MBTI Typing Skill Agent Adapter Lab
Local-first Agent Adoption Lab
Agent Adoption Lab
Select an agent stack and leave with a copyable installation path.
This page keeps the maintained surface small: the Core Pack is first-class for Codex, Claude Code, Cursor, and opencode. Extended manifest recipes remain searchable for teams that need them, but the default path is lean, local-first, low-typing, and preserves the same candidate set, serious runner-up, evidence ledger, falsifier, and safety boundary across tools.
Targets
0 visible
0
Selected agent targets
0
Estimated exported file groups
Export Path
Copyable route1
Manifest source
agent-adapters/manifest.json records source URLs, install notes, entrypoints, and the shared adapter contract.
2
Pack export
scripts/export_agent_pack.py copies selected adapters plus the canonical skill and writes AGENT_PACK_MANIFEST.json.
3
Audit loop
scripts/agent_adapter_audit.py, scripts/agent_pack_export_audit.py, and scripts/agent_adapter_lab_audit.py keep the public surface aligned.
source-of-truth sync
Generated data must match the manifest exactly before release.
Safety boundary
No clinical, hiring, legal, medical, financial, or deterministic use.
Sticky precision
Adoption is anchored in falsifiable typing, not flattery or label lock.
Low-typing questions
Use native question UI when present, otherwise compact choices with an Other escape hatch.