Skill Budget Alert (per-engine skill overload)

What it is

When Omniscio syncs your skills into other AI CLIs (Codex, Gemini, Cursor), a strict-loader engine like Codex can receive more skills than it can handle and warns "skills over length". This feature notices that and lets you decide what to hold back — it never cuts skills on its own.

It is the fan-out sibling of the Skill Overhead alert: that one watches the skills loaded into your Claude context; this one watches how many skills get copied to each other engine.

Where to find it

The review screen arrives as an inbox row; the budgets and exclusions live under Settings → Sync Config to Other AI CLIs.

How it behaves

How it works

• A per-engine budget. Each copying engine has a skill-count budget you can tune. Codex ships with a real default (its loader actually has a limit); Gemini and Cursor default to off (no known aggregate limit) until you set one. A budget of 0 means off for that engine.
• An alert when an engine is over budget. After a sync, if an engine is set to receive more skills than its budget, an inbox alert appears ("Codex may be getting too many skills"). It clears itself once you're back under budget. Its one button opens the review screen.
• You pick what to exclude. The review screen (Settings → Sync Config to Other AI CLIs → Skill budget per engine) lists each engine's skills with a toggle to hold each one back from that engine. Excluding a skill:
  • keeps it fully installed and usable everywhere else (in Claude and the other engines) — it's only held back from the one engine you toggled;
  • removes its already-copied folder from that engine on the next sync;
  • is reversible — toggle it back on to restore it.

Nothing is ever excluded automatically — the alert only points you at the screen; the choices are yours.

Which engines it covers

Only the engines that copy a per-engine set of skills: Codex and Gemini (into a global skills folder) and Cursor (into each picked project's .cursor/skills). OpenCode is not covered — it reads your ~/.claude/skills folder directly, so there's no separate copy to hold a skill back from; its load is the one the Skill Overhead alert already watches.

For agents

Settings

• Budgets + exclusions live under Settings → Sync Config to Other AI CLIs (only shown when "Sync skills" is on). Both are also controllable from the command line, since they're plain settings (providerConfigSyncSkillBudgets, providerConfigSyncSkillExclusions).

Related

• skills-bloat-alert.md — the sibling alert for your Claude skill context.
• provider-config-sync.md — the feature that syncs your config (MCP, instructions, skills) into the other AI CLIs, where this review screen lives.

See the contract for the full invariant list: .claude/memory/contracts/skill-budget-alert-contract.md.