How do I install the Fi Mesh Check skill?

Run npx skillmds@latest add ariffazil/fi-mesh-check 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 Mesh Check safe to use?

This skill has not completed SkillMD's automated safety review yet. Capability flags: docs only. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Fi Mesh Check?

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 Mesh Check free to use?

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

Who published Fi Mesh Check?

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

Fi Mesh Check by ariffazil — Agent Skill

Name: Fi Mesh Check
Rating: 4.2 (5 reviews)
Author: ariffazil

Fi Mesh Check

Probe all FI coder CLIs live (qwen/kimi/opencode/codex/claude/grok/gemini) with a minimal falsification prompt and report the mesh matrix. Use when Arif says 'mesh check', 'test all coder CLIs', 'FI mesh health', 'are all harnesses alive', or before/after cross-FI plumbing work.

ariffazil Updated Sep 10, 2026

File contents

FI Mesh Check — falsification probe

Probe every FI coder CLI in the federation with a minimal falsification prompt. For each: run the CLI's one-shot mode asking it to reply with an exact marker string, capture output tail, and classify PASS / FAIL / EXTERNAL.

Current invocations (verify against disk before trusting — these drift)

Kimi: kimi -m zai-coding-plan/glm-5.3 -p "Reply with exactly: KIMI-MESH-OK"
OpenCode: opencode run --model litellm-federation/forge-777 "Reply with exactly: OPENCODE-MESH-OK"
Codex: codex exec --skip-git-repo-check "Reply with exactly: CODEX-MESH-OK" (goes through :4010 middleware)
Grok: grok -p "..." (402 = balance exhausted → EXTERNAL flag, not a mesh defect)
Gemini: GEMINI_CLI_TRUST_WORKSPACE=true gemini -p "..." (429 = quota → EXTERNAL)

Rules

Timeout each probe at 90s; one retry only for transient network.
Never mask an error as pass. One marker mismatch = FAIL.
Classify external billing/quota failures (402/429) separately from mesh defects — EXTERNAL ≠ FAIL.
Probe the correct layer before claiming absence: 401 on a health endpoint = service UP (auth-gated); conn-refused/timeout = DOWN.
When a CLI reports "high demand" / load errors, probe its endpoint directly — middleware faults masquerade as vendor load (scar: :4010 responses-path, 2026-08-21).

Output

End with the matrix table (FI × verdict × latency × hcsvog × note) + any root-cause fix performed. Root-cause fixes go to the responsible FI lane, not self-assigned cross-lane edits.

hcsvog columns (ETCSOVG harness metadata, arxiv 2605.23950)

Capture per-FI harness identity during probe. Minimum: h_fingerprint + h_tools + h_gov.

Column	Source	Example
h_fingerprint	SHA256-first-8 of harness config	`a3f1c902`
h_tools	Tool availability tier	`full`, `mcp-core`, `minimal`, `none`
h_gov	Governance config	`333:yolo`, `333:confirm`

If harness metadata cannot be determined for an FI, mark as — (never infer).

---name: fi-mesh-checkdescription: Probe all FI coder CLIs live (qwen/kimi/opencode/codex/claude/grok/gemini) with a minimal falsification prompt and report the mesh matrix. Use when Arif says 'mesh check', 'test all coder CLIs', 'FI mesh health', 'are all harnesses alive', or before/after cross-FI plumbing work.---# FI Mesh Check — falsification probeProbe every FI coder CLI in the federation with a minimal falsification prompt. For each: run the CLI's one-shot mode asking it to reply with an exact marker string, capture output tail, and classify **PASS / FAIL / EXTERNAL**.## Current invocations (verify against disk before trusting — these drift)- Kimi: `kimi -m zai-coding-plan/glm-5.3 -p "Reply with exactly: KIMI-MESH-OK"`- OpenCode: `opencode run --model litellm-federation/forge-777 "Reply with exactly: OPENCODE-MESH-OK"`- Codex: `codex exec --skip-git-repo-check "Reply with exactly: CODEX-MESH-OK"` (goes through :4010 middleware)- Grok: `grok -p "..."` (402 = balance exhausted → EXTERNAL flag, not a mesh defect)- Gemini: `GEMINI_CLI_TRUST_WORKSPACE=true gemini -p "..."` (429 = quota → EXTERNAL)## Rules- Timeout each probe at 90s; one retry only for transient network.- Never mask an error as pass. One marker mismatch = FAIL.- Classify external billing/quota failures (402/429) separately from mesh defects — EXTERNAL ≠ FAIL.- Probe the correct layer before claiming absence: 401 on a health endpoint = service UP (auth-gated); conn-refused/timeout = DOWN.- When a CLI reports "high demand" / load errors, probe its endpoint directly — middleware faults masquerade as vendor load (scar: :4010 responses-path, 2026-08-21).## OutputEnd with the matrix table (FI × verdict × latency × hcsvog × note) + any root-cause fix performed. Root-cause fixes go to the responsible FI lane, not self-assigned cross-lane edits.### hcsvog columns (ETCSOVG harness metadata, arxiv 2605.23950)Capture per-FI harness identity during probe. Minimum: `h_fingerprint` + `h_tools` + `h_gov`.| Column | Source | Example ||---|---|---|| h_fingerprint | SHA256-first-8 of harness config | `a3f1c902` || h_tools | Tool availability tier | `full`, `mcp-core`, `minimal`, `none` || h_gov | Governance config | `333:yolo`, `333:confirm` |If harness metadata cannot be determined for an FI, mark as `—` (never infer).

Frequently asked questions