Skip to content

Doctor's re-init fix keeps --no-hooks; deploy compose runs without tracekit on root's PATH (#464) - #845

Merged
LoopGlitch26 merged 2 commits into
mainfrom
agent/issue-464-reinit
Nov 14, 2026
Merged

LoopGlitch26 merged 2 commits into
mainfrom
agent/issue-464-reinit

Conversation

@LoopGlitch26

Copy link
Copy Markdown
Member

The two items still open on #464 (its 11-13 status comment).

1. Doctor's re-init fix keeps --no-hooks

plugin/README.md tells Claude Code plugin users to run system init with --no-hooks, but doctor's re-init fix (REINIT) has no --no-hooks. System init doesn't detect the plugin, so a plugin user who followed the fix got init's hooks beside the plugin's (doctor then warns, D-HOOKS-PRESENT), and a host that declined hooks (an SDK agent) got Claude Code hooks back.

In system mode, doctor.collect now adds --no-hooks to every fix that names the re-init when client.json records no_hooks (init was told --no-hooks), or when the Claude Code plugin is on for the agent's user and the agent is Claude Code. Another agent's re-init (--agent codex) is left alone, since the Claude Code plugin doesn't give it hooks.

2. deploy compose without tracekit on root's PATH

docs/deploy-compose.md (L9, L80) ran sudo tracekit deploy compose. Root's PATH has no tracekit unless pip installed it into the system Python, so from a checkout or a venv it fails with "command not found". The page now runs sudo "$(command -v python3)" -m tracekit deploy compose … from the checkout, as deploy/compose/e2e.sh does, and says why root and why that form. tests/test_docs.py no longer exempts sudo tracekit deploy.

Checks

  • tests/test_doctor_v2.py::NoHooksChosen::test_the_reinit_fix_keeps_no_hooks_where_init_was_told_it_or_the_plugin_hooks and tests/test_docs.py::DocCommands::test_no_command_needs_tracekit_on_roots_path both fail on main's doctor.py and deploy-compose.md and pass here.
  • By hand: python -m tracekit deploy compose --dir <tmp> from the checkout (without sudo) writes the stack and prints the note about running as root.
  • Not checked by hand: doctor in a real system-mode install (it needs root and /etc/tracekit); the test drives collect with system mode's client.json mocked.
  • Gates in the worktree on macOS: make test-ts lint build test pass (5219 passed, 395 skipped). Linux CI is the merge gate.

Fixes #464.

…acekit on root's PATH (#464)

A plugin user who ran system init with --no-hooks (plugin/README.md) and
followed doctor's re-init fix got init's hooks beside the plugin's, and a
host that declined hooks got them back. In system mode the fix now adds
--no-hooks when client.json records it or the Claude Code plugin is on for
the agent's user.

docs/deploy-compose.md ran `sudo tracekit deploy compose`, which finds no
tracekit on root's PATH unless pip installed it system-wide; it now runs the
checkout with your Python by its path, as deploy/compose/e2e.sh does, and
the docs test no longer exempts deploy.

Signed-off-by: LoopGlitch26 <53336715+LoopGlitch26@users.noreply.github.com>
)

With init --managed and the Claude Code plugin on, every re-init fix
became `init ... --no-hooks --managed`, and running it removed the
root-owned managed hooks, leaving only the plugin's, which the agent's
user can turn off. The plugin check reads the agent's own settings, so
the agent could make doctor recommend that. The fix no longer adds
--no-hooks when init's hooks are the managed ones.

The re-init fix also named no --agent, so on another coding agent's host
(init --agent codex, with or without --no-hooks) it set the host up
again for Claude Code. collect() now adds init's agent to every re-init
fix; the harness hook check no longer adds its own.

docs/deploy-compose.md: the witness-class command follows a block that
ends in `cd /srv/tracekit`; it now says to run it in the checkout.

Signed-off-by: LoopGlitch26 <53336715+LoopGlitch26@users.noreply.github.com>
@LoopGlitch26
LoopGlitch26 enabled auto-merge (squash) November 14, 2026 06:58
@LoopGlitch26
LoopGlitch26 merged commit 6a3ec29 into main Nov 14, 2026
32 of 34 checks passed
@LoopGlitch26
LoopGlitch26 deleted the agent/issue-464-reinit branch November 14, 2026 09:18
LoopGlitch26 added a commit that referenced this pull request Nov 15, 2026
#845) (#858)

On a host set up with init --managed, only the hook checks' fixes
(D-HOOKS-MANAGED, D-HOOKS-PRESENT, D-HOOKS-VENV) named --managed. The
other re-init fixes, D-PROCESS-BOUNDARY, D-CODE-TRUST, D-UNIT-HARDENING,
D-TAILER-CONFIG among them, ended without it, so a person who followed
one ran init without --managed: init wrote its hooks into the agent's
own ~/.claude/settings.json, which the agent can edit, and took them out
of the root-owned managed file. With the Tracekit plugin on, each call
was then decided twice. Without --managed-only, init also turned off the
allowManagedHooksOnly an init with it had turned on.

collect() now adds --managed to every re-init fix when init's hooks are
in Claude Code's managed settings, and --managed-only when init recorded
it there (its _tracekit_allowManagedHooksOnly marker; an admin's own
allowManagedHooksOnly has no marker, and init leaves it). The hook
checks' fix no longer adds its own, so it names --managed once.
--no-hooks and --agent stay as #845 left them.

Signed-off-by: LoopGlitch26 <53336715+LoopGlitch26@users.noreply.github.com>
LoopGlitch26 added a commit that referenced this pull request Nov 15, 2026
* doctor: a re-init fix on a --managed host keeps --managed (follow-up to #845)

On a host set up with init --managed, only the hook checks' fixes
(D-HOOKS-MANAGED, D-HOOKS-PRESENT, D-HOOKS-VENV) named --managed. The
other re-init fixes, D-PROCESS-BOUNDARY, D-CODE-TRUST, D-UNIT-HARDENING,
D-TAILER-CONFIG among them, ended without it, so a person who followed
one ran init without --managed: init wrote its hooks into the agent's
own ~/.claude/settings.json, which the agent can edit, and took them out
of the root-owned managed file. With the Tracekit plugin on, each call
was then decided twice. Without --managed-only, init also turned off the
allowManagedHooksOnly an init with it had turned on.

collect() now adds --managed to every re-init fix when init's hooks are
in Claude Code's managed settings, and --managed-only when init recorded
it there (its _tracekit_allowManagedHooksOnly marker; an admin's own
allowManagedHooksOnly has no marker, and init leaves it). The hook
checks' fix no longer adds its own, so it names --managed once.
--no-hooks and --agent stay as #845 left them.

Signed-off-by: LoopGlitch26 <53336715+LoopGlitch26@users.noreply.github.com>

* doctor: re-init advice keeps init's own flags (follow-up to #858)

A fix that re-ran system init gave back init's agent, --no-hooks and
--managed only. Following one on a host set up with --policy, --harness,
--approver, --egress and the like rewrote signer.yaml with init's defaults
(its own policy, no harness binding), and doctor didn't say.

- init --v2 records the rest of its flags in client.json (hooks.init_flags,
  and the directory of a --project init); v1 init records its own at the
  top level. Doctor reads them from that root-owned file only and gives
  them back shell-quoted. A host whose client.json predates the record gets
  a fix that says doctor can't tell init's flags, instead of a re-init that
  drops them.
- v1: V1_REINIT never named --managed; doctor now gives v1 init's flags too,
  and --managed where the managed settings hold v1 hooks on an older host.
- D-HOOKS-PRESENT no longer fails "every call is decided twice" on an
  init --managed host with allowManagedHooksOnly on when only the agent's
  own settings enable the plugin (Claude Code doesn't run it then). On a
  managed host its fix no longer names tracekit uninstall, which never
  takes managed hooks out.
- D-SYSTEM-CONFIG keeps --managed when the managed settings hold init's v2
  hooks.
- --managed-only is given again only while init's marker is there and
  allowManagedHooksOnly is still on, init's own rule.

Signed-off-by: LoopGlitch26 <53336715+LoopGlitch26@users.noreply.github.com>

* doctor: re-init fixes run from the clone; init takes --project DIR (review of #860)

- init --project [DIR]: a re-init fix for a --project host named the
  directory and had the person run `sudo python3 -m tracekit init` there,
  in the agent's own repo, where -m imports a planted tracekit/ before the
  release's, as root. init now takes the directory (no DIR keeps the
  working directory), init records it, and the fix runs from the
  root-owned clone like every other re-init.
- --harness is recorded as given (absolute, symlinks kept), not as its
  resolved real path, which a package upgrade removes or leaves pointing
  at the old binary. The signer still binds the resolved program.
- --managed-only is recorded (v1 and v2). When allowManagedHooksOnly is off
  now (an admin, or MDM replacing the managed file), the fix keeps it and
  says to drop it only if it was turned off on purpose; a fix's own
  --managed-only isn't given twice.
- --i-understand-agent-is-privileged, --no-witness and --public-witness
  aren't recorded or replayed.
- migrate --system keeps v1's init_flags.
- install no longer imports doctor's private helpers: install.reinit_flags
  and install.setting are shared. doctor --config with no client.json says
  there is none.

Signed-off-by: LoopGlitch26 <53336715+LoopGlitch26@users.noreply.github.com>

* doctor: keep --harness as given only where root alone can repoint it (re-review of #860)

- harness_spec kept the given path, links and all, but trusted_file checks
  only the real path's chain: a link in a directory the agent's user can
  write could be repointed before a re-init, which then bound any
  root-owned program. The given path is now kept only when every
  directory its lookup reads, its links' targets included, is root-owned
  and not group- or world-writable; else init records the real path.
- The no-record note names --project DIR: a bare --project run from the
  release's clone puts the hooks in the clone.
- uninstall takes --project [DIR] as init does, and dev init's plugin
  check reads DIR's .claude (and doesn't take a named project's .claude
  for a user config root, as it didn't the working directory's).

Signed-off-by: LoopGlitch26 <53336715+LoopGlitch26@users.noreply.github.com>

* cli: --project [DIR] resolves against os.getcwd() as init runs, on every OS

CI's py3.9 Windows job failed five init --dev --v2 --project tests: the
hooks went into the checkout, not the test's working directory. --project
with no DIR went through os.path.abspath(".") at run time, and on Windows
abspath asks the OS for the working directory (nt._getfullpathname)
instead of calling os.getcwd(), which those tests patch to set it and
which the code read before --project took a DIR. DIR is now joined to
os.getcwd() and normalized, as POSIX abspath does. A test stands in an
abspath that ignores a patched os.getcwd, as Windows' does, and checks a
parser built before a chdir. The two CliProject tests compare against
os.path.abspath of the input (C:\srv\my repo on Windows).

Signed-off-by: LoopGlitch26 <53336715+LoopGlitch26@users.noreply.github.com>

* tests: HarnessSpec skips on Windows, as this file's other system-mode tests do

CI's py3.13 windows-latest job failed it: resolve_harness couldn't take the
temp script's "#!/bin/sh" interpreter, as Python 3.13's ntpath.isabs no
longer counts a drive-less rooted path. Harness binding is system mode's,
which _require_system refuses on Windows before any harness code runs.

Signed-off-by: LoopGlitch26 <53336715+LoopGlitch26@users.noreply.github.com>

---------

Signed-off-by: LoopGlitch26 <53336715+LoopGlitch26@users.noreply.github.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

System mode: hints, messages and help text that don't match what init does

1 participant