skills.rest
MJ-AgentLab
MJ-AgentLabOFFICIAL
Explore all skills by this author →
GitHub

mj-agent-infra-studio-probe

Runs LangGraph Studio walkthrough probes validating mj-agent happy-path queries and data boundary red lines.

1|Updated Mar 4, 2026
Search Tags:#verification#regression#langgraph#agent testing#probe#studio#data boundary
One-click install
npx skills add https://github.com/MJ-AgentLab/mj-agent --skill mj-agent-infra-studio-probe-mj-agentlab
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: mj-agent-infra-studio-probe
Source: https://github.com/MJ-AgentLab/mj-agent/tree/main/.claude/skills/mj-agent-infra-studio-probe
Command: npx skills add https://github.com/MJ-AgentLab/mj-agent --skill mj-agent-infra-studio-probe-mj-agentlab

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Verifying that the mj-agent data agent behaves correctly end-to-end requires manually starting LangGraph Studio and checking both happy-path analytics queries and data-governance red lines; this Skill standardizes that verification into a repeatable 5-step probe matrix. ## Core Features & Use Cases - Studio Startup & Health Checks: Launches uv run langgraph dev, validates environment prerequisites, and diagnoses common startup failures such as missing API keys or port conflicts. - H1/H2/H3 Happy-Path Probes: Walks through catalog-first table discovery, 7-day trend queries, and Top-N institution joins against the biz_dws domain. - R1/R2 Red-Line Probes: Confirms the agent refuses biz_ods access with an explicit boundary declaration and responds to unbounded export requests with a zero-tool-call clarifying turn. - Use Case: After editing the agent system prompt, run the full probe to regression-test LLM behavior and produce a Probe Report aligned with the dev_studio_walkthrough runbook before self-review. ## Quick Start Ask the agent to start LangGraph Studio and run the full H1/H2/H3/R1/R2 Studio probe for mj-agent, then report the verification matrix results.

Frequently Asked Questions about mj-agent-infra-studio-probe

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I verify LangGraph Studio agent behavior end-to-end?▼

Start Studio with uv run langgraph dev, select the mj_agent graph, then run the H1/H2/H3 happy-path questions and R1/R2 red-line prompts. Compare each trajectory against the expected tool-call sequence and record results in the Probe Report matrix.

How to test data boundary enforcement in a LangGraph SQL agent?▼

Send a red-line prompt requesting a forbidden biz_ods table and an unbounded full-data export. The agent should explicitly declare the governance boundary without executing ODS queries, and respond to the export request with a clarifying question using zero tool calls.

What prerequisites are needed before running a Studio probe?▼

The .env file must be configured with Postgres analyst credentials and an LLM API key, uv sync must be complete, and uv run mj-agent check must pass. If using a local OpenAI-compatible endpoint, run the LLM endpoint probe first.

Why does LangGraph Studio fail to start with LLMConfigError?▼

LLMConfigError means the ARK_API_KEY or LLM_BASE_URL is missing from the environment. Confirm you are running from the worktree root, that .env exists, and rerun the environment setup script to inject team credentials.

When should I not run the red-line boundary probes?▼

Do not run R1/R2 probes on test or production profiles because they deliberately attempt to access restricted ODS tables. Run them only on the dev profile, and keep LangSmith tracing enabled so any boundary breach can be root-caused.

Related Skills

Explore More Software Engineering Skills→
IPADS-SAIIPADS-SAI
44

setup

Install and configure the MobiClaw project environment with dependency and gateway setup.

Official
Advanced
superluminar-iosuperluminar-io
10

api-change-guardian

Detect API contract changes against a published registry spec and classify version bumps.

Official
Advanced
mohitmishra786mohitmishra786

test

Generate unit, integration, and end-to-end tests with Pytest, Jest, Playwright, and k6.

Community
Advanced
artgaurav16420-ossartgaurav16420-oss

langsmith-fetch

Fetch and analyze recent LangSmith traces to debug LangChain and LangGraph agents.

Community
Intermediate