Agentloom integration

Agentloom support is beta.

tokenjuice install agentloom writes .agents/rules/tokenjuice-agentloom.md in the current workspace. Agentloom treats .agents/ as the canonical source for agents, commands, rules, skills, and MCP definitions, then syncs those definitions into provider-native config files.

Install

tokenjuice install agentloom
agentloom sync
tokenjuice doctor agentloom

Uninstall

tokenjuice uninstall agentloom
agentloom sync

Behavior

• The source rule tells synced provider-native agent configs to prefer tokenjuice wrap -- <command> for terminal commands likely to produce long output.
• The source rule tells downstream agents to treat compacted output as authoritative.
• The only documented escape hatch is tokenjuice wrap --raw -- <command>.
• Existing .agents/rules/tokenjuice-agentloom.md content is backed up before install.
• Uninstall removes the source rule; run agentloom sync afterward so provider-native configs drop any previously synced tokenjuice guidance.
• The file name is host-specific so it can coexist with other .agents/rules/ integrations.

Current beta caveat

Agentloom is a config sync layer, not a command runtime. This integration does not intercept shell output; it adds a tokenjuice source rule that Agentloom can sync to the provider configs it manages.