The agy knowledge allowlist walked the store with `**` and copied with
plain cp, so a symlink inside the store was both followed and dereferenced.
The extension rule still bounded what kind of file was collected, but not
whose: a link to a home directory hands over settings.json, CLAUDE.md and
every cached .json in it, and those reached a commit. A link to / made the
walk itself unbounded, on the path that runs before every agent launch.
The tree is now walked a level at a time and nothing that is a symlink is
followed or copied.
Autopush had the same shape one layer out. Neither GIT_TERMINAL_PROMPT nor
GIT_ASKPASS closes a socket, and git has no connect timeout to set: against
a blackholed address a push took 135s with http.lowSpeedLimit and
http.lowSpeedTime set as well as without them. ssh can time itself out and
is now told to; the autopush pull and push are additionally capped with
timeout(1). An explicit --push stays uncapped, since it is watched and has
to report what a real transfer really did.
Also: scaffold /.migrate-stash into .gitignore beside /.adopt-stash, which
the comment already claimed was covered; and drop the live memory path
during a slug migration only when it is a link. Reached from the
path-derived fallback candidate it can be a real populated directory, where
rm -f correctly refuses -- but said so in rm's voice, so a --silent run that
had succeeded printed what read as an error.
The guard around the symlink step only linked when the live Claude
project directory already existed, which is exactly backwards for the
clone-onto-a-new-machine restore case: a freshly cloned vault entry
would be silently left unlinked and a starting agent would write fresh,
history-less memory instead. _agents_repo_ensure_symlink already makes
its own parent directories and is idempotent, so nothing depended on
the guard; it is removed and the link is now attempted unconditionally.
Also stop swallowing a refused or failed link as success: the helper's
exit status is now checked, and agents-vault reports its own error and
exits 1 instead of silently continuing with no link in place.
Smaller fixes from the same review pass:
- check the exit status of _agents_repo_install_tools and the
core.hooksPath git config write, instead of discarding both
- give the vmem mkdir failure a stderr message like every other fatal
in the function
- guard hostname with type -q and add it to DEPENDENCIES
- .version creation now sets changed, so --link (which skips the
commit step) reports it in --quiet mode
- reword --link's help/doc text: it still scaffolds the vault and
links memory, it only skips the final commit
- drop the unused c_dim color variable
- move the __fish_agent_vault_dir / __fish_agent_vault_autopush
documentation below Opinionated Components so its NOTE: callout
(now flush-left so it actually renders as a Starlight Aside, per
review) doesn't become the first Note aside in the page and steal
the existing test's assertions about the original 4-bullet one
Adds two tests: pre-seeded vault entry with no live directory at all
(the restore path the guard was breaking), and a forced link failure
asserting agents-vault now exits 1 instead of 0.
Creates the vault repo on demand, reusing the AGENTS version bumper and
hook shims, then links the current project's live memory directory into
its slug-keyed entry and commits.
Because the live directory becomes a symlink into the vault, backup and
restore are the same operation: a cloned vault relinks itself on the next
run in each project, with no manifest and no batch restore step.
Also fixes _agents_repo_install_tools' progress messages, which hardcoded
the literal "AGENTS/.agents-tools/" even for callers writing elsewhere:
they now name repo_dir's own basename, so agents-vault reports its own
directory instead of a false AGENTS/ path.
Verified every claim in the Prompt and Theme section against the actual
source and corrected several inaccuracies: the Starship wrapper's missing
C3 gate, the fallback prompt's vi-mode states and segment order, the FZF
theme's real location (conf.d/theme.fish, not integrations/fzf.fish) and
color set, and the Catppuccin theme-switch example using the wrong
fish_config subcommand (choose, not save). The right-prompt Docker-context
example was rewritten to show that it's independent of exit status.
While reformatting, found that any block build-manual.py couldn't classify
as shell/table/tree fell back to plain 4-space markdown indentation, which
silently renders as squashed, unreadable paragraph text on any page that
also contains an <Aside> or <FileTree> — MDX has no indented-code-block
syntax, unlike plain Markdown. This affected 07-customization.mdx plus four
other pages. Fixed the fallback to emit a fenced ```text block instead,
since fences work in both MDX and plain Markdown; this also gives every
affected block Starlight's normal code-block styling instead of a bare grey
slab.
docs/fish-config.md is intentionally left stale here — CI regenerates and
auto-commits it from docs/manual/** on push to main.
Sub-categories were previously undocumented outside README and the
per-category reference pages. Add a paragraph to 07-customization.md (the
canonical opinionated-components page) explaining the sub-category cascade,
a short note to the 08-components-reference index pointing readers at each
category's sub-category list, and a troubleshooting note that
__fish_config_op_<category>_<subcategory> variables follow the same
check/reset pattern as category variables.
Session logging is the one opinionated category that writes a persistent
record of terminal output to disk, and those logs can contain secrets.
Silent-by-default is the wrong posture for it, so C5 now defaults to off:
__fish_config_op_logging must be set to an explicit truthy value to enable
capture, and the master switch cannot enable it on its own.
Implemented as a single special case in __fish_config_op_enabled, so all
five capture components, the sentinel file, the paru/yay wrappers, the
Kitty watcher, and the config-settings TUI pick it up with no other code
changes. The existing startup sync in conf.d/logging-events.fish
reconciles the sentinel and removes the generated wrappers on the first
shell after upgrading.
BREAKING CHANGE: users relying on the previous default must now run
`set -U __fish_config_op_logging on`. Existing logs in ~/.terminal_history
are left untouched.
Wrapped the XDG variables and PATH setup in config.fish with the C3 overrides toggle so users can fully opt out. Updated the documentation (07-customization.md, 02-path-setup.md, README.md) to reflect this. Also added a Home link to the 404 page and removed a leftover test_sidebar.js file.
Fixed tables and text formatting in the Dependency Catalog and Configuration Variables pages so they render properly as markdown instead of flat code blocks. Added introductory sentences to Troubleshooting and Customization so the Starlight sub-heading navigation cards get injected at the top of the page rather than jumping below the first heading.
Every ruled table in the SSOT used one solid rule under the header
except the "Fish Universal Variables" table, which used the other
(per-column, RST-style) convention build-manual.py just learned to
also accept. Two conventions for the same thing with no reason for
the split, so convert the outlier to match the other six.
The parser keeps supporting both — the per-column style stays a
useful fallback for anyone still authoring that way — this just
makes our own SSOT consistent.
_as_table() only detects tables whose data rows are indented deeper
than a ":"-terminated label. The "Component Reference" tables use a
different convention (header, dashed rule, rows all at the same
indent), so they never matched and fell through to a plain indented
code block on the Starlight site.
Add _as_ruled_table() to recognize that header+rule+rows shape. It
supports N columns, folds word-wrapped continuation lines into the
previous row, and backtick-escapes cells containing "<" or "{" instead
of rejecting the table outright (unlike _as_table's stricter guard,
which those tables would otherwise trip on for angle-bracket
placeholders and brace globs).
Also fixes four ambiguous rows in 07-customization.md where two
columns had collapsed to a single space, making them indistinguishable
from a word-wrapped continuation.
Adds docs/build-manual.py (--concat) and a round-trip test in
verify-manual.py that reproduces the pre-migration fish-config.md
exactly from docs/manual/.
Fixes found by the round-trip check, applied in split-manual.py and
re-run to regenerate docs/manual/:
- NAME/SYNOPSIS/TABLE OF CONTENTS had no sidebar.order, so they sorted
after every numbered section instead of interleaving with DESCRIPTION
in original document order. All manual pages now get sidebar.order
from a single position counter matching source order.
- split_h1/split_h2 used .strip() on section bodies, which also ate
leading indentation on the first body line, corrupting the 4-space
indented code blocks that open SYNOPSIS and TABLE OF CONTENTS.
- The source's leading pandoc metadata block (title/section/header/
date/author) was discarded entirely by the splitter. It's now parsed
off via manualtools.parse and stashed on index.md under a "pandoc"
key; build-manual.py re-emits it verbatim ahead of the first heading.
rootiest