LinkSKILL — Universal Platform API Integration#

LinkSKILL is a config-driven API integration skill for enterprise platforms.

Given a Swagger/OpenAPI endpoint and auth settings, it can discover APIs, authenticate, and execute requests with minimal manual setup.

When to Use This Skill#

Use LinkSKILL when you need to:

• Integrate with a new internal/external platform quickly.
• Explore unknown APIs from Swagger/OpenAPI docs.
• Standardize auth + request execution across multiple platforms.
• Run repeatable API calls from CLI without custom one-off scripts.

Core Workflow (4-Step Closed Loop)#

1. Identify platform

   Read scripts/platform_config.json and select active_platform (or pass --platform explicitly).

2. Discover APIs

   python scripts/swagger_loader.py --config scripts/platform_config.json --mode list

3. Authenticate

   python scripts/auth_manager.py --config scripts/platform_config.json

4. Execute request

   python scripts/http_request_tool.py --config scripts/platform_config.json --method POST --endpoint "/api/xxx" --json-body '...'

Tip: Use search and detail before execution to reduce trial-and-error:

• --mode search --keyword "keyword"
• --mode detail --path "/api/xxx" --method POST

Tool Responsibilities#

ToolResponsibilityswagger_loader.pyFetch and parse Swagger 2.0 / OpenAPI 3.x; supports list, search, detail, tags, cache``auth_manager.pyHandle auth login/token retrieval and local token cache managementhttp_request_tool.pySend HTTP requests, inject auth headers, and auto re-auth + retry on 401

Configuration Model#

All behavior is driven by scripts/platform_config.json.

Required top-level fields#

• active_platform: default platform ID.
• platforms: map of platform definitions.

Typical platform definition#

• name: human-readable platform name.
• gateway: base URL used by request executor.
• swagger.url: OpenAPI/Swagger endpoint URL.
• swagger.auth_required: whether Swagger endpoint needs auth header.
• auth.type: one of bearer_token, api_key, basic, none.
• default_headers: default request headers.

Auth field reference (for token login)#

• auth.login_endpoint: login path (joined with gateway).
• auth.login_body: login payload.
• auth.token_field: token extraction path (for example data.token).
• auth.token_header: header name (default Authorization).
• auth.token_prefix: header prefix (default Bearer ).

Authentication Strategies#

LinkSKILL supports:

• none: no auth required.
• api_key: static key from config.
• basic: build Basic credential from username/password.
• bearer_token: login API + token extraction + local cache.

Token cache file: scripts/.token_cache.json

Discovery Modes ( swagger_loader.py)#

• list: print all available endpoints.
• search: fuzzy-match endpoints by path/summary/description/tags.
• detail: show parameters, request body schema/example, and response schema.
• tags: show API groups.
• cache: print local swagger cache location.

Swagger cache directory: scripts/.swagger_cache/

Request Execution ( http_request_tool.py)#

Supports:

• Methods: GET, POST, PUT, DELETE, PATCH
• URL styles: full --url or --endpoint + gateway
• Payload: --json-body or --json-file
• Query params: --params
• Extra headers: --headers

On 401, the tool can trigger auth_manager.py --force-login and retry once automatically.

Recommended Operating Pattern#

For stable integrations, run in this order:

1. list or search to locate target endpoint.
2. detail to inspect params/schema.
3. Authenticate and verify token cache.
4. Execute with minimal body first, then iterate.
5. Use cached swagger/token for faster repeated calls.

Troubleshooting#

• Platform not found: check active_platform and platforms keys.
• Swagger fetch failed: verify swagger.url, network, and TLS/headers.
• Token extraction failed: correct auth.token_field based on real login response.
• 401 on request: verify token TTL/login payload and header prefix.
• Non-JSON response: inspect raw response text and endpoint behavior.

Onboarding a New Platform#

1. Add a new platform under platforms in scripts/platform_config.json.
2. Fill gateway, swagger, auth, and optional default_headers.
3. Set active_platform (or use --platform).
4. Run the 4-step workflow end-to-end and validate one successful API call.

Notes and Constraints#

• Designed for CLI-driven API integration workflows.
• Assumes APIs are discoverable via Swagger/OpenAPI endpoint.
• Uses local JSON files for cache; suitable for local/dev automation.