feat: add recommendation pipeline and fix command prefix for Codex#35
Open
gabelul wants to merge 3 commits intopbakaus:mainfrom
Open
feat: add recommendation pipeline and fix command prefix for Codex#35gabelul wants to merge 3 commits intopbakaus:mainfrom
gabelul wants to merge 3 commits intopbakaus:mainfrom
Conversation
added 2 commits
March 15, 2026 21:08
The build system hardcoded `/` as the command prefix for every provider, but Codex CLI uses `$`. Added command_prefix to PROVIDER_PLACEHOLDERS so replacePlaceholders and prefixSkillReferences use the right one. Now `$normalize` shows up in Codex output instead of `/normalize`.
The problem: you run /audit, it tells you to run /normalize, /harden, /polish — but those recommendations live only in conversation text. Context gets compressed, you forget what to run next, and the whole audit was basically wasted effort. Now /audit and /critique write their recommendations to .impeccable-plan.md — a structured plan file that sticks around. Two new commands consume it: - /next: runs the top pending item, marks it done, shows what's left - /fix-all: runs everything in priority order, start to finish The plan file is just markdown, so it works across all providers.
f577711 to
e4bfbe1
Compare
Without this, the AI just sees "execute the command" and might try to wing it instead of invoking the actual skill. Now the instructions explicitly say to use skill/slash command invocation, and fall back to reading the SKILL.md file directly if needed.
48addb4 to
237a8b4
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What's this about
Two things that were bugging me:
1. Codex was showing
/instead of$for commands. The build system hardcoded/as the command prefix for every provider, but Codex uses$. So all the "suggested commands" in Codex skill output said/normalizewhen they should've said$normalize. Addedcommand_prefixto the provider config and threaded it throughreplacePlaceholdersandprefixSkillReferences.2. Audit recommendations get lost between commands. You run
/audit, it tells you "run/normalizefor theming,/hardenfor a11y,/polishat the end" — but those suggestions only exist as text in the conversation. Context gets compressed, you move on, and the whole audit was basically wasted effort.Now
/auditand/critiquewrite their recommendations to.impeccable-plan.md— a structured plan file that sticks around. Two new commands consume it:/next— runs the top pending item, marks it done, shows what's left/fix-all— runs everything in priority order, start to finishThe plan file is just markdown, so it works the same across all 8 providers. No runtime dependencies, no special APIs — just an AI reading a file and following instructions.
What changed
Bug fix:
scripts/lib/utils.js—command_prefixinPROVIDER_PLACEHOLDERS, used inreplacePlaceholdersandprefixSkillReferencesscripts/lib/transformers/codex.js— passes$toprefixSkillReferencesNew commands:
source/skills/next/SKILL.mdsource/skills/fix-all/SKILL.mdUpdated skills:
source/skills/audit/SKILL.md— emits.impeccable-plan.mdafter the reportsource/skills/critique/SKILL.md— sameDocs/metadata:
plugin.json,marketplace.json,README.md,AGENTS.md,NOTICE.md,index.html— 17 → 19 commandsTested
$everywhere —$audit,$normalize,$next, etc.//auditon a test page →.impeccable-plan.mdcreated with 9 prioritized items/next→ picked up first item (/harden), executed it, moved to Completed, renumbered the rest/fix-alluses the same read→execute→update loop, verified through shared mechanism._file issues, nothing new)