CLI Enhancement: Complete agent/developer workflow support #679
Description
Summary
The AD4M CLI already has solid foundation for programmatic interaction, but several gaps prevent complete workflows for:
- Autonomous AI agents (like Data, Hex)
- Human developers scripting
- CI/GitHub Actions automation
- Test frameworks
Current CLI Capabilities ✅
Neighbourhoods: create, join
Languages: list, info, publish, settings, template-apply
Agent: me (DID), status, lock/unlock, lookup by DID, generate
Perspectives: comprehensive link operations
Proposed Additions
High Priority
- Neighbourhood leave - Exit a neighbourhood
- Neighbourhood edit - Update neighbourhood metadata
- Language install - Install language from address (vs publish from file)
- Language uninstall - Remove installed language
- Subject class CRUD - Create/read/update/delete typed objects via CLI
- AD4M_EXECUTOR_URL env var - Default executor URL for cleaner CI scripts
Medium Priority
- Output formatting options - JSON, table, plain text flags
- Better error messages - Actionable guidance on failures
- Batch operations - Process multiple items in single command
Nice to Have
- SKILL.md documentation - Agent-friendly CLI reference
- Shell completions - Bash/Zsh/Fish autocomplete
Context
Discussion with @HexaField about making AD4M accessible to AI agents and automation workflows. The CLI already supports --executor-url for multi-node scenarios.
Related: MCP integration (PR #665) provides another interface, but CLI + SKILL.md approach has advantages for agents that work via shell commands.
cc @lucksus @HexaField
depends on d9011e34-8219-4ba1-bdd1-7d104a25cd1f/91215135-9143-46eb-8f5d-b38504a40392#682