feat(hooks): enforce the mechanical rules with PreToolUse hooks

README.md

@@ -1807,33 +1807,61 @@ The [getting started guide](docs/getting-started.md) covers what happens next.
 quest format and workflow they read before working on a quest outside a skill;
 `quest guide prompting` prints one section.
 
-### Configure Claude Code
+### Configure your agent
 
-`quest setup` configures your own Claude Code settings for quest work, in every
-repository:
+`quest setup` configures your own Claude Code and Codex settings for quest
+work, in every repository:
 
 ```sh
 quest setup
 ```
 
-It sets `env.BASH_MAX_TIMEOUT_MS` to 3300000, so a foreground command such as
-`quest wait` can block for 55 minutes, just under the 1h prompt-cache TTL,
-instead of the 10-minute default.
+It sets Claude Code's `env.BASH_MAX_TIMEOUT_MS` to 3300000, so a foreground
+command such as `quest wait` can block for 55 minutes, just under the 1h
+prompt-cache TTL, instead of the 10-minute default.
 
 It also loads the rules once. The repository's `Quests: ` line has agents run
 `quest context` before any work, which costs a turn and is summarized away
 when the session compacts. Instead, `quest setup` installs a `SessionStart`
-hook (startup, resume, clear, and compact) and a `SubagentStart` hook, each
-running `quest hook`, which adds `quest context` to the session in a
-repository whose root `AGENTS.md` or `CLAUDE.md` has the `Quests: ` line, and
-nothing elsewhere. `quest` must be on the `PATH` Claude Code starts with.
-`quest setup --worktrees` also installs the
+hook (startup, resume, clear, and compact), and for Claude Code a
+`SubagentStart` hook, each running `quest hook`, which adds `quest context` to
+the session in a repository whose root `AGENTS.md` or `CLAUDE.md` has the
+`Quests: ` line, and nothing elsewhere. Codex holds the rules in full, rather
+than spilling them to a file, through the hook's `additionalContextLimit`.
+`quest setup --worktrees` also installs Claude Code's
 [worktree hooks](#claude-code-checkouts).
 
-It edits `$CLAUDE_CONFIG_DIR/settings.json`, or `~/.claude/settings.json`,
-keeps your other settings, refuses a malformed file unchanged, writes once,
-and prints each change. Rerun it after upgrading. `quest setup --remove` takes
-the hooks out, and the timeout while it still holds Quest's value.
+The rules a program can check are enforced instead of read. A `PreToolUse`
+hook on shell commands and file edits denies, with the rule as the reason:
+
+- an edit, commit message, or new GitHub post that adds an em dash outside a
+  `>` quote line;
+- a `git push` whose refspecs lack a `:`;
+- a force-push, `git rebase`, or `git pull --rebase`, unless the command
+  starts with `QUEST_USER_ASKED=1` because the user asked for it;
+- a new GitHub post (`gh pr create`, `comment`, or `review`, `gh issue create`
+  or `comment`, or `gh api` with a `body` field) without a `(Written by ...)`
+  line;
+- a scratch file written under `/tmp`, outside the checkout.
+
+It checks only a body on the command line itself: `--body`, a `body=` field,
+`git commit -m`, or `--body-file -` with a here-document. A body file, or one
+piped into `gh`, passes unread, since the file may hold something else by the
+time the command runs; `quest context` keeps the footer and em-dash rules in
+prose for that case. `quest context` still prints these rules,
+under a last `Checked` section that the hook leaves out, for agents without
+it.
+
+`quest` must be on the `PATH` the agent starts with, as for the worktree
+hooks. Codex asks you to trust the hooks once, and after each upgrade that
+changes them, through `/hooks`; `quest review` runs Codex without your
+config, so its reviewers run none of them.
+
+It edits `$CLAUDE_CONFIG_DIR/settings.json`, or `~/.claude/settings.json`, and
+`$CODEX_HOME/hooks.json`, or `~/.codex/hooks.json`, when that directory exists.
+It keeps your other settings, refuses a malformed file unchanged, writes each
+once, and prints each change. Rerun it after upgrading. `quest setup --remove`
+takes the hooks out, and the timeout while it still holds Quest's value.
 
 ## Where this is going
 

SETUP.md

@@ -62,16 +62,20 @@ to proceed; never delete or rename their files to get past it.
 
 ## Configure the user's agent
 
-`quest setup` configures the user's own Claude Code settings, not the
-repository's, in `$CLAUDE_CONFIG_DIR/settings.json` (or
-`~/.claude/settings.json`). It sets `env.BASH_MAX_TIMEOUT_MS` to 55 minutes, so
-agents can block on a long wait in one call instead of polling. It also adds
-the rules the `Quests: ` line points at to every session and subagent, and
-again after compaction, which would otherwise summarize them away, through
-hooks that need `quest` on the `PATH` Claude Code runs hooks with. Offer it,
-explaining that it applies in every repository the user opens in Claude Code
-(the hooks only in those with the `Quests: ` line), and run it with consent. It
-keeps the file's other settings and prints what it changed.
+`quest setup` configures the user's own agent settings, not the repository's:
+Claude Code's in `$CLAUDE_CONFIG_DIR/settings.json` (or
+`~/.claude/settings.json`), and Codex's in `$CODEX_HOME/hooks.json` (or
+`~/.codex/hooks.json`) when that directory exists. It sets Claude Code's
+`env.BASH_MAX_TIMEOUT_MS` to 55 minutes, so agents can block on a long wait in
+one call instead of polling. It also adds the rules the `Quests: ` line points
+at to every session, and again after compaction, which would otherwise
+summarize them away, and blocks tool calls that break the rules a program can
+check, such as a `git push` without an explicit refspec, through hooks that
+need `quest` on the `PATH` the agent runs hooks with. Offer it, explaining that
+it applies in every repository the user opens (the hooks act only in those with
+the `Quests: ` line), and run it with consent. Codex then asks the user to
+trust the hooks once through `/hooks`. It keeps the files' other settings and
+prints what it changed.
 
 ## Opt into worktree hooks
 

assets/CONTEXT.md

@@ -15,8 +15,6 @@ Engineering rules for every task, as `quest context` prints them. The repository
 - Every test runs in CI. The merge gate holds only fast, high-signal tests; slow ones run nightly.
 - Be adversarial: attack your own approach, reports, reviews, and other PRs as hard as anyone else's, and argue for the best approach over the first one that works.
 - Never add or change a rule in an `AGENTS.md`, `CONTRIBUTING.md`, or skill without being prompted. When you do, follow `quest guide prompting`.
-- No em dashes.
-- End every GitHub post you author with `(Written by <model> with <effort>)`, naming the running model and its reasoning effort, or `(Written by <model>)` when the harness doesn't report the effort; never guess it. Commit messages with a `Co-Authored-By:` trailer are exempt.
 - In a public repository, never mention customers, partners, or prospects, or link to private repositories. Justify work on its merits.
 
 ## Important
@@ -56,8 +54,8 @@ Work in your own worktree, reusing a suitable one you already own.
 Read-only discovery needs no new checkout when your context or own checkout answers it.
 Put PR bodies, review drafts, logs, patches, helper scripts, and temporary clones in its `.tmp/`, so they disappear with the worktree.
 If the repository doesn't ignore `.tmp/`, add it to the file `git rev-parse --git-path info/exclude` names.
-Never write them to `/tmp`, elsewhere in or beside the checkout, or in a directory other agents share, such as a session scratchpad.
-Pass PR bodies to `gh` on stdin (`--body-file -`).
+Never write them elsewhere in or beside the checkout, or in a directory other agents share.
+Pass PR bodies to `gh` on stdin (`--body-file -`) with a here-document, where the hook checks them; one posted from a file goes unchecked, so it must still end with the `(Written by ...)` line and hold no em dashes.
 
 Before ending a task, pausing for a decision, or handing off, commit a checkpoint and report what remains and which checks passed or did not run.
 Finish with a clean checkout: intended changes committed and pushed, incidental generated changes restored, and pre-existing or unrelated changes preserved.
@@ -87,9 +85,7 @@ A decision about an open PR never authorizes closing it. Confirm a close as its
 
 ## Pull requests
 
-Push with an explicit refspec: `git push <remote> refs/heads/<branch>:refs/heads/<branch>`.
-A bare branch name can push to the base instead, as under `push.default=upstream` when the branch tracks it.
-Never rebase or force-push, even a branch only you hold; merge the base instead, only when the branch conflicts with it or a landing step, such as `quest merge` or a merge queue, refuses the head, and resolve the conflicts in that merge; a branch only behind needs none, since landing tests it on the current base.
+Merge the base only when the branch conflicts with it or a landing step, such as `quest merge` or a merge queue, refuses the head, and resolve the conflicts in that merge; a branch only behind needs none, since landing tests it on the current base.
 A PR that another session pushed to in the last hour is theirs: report its events instead of pushing to it.
 Before each push to a PR's branch, confirm the PR is still open, since a push after it merges restores the deleted branch.
 Run the repository's checks locally while the PR is a draft; CI may run only on PRs ready for review.

quest/m0/token-waste/README.md

@@ -41,7 +41,8 @@ Decision prompts, chosen options marked ✅:
   (Theo's tstack, jdx's jouet, Matt Pocock's skills, HumanLayer, Mitchell
   Hashimoto), chosen ✅ and rejected listed:
   - ✅ Mechanical rules become hooks or lint, so their prose leaves the
-    always-loaded rules ([rule checks](/quest/m0/token-waste/rule-checks.md)).
+    always-loaded rules, done in
+    [moq-dev/quest#553](https://github.com/moq-dev/quest/pull/553).
   - ✅ A review round budget by severity, after Theo's `sol-loop`
     ([moq-dev/quest#514](https://github.com/moq-dev/quest/pull/514)).
   - ✅ `quest usage` moves from m1 into this epic, so its closing