Close the blockers from the fifth audit
Build App (Preview) / compute-version (pull_request) Successful in 3s
Build Container / build-container (pull_request) Successful in 10m5s
Build App (Preview) / create-release (pull_request) Successful in 1s
Build App (Preview) / build-macos (pull_request) Successful in 4m31s
Build App (Preview) / build-linux (pull_request) Successful in 5m21s
Build App (Preview) / build-windows (pull_request) Successful in 19m1s
Build App (Preview) / prune-previews (pull_request) Successful in 1s
Build App (Preview) / compute-version (pull_request) Successful in 3s
Build Container / build-container (pull_request) Successful in 10m5s
Build App (Preview) / create-release (pull_request) Successful in 1s
Build App (Preview) / build-macos (pull_request) Successful in 4m31s
Build App (Preview) / build-linux (pull_request) Successful in 5m21s
Build App (Preview) / build-windows (pull_request) Successful in 19m1s
Build App (Preview) / prune-previews (pull_request) Successful in 1s
Docs and disclosure. HOW-TO-USE.md's settings table still described the pre-fix behaviour — and help_commands.rs fetches that file from GitHub main at runtime, ahead of the embedded copy, so it would have reached every user's Help dialog the moment this merged. The Config tab named three settings that need a base-image update; there are four, and the omitted one (Session recap) is the one that fails *without* the "won't switch off" symptom the warning teaches. Both now also state the cost nobody had written down: changing any of these recreates the container, which commits a layer. Two stale comments that told a reviewer the code was safe when it was not. compute_claude_code_settings_fingerprint still claimed the historical fingerprint is preserved so an upgrade cannot churn every container — carried over from before the widening, false since the format string changed. And capabilities/default.json, which is the reviewed threat model of record, described a "Save to host…" action this branch deletes. Security and correctness. update_settings validated env vars and nothing else, so the *global* default_ssh_key_path — the fallback for every project without an override — took `/` and read-only bind-mounted the host, which entrypoint.sh then copies into the home volume. classify_ mount_source ran canonicalize on the raw string, which resolves a relative path against Triple-C's own cwd, so `.` and `..` were accepted or refused depending on where the app was launched; the daemon then refuses the mount and the project can never start. Its test passed only because its examples did not exist under app/src-tauri. bind_mount_exclusions still derived a path from every row while project_path_mounts had learned to skip unmountable ones, so a legacy row made /workspace/<name> ordinary container content that a migration would then exclude from staging and destroy. The skip is also logged now rather than silently dropping a folder. The terminal's file-in path checked is_dir() but not file type, so a dropped FIFO blocked forever with no timeout — and it is the only route in now. The web terminal labelled sessions from a global set at request time, so two quick opens swapped them; harmless until Shift+Enter became type-dependent, at which point a mislabelled Claude session submitted a half-written prompt. Opened now carries the type. Every ~/.claude.json write goes through one atomic helper. The awsAuthRefresh branches still truncated in place — the same corruption the Shift+Enter block was fixed for twenty lines later, and its own comment said so. Demonstrated: a failed write now leaves the original byte-identical. And the registration test I added yesterday could pass while the property was false: an audit got five real unregistered commands past its exact-string attribute match, and "exactly once" was in its name but not its body. Mutation-checked against all six shapes. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01GBq2rGum6GX7xXgsas1fDc
This commit is contained in:
+33
-33
@@ -454,27 +454,48 @@ fi
|
||||
# previous Bedrock-profile session — ~/.claude.json lives in the persisted home
|
||||
# volume, so without this the container keeps trying to run the SSO refresh even
|
||||
# after switching to a non-SSO backend (Anthropic/Ollama) or to static creds.
|
||||
# Replace ~/.claude.json atomically: write a sibling temp file, then rename.
|
||||
#
|
||||
# `> "$CLAUDE_JSON"` truncates before it writes, so a write that fails part-way
|
||||
# — a full home volume being the obvious way, and bounding that volume is what
|
||||
# half this release is about — leaves the file unparseable. It holds the OAuth
|
||||
# account, and the damage does not self-heal: the next start's `jq` fails on the
|
||||
# corrupt file, `MERGED` comes back empty, and the `[ -n "$MERGED" ]` guard
|
||||
# skips the write that would have repaired it. `triple-c-task-runner` has done
|
||||
# it this way all along.
|
||||
write_claude_json() {
|
||||
_wcj_tmp="${CLAUDE_JSON}.triple-c-tmp"
|
||||
if printf '%s\n' "$1" > "$_wcj_tmp" 2>/dev/null; then
|
||||
mv -f "$_wcj_tmp" "$CLAUDE_JSON" 2>/dev/null || rm -f "$_wcj_tmp"
|
||||
else
|
||||
rm -f "$_wcj_tmp"
|
||||
echo "entrypoint: warning — could not write $CLAUDE_JSON (leaving it as it was)"
|
||||
return 1
|
||||
fi
|
||||
# By name, after the rename, so these land on the new inode.
|
||||
chown claude:claude "$CLAUDE_JSON"
|
||||
chmod 600 "$CLAUDE_JSON"
|
||||
}
|
||||
|
||||
CLAUDE_JSON="/home/claude/.claude.json"
|
||||
if [ -n "$AWS_SSO_AUTH_REFRESH_CMD" ]; then
|
||||
if [ -f "$CLAUDE_JSON" ]; then
|
||||
MERGED=$(jq --arg cmd "$AWS_SSO_AUTH_REFRESH_CMD" '.awsAuthRefresh = $cmd' "$CLAUDE_JSON" 2>/dev/null)
|
||||
if [ -n "$MERGED" ]; then
|
||||
printf '%s\n' "$MERGED" > "$CLAUDE_JSON"
|
||||
write_claude_json "$MERGED"
|
||||
fi
|
||||
else
|
||||
printf '{"awsAuthRefresh":"%s"}\n' "$AWS_SSO_AUTH_REFRESH_CMD" > "$CLAUDE_JSON"
|
||||
# No existing file, so there is nothing to destroy — but go through the
|
||||
# same helper so the owner and mode are set in one place.
|
||||
write_claude_json "$(printf '{"awsAuthRefresh":"%s"}' "$AWS_SSO_AUTH_REFRESH_CMD")"
|
||||
fi
|
||||
chown claude:claude "$CLAUDE_JSON"
|
||||
chmod 600 "$CLAUDE_JSON"
|
||||
unset AWS_SSO_AUTH_REFRESH_CMD
|
||||
elif [ -f "$CLAUDE_JSON" ] && grep -q '"awsAuthRefresh"' "$CLAUDE_JSON" 2>/dev/null; then
|
||||
# Only rewrite when the key is actually present, to avoid a needless jq
|
||||
# reformat of ~/.claude.json on every start of a non-SSO backend.
|
||||
MERGED=$(jq 'del(.awsAuthRefresh)' "$CLAUDE_JSON" 2>/dev/null)
|
||||
if [ -n "$MERGED" ]; then
|
||||
printf '%s\n' "$MERGED" > "$CLAUDE_JSON"
|
||||
chown claude:claude "$CLAUDE_JSON"
|
||||
chmod 600 "$CLAUDE_JSON"
|
||||
write_claude_json "$MERGED"
|
||||
fi
|
||||
fi
|
||||
|
||||
@@ -491,38 +512,17 @@ if [ -f "$CLAUDE_JSON" ]; then
|
||||
# Only rewrite when the value isn't already true, to avoid a needless jq
|
||||
# reformat of ~/.claude.json on every single start.
|
||||
if ! grep -q '"shiftEnterKeyBindingInstalled"[[:space:]]*:[[:space:]]*true' "$CLAUDE_JSON" 2>/dev/null; then
|
||||
# Write to a temp file and rename, never `> "$CLAUDE_JSON"`.
|
||||
#
|
||||
# `>` truncates before it writes, so a write that fails part-way — a
|
||||
# full home volume is the obvious way, and bounding that volume is
|
||||
# what half this release is about — leaves the file unparseable. This
|
||||
# file holds the OAuth account, and the damage does not self-heal: the
|
||||
# next start's `jq` fails on the corrupt file, `MERGED` is empty, and
|
||||
# the guard below skips the write that would have repaired it. So the
|
||||
# failure mode is a permanently lost login, for a purely cosmetic flag
|
||||
# that suppresses a "run /terminal-setup" tip.
|
||||
#
|
||||
# `triple-c-task-runner` already does it this way; this block was
|
||||
# modelled on the awsAuthRefresh one above, which has the same flaw but
|
||||
# only fires when a Bedrock SSO command is configured.
|
||||
# Atomic, via `write_claude_json` — see its comment for why a plain
|
||||
# `>` on this file can permanently destroy the OAuth login.
|
||||
MERGED=$(jq '.shiftEnterKeyBindingInstalled = true' "$CLAUDE_JSON" 2>/dev/null)
|
||||
if [ -n "$MERGED" ]; then
|
||||
CLAUDE_JSON_TMP="${CLAUDE_JSON}.triple-c-tmp"
|
||||
if printf '%s\n' "$MERGED" > "$CLAUDE_JSON_TMP" 2>/dev/null; then
|
||||
mv -f "$CLAUDE_JSON_TMP" "$CLAUDE_JSON" 2>/dev/null || rm -f "$CLAUDE_JSON_TMP"
|
||||
else
|
||||
# Out of space, or the volume went read-only. The original is
|
||||
# untouched, which is the whole point.
|
||||
rm -f "$CLAUDE_JSON_TMP"
|
||||
echo "entrypoint: warning — could not set shiftEnterKeyBindingInstalled (leaving ~/.claude.json as it was)"
|
||||
fi
|
||||
write_claude_json "$MERGED"
|
||||
fi
|
||||
fi
|
||||
else
|
||||
printf '{"shiftEnterKeyBindingInstalled":true}\n' > "$CLAUDE_JSON"
|
||||
# Nothing to destroy, but the helper owns the owner/mode too.
|
||||
write_claude_json '{"shiftEnterKeyBindingInstalled":true}'
|
||||
fi
|
||||
chown claude:claude "$CLAUDE_JSON"
|
||||
chmod 600 "$CLAUDE_JSON"
|
||||
|
||||
# ── Docker socket permissions ────────────────────────────────────────────────
|
||||
if [ -S /var/run/docker.sock ]; then
|
||||
|
||||
Reference in New Issue
Block a user