plugin-readiness-audit

name	plugin-readiness-audit
description	Audit Antigravity workflows, skills, agents, and command surfaces to decide whether they should stay as prompts/workflows/skills or be packaged as repo-local or global Codex plugins. Use when the user asks about plugin readiness, workflow packaging, skill bloat, missing operating context, fresh-thread reliability, or Nate B Jones style plugin restructuring.

Plugin Readiness Audit

Use this skill to decide what should become a Codex plugin and what should stay as a workflow, skill, reference, or prompt.

Load Order

1. Read semantic_libraries/antigravity/primitives/workflow-packaging-ladder.md.
2. If the user is asking about broad restructuring, performance, or overengineering, run python3 execution/system_efficiency_benchmark.py first.
3. Run python3 execution/plugin_readiness_audit.py for the target workflow family.
4. Read the generated scorecard before recommending any plugin scaffold.

Default Candidate Set

When the user asks about the core Antigravity operating layer, score:

• autopilot
• mission
• orchestrate
• extraction-governor-agent
• knowledge-librarian
• health-check
• routing-intelligence
• self-evolve

Decision Rule

• Package only candidates scoring 80+ or a tightly related bundle whose average score is 80+.
• If a candidate scores 65-79, fix the workflow/skill layer before adding plugin packaging.
• If the issue is command bloat, routing drift, or underuse, prefer routing and usage instrumentation over a new plugin.
• If the benchmark says router/metadata cleanup performs as well as plugin packaging, cleanup comes first.

Required Output

Return:

• the scorecard path
• package now / improve first / keep as workflow decisions
• the first plugin candidate, if any
• required fresh-thread tests
• any path, permission, or marketplace risk