| name | disk-space-assessment |
| description | Measure and explain disk usage without deleting data, producing reusable compressed ncdu snapshots and an evidence-backed cleanup handoff. Use when disk is low, a filesystem is unexpectedly large, builds fail from space pressure, the user asks what is consuming space, or deeper iterative disk analysis is needed before cleanup. |
Disk Space Assessment
Build a reproducible picture of disk usage. Stop at findings and proposed actions; use disk-space-cleanup when the user authorizes remediation.
Read references/ignore-paths.md before scanning. Read references/direnv-gc-roots.md for Nix development-root attribution. Read references/observed-heavy-hitters.md when triaging this machine or interpreting familiar paths.
Required Artifact Contract
Always create a reusable, timestamped ncdu export for every filesystem or major root assessed. Never make an interactive ncdu session, transient /tmp export, or du output the only record of an assessment.
Always scan / with privilege. An unprivileged root scan is incomplete by design because it cannot measure private service state. Also scan each separately mounted pressured filesystem because safe_ncdu -x does not cross mount boundaries:
/run/wrappers/bin/sudo -n env HOME=/home/imalison safe_ncdu /
safe_ncdu /home
safe_ncdu /nix/store
safe_ncdu writes these durable artifacts under /srv/disk-space-assessments/ (a shared, users-group-readable location — $SAFE_NCDU_DIR, overridable via env var — not a per-user ~/.cache dir):
safe-ncdu-<root>-<timestamp>.json.zst: compressed importable ncdu data<snapshot>.excludes: exact exclusions used<snapshot>.meta: scan root, timestamps, hostname, ncdu version, status, and artifact pathslatest-<root>.json.zst plus matching sidecar symlinks
The latest-<root>.json.zst symlink (e.g. latest-root.json.zst, latest-home.json.zst, latest-nix_store.json.zst) is the intentional predictable-location contract: it always points at the newest snapshot for that root, so any session can find "the most recent ncdu run" without searching timestamped filenames or being told the path. Prefer querying latest-<root>.json.zst over a specific timestamped snapshot unless you deliberately need an older run for comparison. Every skill or ad-hoc command that produces an ncdu export (including from disk-space-cleanup) must go through safe_ncdu so it lands under this same location instead of a one-off /tmp export or an unrecorded interactive session.
Keep timestamped files intact for iterative analysis. Report their absolute paths in the handoff. If a scan cannot finish, retain and report the failed manifest, mark the coverage gap, and use bounded probes as supplemental evidence.
Workflow
- Record filesystem pressure and topology.
- Run a privileged
/ scan and choose additional roots for separately mounted filesystems.
- Create reusable
safe_ncdu snapshots before any cleanup.
- Analyze snapshots repeatedly with
top and open; do not rescan for every question.
- Attribute special stores such as Nix separately.
- Produce findings, confidence, cleanup candidates, and artifact paths.
1. Establish the Baseline
df -h /
df -h /home
df -h /nix
findmnt -rn -o TARGET,FSTYPE,SOURCE,OPTIONS
Record used/free space and whether /home, /nix, or other large paths are separate mounts. Note active builds or services that may cause measurements to move during the assessment.
2. Select Scan Roots
Require one-filesystem coverage:
- Always scan
/ with /run/wrappers/bin/sudo -n env HOME=/home/imalison; never substitute an unprivileged root scan. On NixOS, use the setuid wrapper explicitly because a non-setuid sudo store binary may appear earlier on PATH.
- Scan separately mounted
/home and /nix/store independently.
- Add a focused root such as
~/Projects when the first snapshot identifies it as dominant.
- Treat a failed privileged root scan as an explicit coverage gap; do not silently fall back to an unprivileged scan.
Inspect exclusions before a long scan:
safe_ncdu excludes /
Update references/ignore-paths.md and the implementation of safe_ncdu together when a newly discovered remote, recursive, or special mount needs a permanent exclusion.
3. Create the Snapshots
Run scans early enough that cleanup does not destroy the evidence:
/run/wrappers/bin/sudo -n env HOME=/home/imalison safe_ncdu /
safe_ncdu /home
safe_ncdu /nix/store
If safe_ncdu is unavailable, run /srv/dotfiles/dotfiles/lib/functions/safe_ncdu <subcommand> <args> as a script (e.g. zsh /srv/dotfiles/dotfiles/lib/functions/safe_ncdu top /srv/disk-space-assessments/latest-root.json.zst 30) — do NOT plain-source it. The file ends with a bare safe_ncdu "$@" tail-call meant for zsh's autoload mechanism; sourcing it directly in a non-interactive shell re-invokes that tail-call with the sourcing command's own (usually empty) positional params, silently kicking off a full unprivileged / scan and clobbering the latest-root symlink with inferior data. If that happens, re-point latest-* symlinks at the correct privileged snapshot's .json.zst/.excludes/.meta files. If ncdu itself is missing, use Nix temporarily rather than substituting a non-reusable interactive scan.
The interactive shell's safe_ncdu function is not sourced live from this repo file — home-manager bakes it into /nix/store/*-dotfiles-zsh-lib/functions/safe_ncdu at build time. After editing lib/functions/safe_ncdu, the change is invisible to the shell function (and even a brand-new zsh -c '...' process, since it re-autoloads from the same stale Nix store path) until just switch (from /srv/dotfiles/nixos) rebuilds and re-links it. To validate an edit immediately, invoke the repo file directly as a script (zsh /srv/dotfiles/dotfiles/lib/functions/safe_ncdu ...), same as the "unavailable" fallback above.
Artifacts always land in the shared /srv/disk-space-assessments/ location regardless of which user or sudo ran the scan, so a privileged scan cannot orphan its output in root's home. Set HOME=/home/imalison on the privileged invocation anyway — safe_ncdu's exclude computation reads $HOME to skip imalison's keybase mount, and without it a bare sudo would compute excludes against root's home instead. This host's default umask (077) makes every newly created file 0600 regardless of the directory's group ownership, which would silently defeat group read; safe_ncdu and direnv_gc_roots_audit.py both explicitly chgrp users + chmod 0640 each artifact after writing it to compensate. If sudo or a manual run ever leaves root-owned or 0600 snapshot files behind, restore them to imalison:users mode 0640 (directory 2750) so the users group keeps read access.
4. Analyze Iteratively
Query the same export at multiple depths:
safe_ncdu top /srv/disk-space-assessments/latest-root.json.zst 30
safe_ncdu top /srv/disk-space-assessments/latest-root.json.zst 30 /home/imalison
safe_ncdu top /srv/disk-space-assessments/latest-home.json.zst 30 /imalison/Projects
safe_ncdu open /srv/disk-space-assessments/latest-root.json.zst
Use du only for bounded confirmation, live-change checks, or paths missing from the snapshot:
timeout 30s du -xh --max-depth=1 "$HOME/.cache" 2>/dev/null | sort -h
timeout 30s du -xh --max-depth=1 "$HOME/.local/share" 2>/dev/null | sort -h
Distinguish physical allocation from logical/apparent size, especially for sparse files, hardlinked Nix store data, copy-on-write storage, and container layers. Do not sum overlapping Nix closure sizes as reclaim estimates.
5. Attribute Nix Store Usage
When /nix/store is large, first quantify whether garbage collection can help:
nix_store_audit --top 30
nix-store --gc --print-dead
nix-store --gc --print-roots
Use /srv/dotfiles/dotfiles/lib/functions/find_store_path_gc_roots and nix why-depends to explain why a large path is retained. Inspect .direnv/flake-profile-*, result* symlinks, system generations, and current/booted system closures. Prefer nix_store_audit over an initial du -sh /nix/store, which is slow and does not explain retention.
Always create a reusable .direnv GC-root audit artifact when direnv roots exist:
python /srv/dotfiles/dotfiles/agents/skills/disk-space-assessment/scripts/direnv_gc_roots_audit.py --top 30
This separates collectively direnv-only paths from paths retained by non-direnv roots and estimates each project's marginal uniquely retained footprint. Read references/direnv-gc-roots.md before interpreting or acting on the result.
For very large /nix/store exports, use safe_ncdu open for read-only navigation and the Nix audit tools for noninteractive attribution. safe_ncdu top refuses snapshots over 128 MiB compressed by default because its jq parser may exhaust memory. Do not estimate physical store usage by summing raw ncdu JSON dsize fields; use du -sx --block-size=1 /nix/store when a hardlink-aware physical total is necessary.
Assessment Handoff
Return:
- Baseline filesystem usage and scan time.
- Absolute paths to every snapshot,
.excludes, and .meta artifact.
- Coverage and privilege limitations.
- Largest consumers with evidence from snapshot queries or store attribution.
- Ranked cleanup candidates, expected reclaim range, risk, and whether data is rebuildable.
- Volatile paths or active processes that cleanup must preserve.
Do not delete, prune, garbage-collect, truncate, or mutate data in this skill.
Learning Loop
After each assessment:
- Add new permanent scan exclusions to
references/ignore-paths.md and safe_ncdu.
- Add stable machine-specific diagnostic knowledge to
references/observed-heavy-hitters.md.
- Remove stale observations instead of accumulating an unbounded chronological log.
