Add wait skill and AI writing guard

[Miyamura80]

Summary

• add an agent-facing wait skill that issues minute-by-minute bash sleep 60 commands so CLI timeouts stay under 2 minutes
• introduce an ai writing checker (em dash detection) plus integrate it into Prek pre-commit and the linter workflow
• document the never-force-push policy in CLAUDE.md

Testing

• uv run python scripts/check_ai_writing.py

[greptile-apps]

Greptile Summary

Introduces two developer-facing quality improvements: a wait skill that works around CLI timeouts by executing sequential 60-second sleeps, and an em dash detection check integrated into pre-commit hooks and CI to catch AI-generated writing.

• Wait skill (.claude/skills/wait/SKILL.md): Documents an agent-facing skill that executes multiple bash sleep 60 commands sequentially rather than a single long-running sleep to avoid the 2-minute shell timeout
• AI writing guard (scripts/check_ai_writing.py): Scans repository text files for em dash characters (—) as a heuristic to detect AI-written content
• CI integration (.github/workflows/linter_require_ruff.yaml, .pre-commit-config.yaml): Adds the AI writing check to both pre-commit hooks and the lint workflow
• Git policy documentation (CLAUDE.md): Explicitly documents the never-force-push policy

Confidence Score: 4/5

• Safe to merge with one known minor issue in directory filtering logic
• Implementation is straightforward and well-integrated into existing tooling. The directory filtering false negative issue in scripts/check_ai_writing.py was already flagged in previous review threads and is a minor concern - it may skip checking user directories with names like cache/ or node_modules/ that match excluded patterns, but this is unlikely in practice
• scripts/check_ai_writing.py has a directory filtering issue already noted in previous threads

Important Files Changed

Filename	Overview
.claude/skills/wait/SKILL.md	Adds new wait skill documentation for executing minute-by-minute sleeps to avoid CLI timeout issues
scripts/check_ai_writing.py	Implements em dash detection script to catch AI-written text - directory filtering has known false negative issue (already flagged)
.github/workflows/linter_require_ruff.yaml	Adds AI writing check step to lint workflow to detect em dashes in CI
.pre-commit-config.yaml	Integrates AI writing check into pre-commit hooks for local validation
CLAUDE.md	Documents never-force-push policy in git workflow section

_{Last reviewed commit: c23f905}

[greptile-apps]

_{5 files reviewed, 2 comments}

_{Edit Code Review Agent Settings | Greptile}

[greptile-apps]

Additional Comments (1)

.github/workflows/linter_require_ruff.yaml
This workflow installs uv via curl ... | sh and then hard-codes $HOME/.cargo/bin onto PATH. Depending on how uv is installed (and future changes to the installer), uv may not land in that directory, leading to intermittent "uv: command not found" CI failures. Consider using the installer’s documented install dir output (or ~/.local/bin) / verifying uv --version after install so PATH is correct.

Prompt To Fix With AI

This is a comment left during a code review.
Path: .github/workflows/linter_require_ruff.yaml
Line: 24:26

Comment:
This workflow installs uv via `curl ... | sh` and then hard-codes `$HOME/.cargo/bin` onto PATH. Depending on how `uv` is installed (and future changes to the installer), `uv` may not land in that directory, leading to intermittent "uv: command not found" CI failures. Consider using the installer’s documented install dir output (or `~/.local/bin`) / verifying `uv --version` after install so PATH is correct.

How can I resolve this? If you propose a fix, please make it concise.