Skip to content

Conversation

@ariane-emory
Copy link
Owner

DUMMY PR, IGNORE.

- Add GLM system prompt (session/prompt/glm.txt)
- Update session/system.ts to route GLM models to the new prompt
- Update session/index.ts to extract token usage from metadata (fixes ZAI/Anthropic token counting)
- Add zai-coding-plan to provider priority list
- Add GLM hint to auth selection
- Add specific prompt instructions for ZAI API key
- Add zai-coding-plan to CUSTOM_LOADERS in provider.ts with correct baseURL
- Inject default zai-coding-plan provider definition in models.ts for CLI visibility
- Implement rigorous system prompt for GLM models
- Add 'Ambition vs Precision' heuristic from Codex
- Add detailed workflow from Beast/Anthropic
- Add strict XML constraints for steerability (no emojis, no logs)
- Align with agent-architect skill best practices
- Override npm to @ai-sdk/anthropic for zai-coding-plan (fixes SDK selection from cache)
- Add default thinking options (budgetTokens: 8000) for glm-4.7
- Add thinking variants: none, low (4k), medium (12k), high (24k), max (64k)
- Merge model variants from ModelsDev into provider variants
- Minor cleanup: remove redundant type casts in session
Address code review feedback:
- Use ??= to reduce duplication
- Fix edge case where provider exists but model is missing
- Cleaner, more maintainable code structure
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants