How do I install the Manage Gpt Image Cases skill?

Run npx skillmds@latest add xianyu110/manage-gpt-image-cases 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 Manage Gpt Image Cases safe to use?

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

Which AI agents work with Manage Gpt Image Cases?

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 Manage Gpt Image Cases free to use?

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

Who published Manage Gpt Image Cases?

xianyu110 (@xianyu110) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Manage Gpt Image Cases by xianyu110

Name: Manage Gpt Image Cases
Author: xianyu110

Manage Gpt Image Cases

Build, validate, search, export, and maintain structured GPT image prompt case libraries with stable IDs, localized categories, prompt fingerprints, source attribution, and license metadata. Use when Codex needs to import prompt/image examples, deduplicate cases, query reusable prompts, update a case-library JSON pipeline, or audit provenance and schema compliance.

xianyu110 Updated Sep 17, 2026

File contents

Manage GPT Image Cases

Use one generated case library as the contract shared by sites, agents, and data jobs. Preserve stable IDs and provenance while keeping source-specific parsing outside consumers.

Start Here

Locate the canonical JSON file and its build script. Prefer data/case-library.json and scripts/build_case_library.py when present.
Run validation before reading or editing data:

python3 <skill-dir>/scripts/case_library.py validate --data <repo>/data/case-library.json

Read references/case-schema.md before adding a source, changing fields, or writing an adapter.
Treat generated JSON as read-only. Edit source documents, curated inputs, or source adapters, then rebuild.

Query Cases

Search title, prompt, category, summary, source, and tags:

python3 <skill-dir>/scripts/case_library.py search \
  --data <repo>/data/case-library.json \
  --query "product image" \
  --locale en \
  --limit 10

Narrow by category key or source key when the request is specific:

python3 <skill-dir>/scripts/case_library.py search \
  --data <repo>/data/case-library.json \
  --category products-ecommerce \
  --source awesome-gpt-image-2

Use stats to inspect locale, source, and view counts. Use export --format jsonl to produce machine-friendly subsets on stdout.

Maintain The Library

For repositories with the bundled pipeline:

python3 scripts/build_case_library.py build
python3 scripts/build_case_library.py check
python3 <skill-dir>/scripts/case_library.py validate --data data/case-library.json

When importing a new source:

Confirm a redistribution license and keep the repository, license, original URL, and upstream case URL.
Pin remote assets or source revisions when the upstream format allows it.
Normalize records into the canonical schema. Do not expose source-specific fields to site code.
Compute promptHash from lowercase, whitespace-normalized prompt text.
Deduplicate within a locale. Keep the first stable ID and map discarded IDs through meta.aliases.
Rebuild deterministically and require check to produce no diff.
Update tests and workflow inputs whenever a source contract changes.

Quality Gates

Require all of the following before delivery:

Unique canonical IDs and valid alias targets.
Prompt hashes matching the normalized prompt text.
Complete repository and license metadata for community records.
Root counts matching the actual records and locale views.
No hand edits to generated output.
Site/runtime verification when consumers change.
No secrets, credential-bearing URLs, or unlicensed private media.

If an upstream source becomes incomplete, below its expected minimum, or loses its license, fail the sync instead of shrinking or silently replacing the library.

---name: manage-gpt-image-casesdescription: Build, validate, search, export, and maintain structured GPT image prompt case libraries with stable IDs, localized categories, prompt fingerprints, source attribution, and license metadata. Use when Codex needs to import prompt/image examples, deduplicate cases, query reusable prompts, update a case-library JSON pipeline, or audit provenance and schema compliance.---# Manage GPT Image CasesUse one generated case library as the contract shared by sites, agents, and data jobs. Preserve stable IDs and provenance while keeping source-specific parsing outside consumers.## Start Here1. Locate the canonical JSON file and its build script. Prefer `data/case-library.json` and `scripts/build_case_library.py` when present.2. Run validation before reading or editing data:```bashpython3 <skill-dir>/scripts/case_library.py validate --data <repo>/data/case-library.json```3. Read [references/case-schema.md](references/case-schema.md) before adding a source, changing fields, or writing an adapter.4. Treat generated JSON as read-only. Edit source documents, curated inputs, or source adapters, then rebuild.## Query CasesSearch title, prompt, category, summary, source, and tags:```bashpython3 <skill-dir>/scripts/case_library.py search \ --data <repo>/data/case-library.json \ --query "product image" \ --locale en \ --limit 10```Narrow by category key or source key when the request is specific:```bashpython3 <skill-dir>/scripts/case_library.py search \ --data <repo>/data/case-library.json \ --category products-ecommerce \ --source awesome-gpt-image-2```Use `stats` to inspect locale, source, and view counts. Use `export --format jsonl` to produce machine-friendly subsets on stdout.## Maintain The LibraryFor repositories with the bundled pipeline:```bashpython3 scripts/build_case_library.py buildpython3 scripts/build_case_library.py checkpython3 <skill-dir>/scripts/case_library.py validate --data data/case-library.json```When importing a new source:1. Confirm a redistribution license and keep the repository, license, original URL, and upstream case URL.2. Pin remote assets or source revisions when the upstream format allows it.3. Normalize records into the canonical schema. Do not expose source-specific fields to site code.4. Compute `promptHash` from lowercase, whitespace-normalized prompt text.5. Deduplicate within a locale. Keep the first stable ID and map discarded IDs through `meta.aliases`.6. Rebuild deterministically and require `check` to produce no diff.7. Update tests and workflow inputs whenever a source contract changes.## Quality GatesRequire all of the following before delivery:- Unique canonical IDs and valid alias targets.- Prompt hashes matching the normalized prompt text.- Complete repository and license metadata for community records.- Root counts matching the actual records and locale views.- No hand edits to generated output.- Site/runtime verification when consumers change.- No secrets, credential-bearing URLs, or unlicensed private media.If an upstream source becomes incomplete, below its expected minimum, or loses its license, fail the sync instead of shrinking or silently replacing the library.

Frequently asked questions