Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/avdlee/xcode-disk-cleanup-agent-skill/xcode-disk-cleanupnpx skills add AvdLee/Xcode-Disk-Cleanup-Agent-Skill --skill xcode-disk-cleanupgit clone --depth 1 https://github.com/AvdLee/Xcode-Disk-Cleanup-Agent-SkillWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00147 | $0.02774 |
| Opus 5 | $0.00073 | $0.01387 |
| Sonnet 5 | $0.00029 | $0.00555 |
| Haiku 4.5 | $0.00015 | $0.00277 |
Grade B, and why
xcode-disk-cleanup scanned grade B with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Recursive force deletemediumDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
- Never use wildcards, unbounded `rm -rf`, `sudo`, SIP changes, or manual deletion Downgraded: this mod is about security review, or the phrase is quoted, so it is likely naming the pattern rather than instructing it.
How it starts
The opening of the file, as written. The whole thing — 251 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Xcode Disk Cleanup
Safety contract
- Treat every cleanup request as permission to audit, not permission to delete.
- Never mutate storage before presenting exact candidate IDs, paths or simulator identifiers, measured sizes, risks, and regeneration costs.
- Treat a category-level request made after an audit (for example, “remove all outdated documentation caches”) as approval of the matching, current frozen selection set when it was already fully enumerated to the user. Execute in that turn after revalidation; do not restate the set and ask again. No special command or repeated confirmation is required. Broad requests made before an audit, such as “clean Xcode” or “delete everything,” remain audit-only.
- Default ordinary files and directories to Trash. Explain that space is not reclaimed until Trash is emptied, and request separate approval before doing so.
- Revalidate identity and running-process usage immediately before mutation.
- Never use wildcards, unbounded
rm -rf,sudo, SIP changes, or manual deletion inside system-managed CoreSimulator and MobileAsset directories. - Preserve archives and dSYMs unless the user proves the exact distributed build is backed up elsewhere and separately approves deletion.
- Never delete
Package.resolved, signing assets, credentials, custom DocSets, the active Xcode, a running simulator, or an Xcode that uniquely provides a required SDK/toolchain. - Report summed candidate sizes separately from actual APFS space recovered.
Read references/safety-model.md before applying cleanup.
Workflow
1. Preflight
- Confirm the host is macOS and the bundled script is available.
- Check for active Xcode, Simulator,
xcodebuild, Swift build, test, archive, and package-resolution processes. - Treat an open Xcode or Simulator application as context, not an automatic cleanup blocker. Defer an affected cleanup when a build, test, archive, or package-resolution process is active, an exact candidate is actively needed by that operation, or a supported deletion API requires the app/device to be stopped.
- Do not ask the user to quit apps merely because they appear in
ps. For regenerable documentation caches, moving an old cache while Xcode is open may temporarily remove local documentation until it reloads, but does not require quitting Xcode or Simulator. State that cost instead. - Determine optional source roots for project-local
.buildand.derived-datascanning. Do not crawl the whole home directory.
What ships with it
6 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 251 lines · 147 tokens per session scan B b070fa1eb851
xcode-disk-cleanup is a skill published in the GitHub repository AvdLee/Xcode-Disk-Cleanup-Agent-Skill (51 stars, last pushed 26d ago), licensed MIT. It adds 147 tokens to every session and 2,774 once invoked, about $0.0007 per session on Opus 5. A static security scan graded it B with 1 finding (recursive force delete). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
baocut
BaoCut-only operator for the installed bcut CLI and .bcut projects. Implicitly trigger only when the request explicitly names BaoCut or bcut, targets a .bcut project or BaoCut Subtitle Studio, or continues a BaoCut workflow already established in the conversation. Do not trigger solely for generic audio or video…
ui-audit-and-predicates
QA a screen with the deterministic audit action and target elements with structured predicates instead of hand-parsing describe-ui trees or eyeballing screenshots. Use for fleet QA sessions checking touch targets, clipping, alignment, spacing, safe-area violations, and missing accessibility labels.
run-the-daemon
Run or restart the manzanasd daemon - flags, warm pool and safety gates, journal settings, 503 overloaded handling, and install/upgrade via deploy/install.sh. Use when starting a daemon on a Mac host, tuning its gates, or debugging boot refusals.
broker-federation
Run manzanas-broker to federate multiple manzanasd daemons behind one endpoint, including daemons only reachable through an SSH tunnel. Use when agents need "any matching simulator in the fleet" placement across several Macs.
journal-evidence-export
Fetch run evidence from the manzanasd journal - list runs, page entries, download screenshot artifacts, upload your own evidence, and export a PR-ready markdown summary. Use when a task needs proof of what was done on a simulator (QA evidence for a PR, debugging a run).
run-evals
Run the manzanas-eval scenario benchmark harness against a daemon and read its determinism report. Use when validating daemon changes end-to-end, measuring action latency, or checking reset determinism.