Repository navigation
Consolidate the Write Guard's Flag-Value Argv Walks #3285
Description
Context
Raised by the local review of #3283. That pull request adds _gh_owner_words and _gh_issue_ref_words to host-setup/agent-safety/claude/gh-write-guard.py.
The Duplication
_gh_owner_words, _gh_issue_ref_words, and _gh_positional_repo_words each walk a gh argv the same way. Each skips a text flag's value, stops at --, splits a word on =, and takes the next word when no value is attached. _gh_repo_words reads the repo flag in a similar way. A fix to one walk, such as how a short-option cluster is read, then has to be made in each copy.
Possible Direction
Add one reader that yields each (flag name, value) pair a gh argv holds for a given set of flag names. Move the walks onto it, and check the move with a develop-versus-branch differential over the guard's self-test commands, expecting no verdict change.
🤖 Generated with Claude Code
Activity
Read Owner Flags, Issue Links, and Run-Time Paths in the Write Guard (#…
ptr727 commentedon Oct 23, 2026
Closed as out of the write guard's scope under the maintainer's decision (ptr727, 2026-10-23) on #3391, item 1. The guard bounds a good-faith agent's mistake rather than an adversary, so a parse gap reached only by a command shape no agent plausibly types in a coding session is closed rather than fixed, and the guard's own growth is a cost each later review pays again. The scope statement lands in the guard's docstring and README under #3391. Reopen only with a command a session actually produced.