feat: add AgentSkill type and skills support to generate methods

[devin-ai-integration]

feat: add AgentSkill type and skills support to generate methods

Summary

Adds support for skill-based requests as an alternative to domain-based requests, aligning the Node SDK with vlm-lab#1621.

Types (types.ts):

• New AgentSkill class, AgentSkillParams interface, and AgentSkillInput union type with skillId, skillName, version, and type fields
• domain made optional in PredictionGenerateParams
• skills field added to GenerationConfigParams, GenerationConfig, AgentExecutionConfigParams, AgentExecutionConfig, AgentCreationConfigParams, AgentCreationConfig
• All config toJSON() methods serialize skills with snake_case keys for API compatibility

Predictions (predictions.ts):

• ImagePredictions.generate() and FilePredictions.generate() now validate that either domain or config.skills is provided
• domain is conditionally included in the request payload (omitted when undefined)
• _castResponseToSchema is only called when domain is present

Review & Testing Checklist for Human

• Verify AgentSkill constructor handles edge cases: Object.assign(this, params) can overwrite class defaults with undefined if params explicitly contains { type: undefined } or { version: undefined }. Consider whether this needs a guard.
• Verify skills: undefined in request payload is acceptable: When no skills are provided, serializedSkills is undefined and gets sent as skills: undefined inside the config object. Confirm the backend handles this (vs. omitting the key entirely).
• End-to-end test with a real skill: Call image.generate() and document.generate() with config.skills instead of domain and confirm the backend resolves the skill and returns a valid response.
• Verify backward compatibility: Existing callers passing domain without skills should continue to work identically.
• Confirm _castResponseToSchema skip is intentional: When using skills without domain, schema casting is now skipped. Verify this is the desired behavior for skill-based responses.

Notes

• No new tests were added — the pre-existing TS errors in chat-completions.test.ts are unrelated.
• Companion PRs: Python SDK and docs are being updated in parallel.
• Requested by: dinesh@vlm.run
• Link to Devin run: https://app.devin.ai/sessions/6d9ce36452de49339048a051a670858e

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment and CI monitoring