Corrects the file extension in the verification report creation guide.
It changes the expected file type from .html to .md,
aligning it with the actual implementation and improving user guidance.

…kills

Based on analysis of Claude Code behavior, subagents only receive skill
headers when the Skill tool is explicitly listed in their tools. Without
it, subagents cannot access any skills - neither Agent OS standards
(when installed as skills) nor user's own custom skills.

By adding Skill unconditionally to relevant subagents, they can now:
- Receive skill headers describing available skills
- Decide whether to load and use specific skills based on context
- Access user's custom skills regardless of standards_as_claude_code_skills setting

Affected subagents:
- implementer: primary code implementation agent
- spec-writer: creates specifications with architectural decisions
- spec-verifier: validates specs against standards
- spec-shaper: gathers requirements with awareness of tech stack
- tasks-list-creator: plans tasks aligned with coding conventions

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

…ls-and-claude-md

fix: add CLAUDE.md and skills reading instructions for subagents

…ntation_prompt

[bug fix] Updates verification report file extension

…e-spec-command

Changed the agent name to spec-initializer in the shape spec command

v3 branding

)

When a bash arithmetic expression ((var++)) is evaluated and var is 0,
the post-increment returns the old value (0), which bash treats as a
failure exit status. Combined with set -e, this silently kills the
script. This caused project-install.sh to exit immediately after
printing "Configuration:" without performing any installation.

Fix by appending `|| true` to all standalone ((var++)) expressions
so the zero-return doesn't trigger set -e.

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

…ds#327)