v0.1.12

Bug fixes

• Fix --append workflow: The setup command's --append flag was blocked by the file-existence check, causing it to exit before reaching the append logic. The documented --append workflow now works correctly for all agents.

Skill improvements

• Fix Python version in skill metadata: Corrected compatibility from "Python 3.8+" to "Python 3.10+" to match pyproject.toml's requires-python = ">=3.10".
• Clarify analyze command exclusion: Added parenthetical explaining that the analyze command requires a local Ollama install and is redundant when the agent is the LLM.
• Add output size warning: Agents are now advised to start with default --pages 3 and only increase if needed, to avoid exceeding context windows.
• Add rate-limiting guidance: Agents are told not to fetch the same app in a loop and to wait before retrying on failure.
• Add cross-platform shell note: Commands assume POSIX (bash/zsh); agents should adapt for PowerShell if needed.

Tests

• 27 new installer and frontmatter validation tests covering: fresh install (all 4 agents), append/force/global behavior, cursor/windsurf --global rejection, bundled script installation, YAML frontmatter validity, compatibility version cross-check against pyproject.toml, and CLI name references.

Full changelog: v0.1.11...v0.1.12