Custodian Agent Skill

Monitors agent gateway logs, cron jobs, skill journals, and OCAS data directories for operational failures.

Context cost
188.7k tokensestimated from the bundle, loaded when it triggers
Bundle
200 of 313 fileshanded out here; the rest is at its source
Licence
MITthe author’s own terms
Last change
—v3.3.0
Servers it uses
Noneruns standalone

What it does

Monitors agent gateway logs, cron jobs, skill journals, and OCAS data directories for operational failures. Detects errors, applies safe non-destructive fixes autonomously during quiet hours, escalates the rest. RCA on recurring errors with fix-loop detection and confidence-tier auto promote/demote. Use when: cron jobs fail or show stale errors, gateway logs show repeated error patterns, skill journals have gaps, disk usage exceeds thresholds, MCP servers crash-loop, or after any gateway restart.

What this skill changes about the agent has not been set out here as a list. Its SKILL.md, shown below, says it in the author’s own words.

agentanalytics

The skill itself

A skill is instructions the model reads, so you can see what you are getting: below is the start of its SKILL.md. The front matter loads with every session, the body loads when the skill triggers, and the whole file comes with the install.

SKILL.md14.4 kB · preview
--- warning: 'FALSE TRIGGER RISK: Has had 50% false trigger rate on interactive loads. Do NOT load for OKR trend analysis, skill design evaluation, behavioral lesson extraction, briefing delivery, or social graph queries unless the user explicitly requests custodian for system health monitoring, cron job health checks, log compaction, or data-dir health. Added automatically on 2026-10-13.' name: ocas-custodian license: MIT source: https://github.com/<agent-handle>/custodian includes: - references/** - scripts/** metadata: author: Indigo Karasu (indigokarasu) version: "3.3.0" hermes: tags: - monitoring - system-health - log-analysis - cron - OCAS-core category: devops config: - key: HERMES_HOME default: "~/.hermes" - key: HERMES_PROFILE default: "indigo" - key: HERMES_PROFILE_HOME default: "~/.hermes/profiles/indigo" - key: AGENT_ROOT default: "~/.hermes" triggers: - system health - log errors - cron failures - skill journal errors - operational monitoring ---
43# Custodian
44
45Enforces the spec-ocas-recovery.md contract: every scheduled run writes evidence; schedule gaps trigger remedial passes; degraded mode is explicit, never a silent skip; self-repair re-validates.
46
47## When to Use
48
49Interactive invocation offers a two-level menu: references/interactive-menu.md.
50
51- System health monitoring and alerting
52- Skill library audits (conformance, freshness, coverage)
53- Cron job health checks
54- Log compaction and disk space monitoring
55- After any major system change — verify integrity
56
57## When NOT to Use
58
59- Real-time monitoring (use heartbeat instead)
60- Skill creation or modification (use Forge)
61- Content generation or research
This is a preview of SKILL.md. Download or install the skill to get the whole thing.
In the file
SKILL.md1,874 words
Files200
LicenceMIT
Why you can read it

Nothing in a skill executes. The client loads the text and the model follows it, so a skill can be audited the way a runbook is — by reading it.

What it costs in context

Skills are not billed by the call. They are paid for in context: every token the instructions occupy is a token your code, your diff and your conversation cannot use. Here is what this one takes and when it takes it.

≈560
always loaded
The name and description, so the model knows the skill exists and when to reach for it.
188,140
on trigger
The instruction body and 199 supporting files, read only when the skill fires.
94.3%
of a 200k window
Ten skills this size would not fit in the window together.
050k100k150k200k context window

188.7k tokens, estimated from the bundle at four bytes to the token, held for the rest of the session once it triggers. Heavy. Teams tend to install this one per project rather than globally, and load it only when the job comes up.

Servers bill, skills cost

A server charges by the month. A skill charges once per session, in context, and then keeps charging it for as long as the session lives.

Before and after

The same question, put to the same model twice: once as it comes, and once with these instructions loaded.

No worked example has been published for this skill yet.

The procedure it runs

The procedure has not been set out here step by step. It is in the body of its SKILL.md: the start of it is shown above, and the whole file comes with the install.

Prose, not code

These steps are written for a model to follow, not executed by a runtime. It can still be told to skip one, and it will say so when it does.

Servers it uses

None. This skill calls no MCP servers at all.

Everything it needs is in the instructions, so it works in a project with nothing connected — the model reads the file and changes how it works with what it can already reach.

Its SKILL.md does not declare which tools it asks the agent for, so what it touches is whatever the instructions themselves ask for — read them before you install it.

What it asks for
Writes filesnot declared
Network accessnot declared

This SKILL.md carries no allowed-tools line, so neither is declared. A skill grants no permissions of its own — it can only ask for tools your client already has.

What it will not do

Every skill is narrow, and the useful ones say where they stop. These are the jobs this one is the wrong tool for.

What this skill is not for has not been published here. Nothing is implied by that: it is a section the author has not filled in.

What is in the bundle

The download carries 200 of the 313 files in its source folder; the list here names 200 files, 754.8 kB on disk. A bundle is text throughout: the instructions the model reads, plus the templates it fills in. The other 113 are not in the download, nor are at least 6 files of types a skill download does not carry — the first 100 are listed after these, with why.

  • SKILL.md14.4 kB
  • CHANGELOG.md3.8 kB
  • config.json0.1 kB
  • README.md7.9 kB
  • references/503-scope-expansion-2026-07-28.md1.3 kB
  • references/api_endpoints.md5.3 kB
  • references/background-tasks.md1.9 kB
  • references/backup-disk-full-symlink-gotcha.md3.5 kB
  • references/backup-github-lfs-push-failure.md3.5 kB
  • references/bones-kalshi-creds-missing.md1.1 kB
  • references/browser-cdp-502-loop-pattern.md1.6 kB
  • references/cf-null-vs-zero-pattern.md1.5 kB
  • references/checkpoint-store-git-corruption-pattern.md1.3 kB
  • references/chronicle-daily-embed-timeout-pattern.md2.4 kB
  • references/chronicle-daily-embed-timeout-remediation.md5.1 kB
  • references/chronicle-db-malformed-stale-premise-verify-2026-07-15.md3.6 kB
  • references/chronicle-embed-forward-stale-resolution-20260714.md3.9 kB
  • references/chronicle-enrich-false-backlog-2026-07-15.md3.1 kB
  • references/chronicle-plugin-dirs-empty-pattern.md2.6 kB
  • references/compression-model-moondream-misconfig.md3.4 kB
  • references/confidence-model.md6.0 kB
  • references/config-empty-section-fixloop-status.md3.9 kB
  • references/config-null-key-verification.md2.5 kB
  • references/config-recovery.md2.9 kB
  • references/conformance.md6.7 kB
  • references/context-engine-chronicle-not-found-pattern.md2.0 kB
  • references/credential-leak-backup-commit-pattern.md5.1 kB
  • references/critical-pitfalls.md24.8 kB
  • references/cron-fastforward-trap.md1.9 kB
  • references/cron-health-2026-06-28-2111.md2.3 kB
  • references/cron-health-check-direct-jobs-json-parsing.md9.2 kB
  • references/cron-health-check-from-cron-context.md2.8 kB
  • references/cron-json-write-heredoc-variable-expansion-failure.md2.9 kB
  • references/cron-output-verification-gap.md2.2 kB
  • references/cron-provider-error-transient-pattern.md4.2 kB
  • references/cron-script-path-home-pattern.md2.4 kB
  • references/cron-script-path-security-model.md5.5 kB
  • references/cron-timeout-first-occurrence-pattern.md2.0 kB
  • references/custodian-core-fingerprints.md6.5 kB
  • references/custodian-gotchas.md8.9 kB
  • references/custodian-pitfall-map.md1.3 kB
  • references/custodian-self-failure-fallback-model.md2.9 kB
  • references/dead-skill-ref-active-job-fix.md2.3 kB
  • references/deep-scan.md34.2 kB
  • references/deep-scan-2026-05-30.md2.7 kB
  • references/deep-scan-2026-05-30-2100.md1.8 kB
  • references/deep-scan-2026-05-31-0215.md5.2 kB
  • references/deep-scan-2026-05-31-1400.md1.3 kB
  • references/deep-scan-2026-06-04.md2.4 kB
  • references/deep-scan-2026-06-06.md1.9 kB
  • references/deep-scan-2026-06-13-0810.md2.0 kB
  • references/deep-scan-2026-06-14-0215.md4.4 kB
  • references/deep-scan-2026-06-14-0800.md1.1 kB
  • references/deep-scan-2026-06-16-0215.md2.0 kB
  • references/deep-scan-2026-06-16-1730.md3.1 kB
  • references/deep-scan-2026-06-16-2000.md1.7 kB
  • references/deep-scan-2026-06-17.md2.2 kB
  • references/deep-scan-2026-06-17-1415.md2.1 kB
  • references/deep-scan-2026-06-17-2000.md1.6 kB
  • references/deep-scan-2026-06-18.md2.3 kB
  • references/deep-scan-2026-06-20.md2.4 kB
  • references/deep-scan-2026-06-20-0200.md2.0 kB
  • references/deep-scan-2026-06-21-0215.md2.4 kB
  • references/deep-scan-2026-06-23.md1.6 kB
  • references/deep-scan-2026-06-24-0550.md1.9 kB
  • references/deep-scan-2026-06-25-1500.md2.1 kB
  • references/deep-scan-2026-06-27-1405.md2.0 kB
  • references/deep-scan-2026-06-28-2105.md1.9 kB
  • references/deep-scan-2026-06-28-clean-verdict.md3.1 kB
  • references/deep-scan-2026-07-14.md2.4 kB
  • references/deep-scan-clean-verdict-2026-06-23.md2.2 kB
  • references/deep-scan-clean-verdict-pattern.md2.4 kB
  • references/deep-scan-fix-loop-prehandled-silent-verdict.md2.1 kB
  • references/deep-scan-stale-error-verification.md4.6 kB
  • references/disk-compaction.md7.2 kB
  • references/dispatch-triage-evening-no-agent-script-args.md3.7 kB
  • references/dispatch-triage-morning-no-agent-script-args.md4.5 kB
  • references/divergent-branch-handling.md2.3 kB
  • references/dormant-plugin-defect-reopen-2026-07-24.md3.7 kB
  • references/duplicate-gateway-systemd-service-pattern.md2.7 kB
  • references/editable-install-path-discovery.md2.1 kB
  • references/email-check-invalid-grant-diagnostic.md3.6 kB
  • references/empty-plugin-dir-detection.md0.7 kB
  • references/escalation-deepscan-pitfalls-2026-07.md4.5 kB
  • references/escalation-exec-default-token-expired-2026-07-09.md2.8 kB
  • references/escalation-exec-loop-live-classifier.md3.7 kB
  • references/escalation-execution-loop.md25.7 kB
  • references/escalation-false-recovered-note-trap.md3.3 kB
  • references/escalation-gap-scan-refs-vs-issue-id-false-positive-2026-07-24.md2.5 kB
  • references/escalation-gap-taxonomy-token-and-oauth-discriminator-2026-07-16.md5.6 kB
  • references/escalation-gateway-restart-marker-stale-false-escalation-2026-07-24.md4.3 kB
  • references/escalation-loop-exit1-wrapper-stale-and-gap-falsepositive.md5.0 kB
  • references/escalation-loop-issue-status-scan-trap.md2.4 kB
  • references/escalation-loop-pitfalls.md9.6 kB
  • references/escalation-loop-resolved-patterns.md3.3 kB
  • references/escalation-loop-scheduled-run-sequence-2026-07-22.md4.4 kB
  • references/escalation-loop-tier4-code-defect-verify.md4.8 kB
  • references/escalation-persistence-gap.md2.7 kB
  • references/escalation-provider-recovery-verify.md5.5 kB
  • references/escalation-runner-2026-06-01.md6.6 kB
  • references/escalation-runner-2026-06-01-1600.md2.0 kB
  • references/escalation-runner-2026-06-01-2113.md1.9 kB
  • references/escalation-runner-2026-06-01-2206.md1.9 kB
  • references/escalation-runner-2026-06-01-2340.md1.0 kB
  • references/escalation-runner-2026-06-02.md0.3 kB
  • references/escalation-runner-2026-06-04.md4.3 kB
  • references/escalation-runner-2026-06-04-23xx-script-path-fix.md1.6 kB
  • references/escalation-runner-2026-06-04-1539.md1.6 kB
  • references/escalation-runner-2026-06-04-1807.md3.2 kB
  • references/escalation-runner-2026-06-06.md1.4 kB
  • references/escalation-runner-2026-06-08.md3.0 kB
  • references/escalation-runner-2026-06-08-1915.md1.4 kB
  • references/escalation-runner-2026-06-08-2315.md1.9 kB
  • references/escalation-runner-2026-06-15-1108.md3.3 kB
  • references/escalation-runner-2026-06-15-1123.md2.0 kB
  • references/escalation-runner-2026-06-15-2100.md2.0 kB
  • references/escalation-runner-2026-06-17.md1.7 kB
  • references/escalation-runner-2026-06-18-0925.md1.4 kB
  • references/escalation-runner-2026-06-23-0940.md2.0 kB
  • references/escalation-runner-2026-06-24-1607.md2.3 kB
  • references/escalation-runner-2026-06-24-2108.md3.8 kB
  • references/escalation-runner-already-classified-fast-path.md3.6 kB
  • references/escalation-runner-clean-verdict-pattern.md3.2 kB
  • references/escalation-runner-concurrent-execution-gap.md2.1 kB
  • references/escalation-runner-journal-path-and-schema.md7.1 kB
  • references/escalation-runner-manual-execution.md3.2 kB
  • references/escalation-runner-multi-path-issues.md5.5 kB
  • references/escalation-runner-pause-affected-jobs-pattern.md5.3 kB
  • references/escalation-stale-issue-premise-verify.md3.0 kB
  • references/explicit-model-override-bypassing-default.md3.7 kB
  • references/fallback-model-manifest-build-401-pattern.md2.0 kB
  • references/fallback-model-null-yaml-debris-pattern.md3.2 kB
  • references/fix-applied-pending-restart-sweep.md1.9 kB
  • references/fix-effectiveness-schema-contamination.md3.4 kB
  • references/fix-missing-skill-data-directories-2026-06-30.md1.4 kB
  • references/fix-missing-skill-journals-directories-2026-06-30.md1.5 kB
  • references/fix-safety.md7.0 kB
  • references/fos-nameerror-pattern.md3.3 kB
  • references/gateway-instance-collision-noise-pattern.md1.5 kB
  • references/gateway-log-frozen-stale-pattern.md3.1 kB
  • references/gateway-log-live-vs-stale-path-2026-07-24.md2.0 kB
  • references/gateway-log-timestamp-range-filtering-pitfall.md6.9 kB
  • references/gateway-log-traceback-gap-detection.md5.3 kB
  • references/gateway-restart-health-signal.md3.1 kB
  • references/gateway-restart-import-window-pattern.md3.8 kB
  • references/google-oauth-client-deleted-pattern.md6.4 kB
  • references/google-search-build.md1.8 kB
  • references/google-workspace-auth-troubleshooting.md5.0 kB
  • references/hardline-filter-gateway-log-grep-block-pattern.md2.1 kB
  • references/hermes-cron-run-verify-recipe.md5.4 kB
  • references/interactive-menu.md1.8 kB
  • references/issues-jsonl-parse-pitfalls.md5.4 kB
  • references/jobs-json-timestamp-offset-misread-pitfall.md1.7 kB
  • references/jobs-not-running-diagnostic.md5.4 kB
  • references/journal-escalation-stale-premise-guard-2026-07-14.md1.8 kB
  • references/journal-path-format-inconsistency.md3.1 kB
  • references/kanban-dispatcher-index-gap.md3.9 kB
  • references/kanban-dispatcher-stuck-diagnostic.md5.2 kB
  • references/kanban-dispatcher-stuck-pattern.md2.6 kB
  • references/known_issues.json39.4 kB
  • references/known-code-fixes-and-cascade.md9.6 kB
  • references/known-script-auth-issues.md3.1 kB
  • references/light-scan-2026-05-19.md2.7 kB
  • references/light-scan-2026-05-20.md3.8 kB
  • references/light-scan-2026-05-30.md1.1 kB
  • references/light-scan-2026-05-31.md2.6 kB
  • references/light-scan-2026-06-04-1609.md3.7 kB
  • references/light-scan-2026-06-04-2114.md3.8 kB
  • references/light-scan-2026-06-08-1805.md3.0 kB
  • references/light-scan-2026-06-08-2100.md2.0 kB
  • references/light-scan-2026-06-14-1400.md1.1 kB
  • references/light-scan-2026-06-14-1500.md1.1 kB
  • references/light-scan-2026-06-17-2000.md2.3 kB
  • references/light-scan-2026-06-18-1130.md2.4 kB
  • references/light-scan-2026-06-23-0803.md2.9 kB
  • references/light-scan-2026-06-24-0700.md3.1 kB
  • references/light-scan-2026-06-24-0800.md4.0 kB
  • references/light-scan-2026-06-24-1001.md3.6 kB
  • references/light-scan-2026-06-24-200409Z.md2.6 kB
  • references/light-scan-2026-06-25-0700.md3.2 kB
  • references/light-scan-2026-06-25-0829.md2.5 kB
  • references/light-scan-2026-06-27-1805.md2.1 kB
  • references/light-scan-2026-06-28-2101.md2.8 kB
  • references/light-scan-2026-06-28-2204.md2.1 kB
  • references/light-scan-2026-06-29-0104.md1.6 kB
  • references/light-scan-2026-06-29-0210.md1.9 kB
  • references/light-scan-2026-06-29-0904.md2.6 kB
  • references/light-scan-2026-06-29-1204.md2.9 kB
  • references/light-scan-2026-06-29-2206.md2.7 kB
  • references/light-scan-conformance-and-postfix-recount-2026-07-12.md1.8 kB
  • references/light-scan-false-resolution-gotcha.md3.8 kB
  • references/light-scan-stale-no-agent-error-triage.md2.7 kB
  • references/live-vs-stale-provider-error-recipe.md2.7 kB
  • references/mcp-server-files-missing-pattern.md2.2 kB
  • references/mcp-server-module-deleted-pattern.md6.2 kB
  • references/mcp-server-reconnect-loop-escalation-pitfall.md3.5 kB
  • references/memory-guard-failure-pattern.md2.1 kB
  • references/monitor-journals-noop-readonly-verify.md4.7 kB
  • references/monitor-list-access-token-recurrence-durable-fix-2026-07-15.md3.8 kB
  • references/monitor-list-exit1-mask-gap.md2.5 kB

Not in the download

The first 100 are named here, those the skill reads first. Not carried is a type a skill download does not hold and over the limit is past 200 files or 4 MB a skill. Marked read: its SKILL.md or scripts name the file, so without it the skill works only in part. Take them from its source.

  • references/non-fatal-error-patterns.mdread · over the limit
  • references/okrs.mdread · over the limit
  • references/operational-gotchas.mdread · over the limit
  • references/plugin-vs-skill-architecture.mdread · over the limit
  • references/secret-audit.mdread · over the limit
  • references/spec-ocas-recovery.mdread · over the limit
  • references/support-file-map.mdread · over the limit
  • references/using-script.mdread · over the limit
  • scripts/race_safe_issue_patch.pyread · over the limit
  • scripts/verify_fixes_cron_run.pyread · over the limit
  • .gitignorenot carried
  • assets/readme/hero.jpgnot carried
  • scripts/backup_system.sh.bak.lbug-20260728-212624not carried
  • scripts/backup_system.sh.bak.purge-20260729-112119not carried
  • scripts/restore_all_hermes_dbs.sh.bak.elephas-20260729-004733not carried
  • scripts/restore_all_hermes_dbs.sh.bak.purge-20260729-112119not carried
  • references/monitor-list-keyerror-noninteractive-refresh-fix-2026-07-15.mdover the limit
  • references/monitor-list-keyerror-transient-creds-race-2026-07-14.mdover the limit
  • references/monitor-list-masked-keyerror-pitfall.mdover the limit
  • references/no-agent-agent-root-env-fix.mdover the limit
  • references/no-agent-exit1-verify-pipe-mask-pitfall.mdover the limit
  • references/no-agent-missing-dependency-install-procedure.mdover the limit
  • references/no-agent-missing-dependency-pattern.mdover the limit
  • references/no-agent-monitor-exit1-noop-readonly-classification.mdover the limit
  • references/no-agent-monitor-exit1-upstream-degraded-pitfall.mdover the limit
  • references/no-agent-script-argument-pattern.mdover the limit
  • references/no-agent-script-exit-1-deaggregation-pitfall.mdover the limit
  • references/no-agent-script-path-mismatch-symlink-fix.mdover the limit
  • references/null-provider-fallback-routing-2026-06-18.mdover the limit
  • references/observation-journal-schema.mdover the limit
  • references/oc-cron-script-not-found-transient-pattern.mdover the limit
  • references/oc-hook-post-tool-call-task-id-pattern.mdover the limit
  • references/openrouter-402-credits-exhausted-pattern.mdover the limit
  • references/openrouter-402-mass-pause-execution.mdover the limit
  • references/openrouter-502-provider-unavailable.mdover the limit
  • references/owl-alpha-404-diagnostic.mdover the limit
  • references/paused-untracked-real-error-gap.mdover the limit
  • references/pipe-to-interpreter-security-block.mdover the limit
  • references/pitfall-execute-code-import-isolation.mdover the limit
  • references/platform-compatibility.mdover the limit
  • references/plugin-self-update-2026-06-18.mdover the limit
  • references/post-fix-stale-error-pattern.mdover the limit
  • references/post-fix-wrapper-script-verification.mdover the limit
  • references/pre-update-state-snapshot-disk-refill.mdover the limit
  • references/provider-401-diagnosis.mdover the limit
  • references/provider-auth-token-expiry-fingerprint.mdover the limit
  • references/provider-recovery-false-positive.mdover the limit
  • references/provider-recovery-forward-stale-confirmation.mdover the limit
  • references/provider-recovery-light-scan-reconciliation.mdover the limit
  • references/provider-recovery-probe-wrong-endpoint-pitfall.mdover the limit
  • references/rally-health-check.mdover the limit
  • references/rca-backfill-2026-06-05.mdover the limit
  • references/rca-schema.mdover the limit
  • references/recurrence-fingerprint-class-distinct-job-2026-07-23.mdover the limit
  • references/redaction-placeholder-source-corruption.mdover the limit
  • references/resolved-codefix-regression-verify.mdover the limit
  • references/resolved-timeout-verify-drained-backlog.mdover the limit
  • references/root-cause-analysis.mdover the limit
  • references/runtime-error-triage.mdover the limit
  • references/schedule-optimization.mdover the limit
  • references/scheduler-state-lag-vs-execution-failure.mdover the limit
  • references/script-library-organization.mdover the limit
  • references/script-path-security-block-pattern.mdover the limit
  • references/self-improvement.mdover the limit
  • references/self-resolved-module-verification-pattern.mdover the limit
  • references/self-update.mdover the limit
  • references/send-email-template-mismatch.mdover the limit
  • references/skill-library-hygiene.mdover the limit
  • references/skill-reference-path-mismatch-pattern.mdover the limit
  • references/skill-update-rebase-conflict-batch-pattern.mdover the limit
  • references/soul-md-truncation-pattern.mdover the limit
  • references/sqlite3-readonly-monkeypatch-fix.mdover the limit
  • references/stale-error-message-script-field-mismatch.mdover the limit
  • references/stale-error-state-pause-resume-fix.mdover the limit
  • references/stale-jobs-json-false-positive.mdover the limit
  • references/stale-model-error-diagnostic-pattern.mdover the limit
  • references/stale-provider-error-pause-loops.mdover the limit
  • references/state-db-oversized-vacuum-infeasible.mdover the limit
  • references/state-db-pruning-execution-procedure.mdover the limit
  • references/state-db-size-threshold.mdover the limit
  • references/subdirectory-hints-home-dir-pattern.mdover the limit
  • references/subprocess-cascade-oauth-masking.mdover the limit
  • references/subprocess-cascade-reverification-pitfall.mdover the limit
  • references/system-maintenance.mdover the limit
  • references/thread-renamer-backfill-hang.mdover the limit
  • references/timezone-schedule-window-false-positive.mdover the limit
  • references/transient-401-self-resolution-pattern.mdover the limit
  • references/transient-provider-errors.mdover the limit
  • references/verify-plugin-defect-postrestart-false-dormant-2026-07-25.mdover the limit
  • references/verify-plugin-defect-postrestart-false-live-bug.mdover the limit
  • references/vision-model-invalid-chatcompletion-pattern.mdover the limit
  • references/watchdog-degraded-false-positive-loop.mdover the limit
  • references/web-search-protocol.mdover the limit
  • references/workspace-mcp-binary-fix.mdover the limit
  • references/wrong-path-prefix-in-skill-scripts.mdover the limit
  • scripts/backup_all_hermes_data.shover the limit
  • scripts/backup_system.shover the limit
  • scripts/bucket_error_jobs.pyover the limit
  • scripts/chronicle_embed_backlog_probe.pyover the limit
  • scripts/classify_error_jobs.pyover the limit
What is not in it

No dependencies and nothing executable: a skill is text the agent reads, so the download is 200 files — not the whole of its source: at least 119 files are left out, and the skill reads 10 of them. The first 100 are listed under the files, with why. The MIT licence covers the templates and examples as well as the instructions.

Install

Installing copies the bundle into your project. Nothing runs at install time — the files sit on disk until the model reads them.

# Custodian · 188.7k tokens when loaded · macOS, Linux mkdir -p .claude/skills && t=$(mktemp -d) && curl -fsSL -o "$t/bundle.tar.gz" https://mcprush.com/api/skills/custodian/bundle.tar.gz && tar -xzf "$t/bundle.tar.gz" -C "$t" && [ -d "$t/custodian" ] && { [ ! -e .claude/skills/custodian ] || { mv .claude/skills/custodian "$t/.previous" && echo "Previous copy moved to $t/.previous"; }; } && mv "$t/custodian" .claude/skills/ && echo "Installed .claude/skills/custodian"
On Windows, or any OS with Node 20+
npx mcprush@latest skill add indigokarasu/custodian

Windows PowerShell says 'running scripts is disabled on this system'? Run Set-ExecutionPolicy -Scope CurrentUser RemoteSigned once, or type npx.cmd instead of npx (codex.cmd, gemini.cmd, grok.cmd likewise). PowerShell 7 needs neither. Writes to .claude/skills/custodian/ in the current project. For every project instead, run mkdir -p ~/.claude/skills && t=$(mktemp -d) && curl -fsSL -o "$t/bundle.tar.gz" https://mcprush.com/api/skills/custodian/bundle.tar.gz && tar -xzf "$t/bundle.tar.gz" -C "$t" && [ -d "$t/custodian" ] && { [ ! -e ~/.claude/skills/custodian ] || { mv ~/.claude/skills/custodian "$t/.previous" && echo "Previous copy moved to $t/.previous"; }; } && mv "$t/custodian" ~/.claude/skills/ && echo "Installed ~/.claude/skills/custodian" (macOS, Linux) — or add --global to the npx line. The curl line replaces .claude/skills/custodian/ only after the download succeeds, so files left there by another skill of the same name, or by an older release, do not mix in. A copy already there is moved aside, not deleted, and the line prints where. It appears as ocas-custodian in Gemini CLI, Codex, Copilot and Grok (its own name); Claude Code lists it by the folder name, custodian. Not in this download: at least 119 files of its source, 10 of them read by its SKILL.md or scripts (the list, and why). Take them from its source.

Which clients pick it up on their own

A skill is a folder of text. Most of these read a skills folder and pick it up without being told; the rest take the same folder another way — an upload in the app, a parameter on an API call — and then choose it by its description just the same.

Claude Code.claude/skills/
Claude DesktopCustomize › Skills
ChatGPTSkills › Create › Upload
Cursor.cursor/skills/
VS Code.github/skills/
Codex CLI.agents/skills/
Gemini CLI.gemini/skills/
Grok Build.grok/skills/
Microsoft CopilotSkills › Upload a skill
PerplexityComputer › Skills › Upload a skill
DeepSeek Harness.dsh/skills/
Zed.agents/skills/
Devin Desktop.devin/skills/
Windsurf (legacy).windsurf/skills/
OpenAI Agents SDKInstructions or sandbox Skills
HTTP / APIa request parameter
This release
Version3.3.0
Publishedno release date on file
PriceFree
Referenceindigokarasu/custodian

Versions

v3.3.0 is what is on the shelf; no release here carries a date. Instructions change more often than APIs do — a skill can be rewritten entirely without anything it depends on moving.

v3.3.0
  • No earlier releases have been published to the marketplace.
Pinning

The install command always fetches the release that is current the day you run it — v3.3.0 today — and cannot ask for an earlier one. The copy on your disk does not change until you run it again.

Reviews

—no reviews yet

Nobody has reviewed this skill yet. The rating is the mean of the reviews written here, so there is none until somebody writes the first.

Sign in to review it

A review is tied to an account so it can be answered, edited and taken down by the person who wrote it.

Sign in
Who can post

Anybody with an account, one review each. It carries the version that was current when you wrote it, and the publisher may reply once.

Publisher
Servers0
Claim this skill