Skip to content
Back to skills

Sdk And Tools

ASecurity

"Use sdk-and-tools for Unstract's shared Python packages, tool

  • 247 stars
  • 0 votes
  • 0 copies
  • 4 views
  • Added September 9, 2026
devopspythonbashtestingapidatabasefrontendbackend

Works with

  • api
  • mcp

Security analysis

A100/100

Pro scans all 3 files and shows the line behind each finding

Scanned September 9, 2026

npx -y skills add VectorSpaceLab/AREX-Skill --skill sdk-and-tools --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Sdk And Tools?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Sdk And Tools
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/vectorspacelab-sdk-and-tools/badge)](https://www.skillsdirectory.com/skills/vectorspacelab-sdk-and-tools)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: "sdk-and-tools"
description: "Use sdk-and-tools for Unstract's shared Python packages, tool
  protocol, tool registry, connectors, and tool authoring workflows."
disable-model-invocation: true
metadata:
  disco-role: operating
license: AGPL 3.0
---

# SDK And Tools

Use this sub-skill when the task is about the reusable Python packages under `unstract/`, the containerized example tools under `tools/`, or the tool-registry / tool-sandbox workflow that connects them.

## Owns

- Shared packages: `unstract/sdk1`, `unstract/core`, `unstract/filesystem`, `unstract/flags`, `unstract/tool-registry`, `unstract/tool-sandbox`, `unstract/workflow-execution`, and `unstract/connectors`.
- Tool authoring under `tools/classifier` and `tools/text_extractor`.
- Tool protocol, tool metadata files, runtime-variable schema, and registry-backed tool loading.
- Shared tool / SDK inspection that does not belong to a specific service or route family.

## Excludes

- Backend API routing and hosted MCP details — use `backend-platform`.
- Full-stack deployment / bootstrap — use `platform-deployment`.
- Worker queue and Celery / PG orchestration — use `workers`.
- Frontend routing or runtime config — use `frontend`.
- Repo test-group manifests and critical-path reporting — use `testing-rig`.

## Start Here

Read `references/package-and-tooling.md` first when the task involves:

- shared package versions or import surfaces,
- the tool protocol or tool metadata schema,
- `tool-registry` / `tool-sandbox` configuration,
- or the classifier / text-extractor example tools.

Read `references/troubleshooting.md` when the issue is a missing optional dependency, registry config failure, tool metadata mismatch, or tool-container runtime problem.

For a safe package-level smoke check, use the root checker from this skill tree:

```bash
python ../../scripts/check_unstract_packages.py
python ../../scripts/check_unstract_packages.py --tool-registry --tool-registry-config <config-dir>
```

## Shared References

- `references/package-and-tooling.md` — shared package map, tool protocol, registry, and example tool guidance.
- `references/troubleshooting.md` — install, optional dependency, registry, and container-tool failures.
- `../../references/package-layout.md` — repo-wide package and service ownership map.
- `../../references/installation-and-env.md` — install and environment matrix for SDK / tool workflows.
- `../../references/repo-provenance.md` — source snapshot used to build this skill.
- `../../scripts/check_unstract_packages.py` — shared package and backend smoke checker.

## Common Task Routing

| User request | Read next |
| --- | --- |
| "How do these shared packages fit together?" | `references/package-and-tooling.md` |
| "How do I use the tool protocol?" | `references/package-and-tooling.md` |
| "How do I validate a tool registry config?" | `references/package-and-tooling.md` |
| "Why is a tool container / SDK import failing?" | `references/troubleshooting.md` |
| "What env vars do the example tools need?" | `references/package-and-tooling.md` |

## Safety Boundaries

- Do not assume optional cloud / database / tool-container dependencies are installed.
- Do not treat registry loading as a smoke test if it would pull images or write JSON to disk.
- Do not leak private registry credentials or tool-provider secrets into generated content.

Files in this skill

  • SKILL.md3.3 KB
  • references/package-and-tooling.md3.2 KB
  • references/troubleshooting.md2.3 KB

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.