Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Official-source-led manual for the TUI `claude` template.
.claude/skills/lingtai-ai-preset-skill-claude/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | -30% | 0% |
| case-02 | ✗→✓ | ▲ Improved | -23% | 0% |
| case-03 | ✗→✓ | ▲ Improved | -32% | 0% |
| case-07 | ✗→✓ | ▲ Improved | -62% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -68% | 0% |
claudeclaudePreset() in tui/internal/preset/preset.go uses canonical provider claude-code, displayed in the TUI as print-mode backend claude-p. It reuses the current Claude Code OAuth login and shows the account email reported by claude auth status --json; the TUI never reads or stores Claude credentials. The default model is opus, and the editor also offers fable (the current Claude Code CLI maps it to full model id claude-fable-5), sonnet, and haiku.
The shipped manifest intentionally has no API-key env-var, base URL, web-search override, or LingTai vision capability.
The underlying Claude models and interactive Claude Code may accept images, but that does not establish image forwarding through LingTai's CLI adapter. No separate Claude plan-level vision MCP was established by the reviewed evidence.
Read the official Claude vision guide and Claude Code common workflows when current CLI capabilities matter. If the local CLI route cannot handle an image, report that limitation; do not invent an HTTP endpoint, guess auth, or auto-load/invoke an MCP. Recheck TUI source for the aliases and conservative capability wiring.
For base URL/API-compat/model/capability declaration shape versus credentials, see reference/operations/endpoint-capabilities/SKILL.md.
Other measured skills in the registry, with their headline benchmark lift.