Codex Crashpad Pending Dumps Taking Disk Space
If Codex Desktop keeps creating .dmp and _sidecar.json files under Crashpad, the disk growth can look like generic System Data. Measure the pending crash-report queue first, avoid sharing dump files publicly, and preserve app state before deleting anything broader.
Find what the AI tool can delete.
Copy the read-only check first. If the output still needs judgment, leave your email and we send the free SafeDisk AI deletion trial step only when the agent cache or session state needs review.
Measure both old and current Codex Crashpad paths.
These checks list counts, sizes, and time ranges. They do not read crash dump contents, and they help separate an old backlog from an active one-report-per-second loop.
paths=(
"$HOME/Library/Application Support/com.openai.codex/web/Crashpad"
"$HOME/Library/Application Support/Codex/Crashpad"
)
for crash in "${paths[@]}"; do
echo "== $crash =="
[ -d "$crash" ] || { echo "missing"; continue; }
du -sh "$crash" "$crash/pending" 2>/dev/null
find "$crash/pending" -maxdepth 1 -type f 2>/dev/null | wc -l
find "$crash/pending" -maxdepth 1 -type f -name '*.dmp' 2>/dev/null | wc -l
find "$crash/pending" -maxdepth 1 -type f -name '*_sidecar.json' 2>/dev/null | wc -l
find "$crash/pending" -maxdepth 1 -type f -print0 2>/dev/null \
| xargs -0 stat -f '%Sm %z %N' -t '%Y-%m-%d %H:%M:%S' 2>/dev/null \
| sort | sed -n '1p;$p'
done
What Is Usually Safe To Remove?
Crashpad pending files are crash reports waiting to be processed or uploaded. They are usually recoverable disk space, but dump files may contain private memory, local paths, prompts, or project data. Do not attach .dmp files to a public issue.
- Quit Codex first.
- Confirm which Crashpad path is growing.
- Move only
pending,new,completed, andattachmentscrash-report files to Trash. - Restart Codex and watch whether the file count grows again.
- If it grows immediately, treat it as an active Crashpad loop, not a one-time cleanup.
Move Crashpad reports to Trash after Codex is stopped.
This does not delete conversations, projects, config, or the whole Codex app support folder.
osascript -e 'quit app "Codex"' 2>/dev/null || true
pkill -f "browser_crashpad_handler.*codex.*/Crashpad" 2>/dev/null || true
for crash in \
"$HOME/Library/Application Support/com.openai.codex/web/Crashpad" \
"$HOME/Library/Application Support/Codex/Crashpad"; do
[ -d "$crash" ] || continue
mkdir -p "$HOME/.Trash/SafeDisk-Codex-Crashpad"
for bucket in pending new completed attachments; do
[ -d "$crash/$bucket" ] || continue
find "$crash/$bucket" -mindepth 1 -maxdepth 1 -exec mv {} "$HOME/.Trash/SafeDisk-Codex-Crashpad/" \;
done
done
Do Not Delete First
- The whole
~/Library/Application Support/Codexfolder, unless you intend to reset local app state. - Conversation/session files before you have backed up anything you need.
- Crash dumps uploaded publicly. Share only metadata unless maintainers provide a private upload path.
- Immutable or permission-blocking workarounds unless you accept that crash reporting may stop working.
About Read-only Workarounds
Making Crashpad/pending read-only or immutable can stop disk growth during an active dump loop, but it also prevents Codex from writing crash reports. Treat it as an emergency workaround only, not a normal cleanup step. The safer default is Trash-only cleanup, then watch whether the pending count starts growing again after restart.
If you do use a read-only workaround, keep the scope narrow: only the specific Crashpad/pending folder that is actively refilling. Do not make the whole ~/Library/Application Support/Codex folder, ~/Library/Application Support/com.openai.codex, or temporary code-sign clone directories permanently read-only unless you are prepared to debug app update, crash reporting, and relaunch failures later.
# Undo a temporary read-only workaround after the loop is fixed.
for pending in \
"$HOME/Library/Application Support/com.openai.codex/web/Crashpad/pending" \
"$HOME/Library/Application Support/Codex/Crashpad/pending"; do
[ -d "$pending" ] || continue
chmod u+w "$pending" 2>/dev/null || true
done
Need a second read before cleaning?
Copy the read-only check first. Request the free SafeDisk AI deletion trial only if the output, app state, or cleanup boundary is still not obvious.
Need a delete / confirm / protect answer for AI tool files?
Submit the form first; the command output or one-line agent storage symptom can follow. We check whether free guidance is enough before asking for the free SafeDisk AI deletion trial.