| name | regen-ai |
| description | Regenerate the embedded LLM reference and verify the result. Use after changing commands, flags, filter paths, help text, JSON output, or the hand-written reference chapters. |
| allowed-tools | Bash(go generate:*) Bash(go test:*) Bash(go build:*) Bash(git status:*) Bash(git diff:*) Read |
regen-ai
Bring internal/llmdocs/ back in line with the code.
git status --short — note what is already dirty.go generate ./... — rewrites 90-commands.md.go build ./... && go test ./....- Report which commands, flags or filter paths changed.
Then check by hand what generation cannot see:
- JSON output changed?
20-schemas.md needs a matching edit. The generator
only sees the command tree, not the shape of what commands print.
- Cost changed? It appears in five places: the
Cost: line in the command's
Long, the credits field of its call, both READMEs and 30-gotchas.md.
The distributed SKILL.md holds no cost table on purpose — only the three
extremes — so check it just for those.
- New endpoint?
00-guide.md has a "which command answers which question"
table that generation does not touch.
This skill is Claude Code-local; it is not part of the distributed plugin.
