZAI_DIRECT_* env — all must reference the same endpoints.
FED SOT: /root/.config/federation-models.json → provider zai-coding-plan must carry anthropic_endpoint_url + live telemetry.

Gotchas (scars, not guesses)

glm-5.2 requests silently return model=glm-5.3 — plan fact, not a bug.
Peak Mon–Fri 14:00–18:00 SGT burns credits at 2×; off-peak 50% off.
If the anthropic endpoint 401s from Kimi: kimi-code does NOT expand ${ENV} in provider api_key — literal key required in a mode-600 file.

Output

Report: endpoint statuses, model count, any drift between the 5 wired surfaces.

Frequently asked questions