Agent Layer integration

Agent Layer support is beta.

After al init, tokenjuice install agent-layer writes .agent-layer/instructions/tokenjuice.md in the current workspace. Agent Layer treats .agent-layer/ as the source of truth and syncs instructions into generated client config files.

Install

al init
tokenjuice install agent-layer
al sync
tokenjuice doctor agent-layer

Uninstall

tokenjuice uninstall agent-layer
al sync

Behavior

• The source instructions tell synced client configs to prefer tokenjuice wrap -- <command> for terminal commands likely to produce long output.
• The source instructions tell downstream agents to treat compacted output as authoritative.
• The only documented escape hatch is tokenjuice wrap --raw -- <command>.
• Existing .agent-layer/instructions/tokenjuice.md content is backed up before install.
• Install refuses to create .agent-layer/ by itself; initialize the Agent Layer project with al init first.
• Uninstall removes the source instructions; run al sync afterward so generated client files drop any previously synced tokenjuice guidance.

Current beta caveat

Agent Layer is a config sync layer, not a command runtime. This integration does not intercept shell output; it adds tokenjuice source instructions that Agent Layer can sync to generated client files.