How do I install the Model Metadata Schema Verification skill?

Run npx skillmds@latest add holobiomicslab/model-metadata-schema-verification 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 Model Metadata Schema Verification safe to use?

SkillMD's automated safety review verdict for this skill is PASS. Independent scanners report: SkillSpector: PASS, Skill Scanner: PASS. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Model Metadata Schema Verification?

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 Model Metadata Schema Verification free to use?

Yes. Installing skills from SkillMD is free. This skill is licensed under CC-BY-4.

Who published Model Metadata Schema Verification?

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

Model Metadata Schema Verification by holobiomicslab

Name: Model Metadata Schema Verification
Author: HolobiomicsLab

Model Metadata Schema Verification

Use when before submitting peak data or other inputs to a machine learning classification API for the first time, after a model update, or if you encounter unexpected prediction errors. It is essential when the underlying model's input names or structure may change and require code updates.

File contents

model-metadata-schema-verification

Summary

Verify the input schema and model version of a deployed machine learning model by querying its metadata endpoint before constructing and submitting prediction requests. This skill ensures that your request payload matches the model's current input specification, preventing schema mismatch errors and outdated code.

When to use

Apply this skill before submitting peak data or other inputs to a machine learning classification API for the first time, after a model update, or if you encounter unexpected prediction errors. It is essential when the underlying model's input names or structure may change and require code updates.

When NOT to use

The model schema is already documented and confirmed to be static in your deployment.
You are working with a local, non-updated model artifact where the schema is hardcoded and never changes.

Inputs

Model metadata endpoint URL (typically /model/metadata on TensorFlow Serving)
HTTP client or SDK configured to query the TensorFlow Serving instance

Outputs

Parsed JSON metadata object containing model name, version, and input schema
Validated field names and data types (e.g., '1H', '13C' as keys for peak dictionaries)

How to apply

Query the /model/metadata endpoint via TensorFlow Serving to retrieve the current model version and input schema (e.g., expected field names such as '1H' and '13C' for NMR peak data). Parse the returned JSON metadata to confirm that the field names and data types match your intended request payload. If discrepancies are found, update your payload construction logic before submitting any classification requests. This prevalidation step prevents silent failures and ensures compatibility with the deployed model version.

Related tools

TensorFlow Serving (Provides the /model/metadata endpoint to retrieve and verify current model input schema and version before classification requests)

Evaluation signals

HTTP response status code is 200 and returns valid JSON metadata.
Returned metadata contains 'model_spec' and input schema information with expected field names (e.g., '1H', '13C').
Model version in metadata matches the version you intend to use for downstream predictions.
Field names and data types in metadata align with the keys and value types in your planned JSON payload.
Subsequent POST requests to /api/smart3/search execute without schema validation errors.

Limitations

Metadata endpoint availability depends on TensorFlow Serving uptime; if unavailable, you cannot verify the schema before submission.
The metadata endpoint may not return human-readable documentation of the expected value ranges or units for each field.
Metadata verification does not guarantee that the model's predictions will be accurate or meaningful for your specific data; it only confirms structural compatibility.

Evidence

[intro] Model input names may change and require code updates: "If the model input names change, then we need to change it in the code"
[intro] Check model metadata via /model/metadata endpoint: "We pass through tensorflow serving at this url: /model/metadata"
[intro] Expected peak data format with 1H and 13C headers: "You can put in your peaks as a json list of dicts, with 1H,13C as headers"

---name: model-metadata-schema-verificationdescription: Use when before submitting peak data or other inputs to a machine learning classification API for the first time, after a model update, or if you encounter unexpected prediction errors. It is essential when the underlying model's input names or structure may change and require code updates.license: CC-BY-4.0---# model-metadata-schema-verification## SummaryVerify the input schema and model version of a deployed machine learning model by querying its metadata endpoint before constructing and submitting prediction requests. This skill ensures that your request payload matches the model's current input specification, preventing schema mismatch errors and outdated code.## When to useApply this skill before submitting peak data or other inputs to a machine learning classification API for the first time, after a model update, or if you encounter unexpected prediction errors. It is essential when the underlying model's input names or structure may change and require code updates.## When NOT to use- The model schema is already documented and confirmed to be static in your deployment.- You are working with a local, non-updated model artifact where the schema is hardcoded and never changes.## Inputs- Model metadata endpoint URL (typically /model/metadata on TensorFlow Serving)- HTTP client or SDK configured to query the TensorFlow Serving instance## Outputs- Parsed JSON metadata object containing model name, version, and input schema- Validated field names and data types (e.g., '1H', '13C' as keys for peak dictionaries)## How to applyQuery the /model/metadata endpoint via TensorFlow Serving to retrieve the current model version and input schema (e.g., expected field names such as '1H' and '13C' for NMR peak data). Parse the returned JSON metadata to confirm that the field names and data types match your intended request payload. If discrepancies are found, update your payload construction logic before submitting any classification requests. This prevalidation step prevents silent failures and ensures compatibility with the deployed model version.## Related tools- **TensorFlow Serving** (Provides the /model/metadata endpoint to retrieve and verify current model input schema and version before classification requests)## Evaluation signals- HTTP response status code is 200 and returns valid JSON metadata.- Returned metadata contains 'model_spec' and input schema information with expected field names (e.g., '1H', '13C').- Model version in metadata matches the version you intend to use for downstream predictions.- Field names and data types in metadata align with the keys and value types in your planned JSON payload.- Subsequent POST requests to /api/smart3/search execute without schema validation errors.## Limitations- Metadata endpoint availability depends on TensorFlow Serving uptime; if unavailable, you cannot verify the schema before submission.- The metadata endpoint may not return human-readable documentation of the expected value ranges or units for each field.- Metadata verification does not guarantee that the model's predictions will be accurate or meaningful for your specific data; it only confirms structural compatibility.## Evidence- [intro] Model input names may change and require code updates: "If the model input names change, then we need to change it in the code"- [intro] Check model metadata via /model/metadata endpoint: "We pass through tensorflow serving at this url: /model/metadata"- [intro] Expected peak data format with 1H and 13C headers: "You can put in your peaks as a json list of dicts, with 1H,13C as headers"

Frequently asked questions