Release Notes

agentsmgr 1.0a3 (2025-10-23)

Enhancements

  • Agent: Add python-annotator agent for Python type annotation tasks including adding type hints, resolving Pyright issues, creating type stubs, and analyzing type: ignore pragmas.

  • CLI: Add instruction file downloading support to populate command with automatic retrieval from Git sources, flexible glob pattern filtering, and per-file preprocessing with header stripping support.

  • CLI: Automatically manage git exclusion of instruction files through .git/info/exclude when populate command runs, respecting dynamic configuration changes.

  • CLI: Populate command now skips items with missing content and logs warnings instead of failing the entire operation, enabling iterative development without complete content sets.

Repairs

  • CLI: Fix template path resolution for OpenCode coder which was causing silent failures during content population due to singular/plural directory name mismatch.

agentsmgr 1.0a2 (2025-10-22)

Enhancements

  • Add support for Git worktrees and GIT_DIR environment variable in populate command, improving compatibility with non-standard Git configurations.

  • Support relative paths in populate command source argument without requiring explicit ./ prefix. The source resolution now uses standard URL scheme extraction for simplified path handling.

Repairs

  • Fix Git source handler URL scheme registration for correct path resolution in populate command source argument.

  • Fix populate command to respect per-coder mode defaults when creating coder directory symlinks and properly handle interrupted operations for idempotent behavior.

agentsmgr 1.0a1 (2025-10-21)

Enhancements

  • CLI: Add –profile parameter to populate command for specifying alternative Copier answers file paths, enabling testing with different configurations.

  • CLI: Add –tag-prefix parameter to populate command for filtering Git version tags by prefix, with semantic version comparison for accurate tag selection.

  • CLI: Add support for Gemini CLI as a target coder for command generation (agent generation not supported by Gemini CLI).

  • CLI: Add support for Qwen Code as a target coder, including agent and command generation with YAML frontmatter templates and MCP server configuration.

  • Git: Automatically add generated symlink names to .git/info/exclude during populate command to prevent accidental commits of generated content.

  • Performance: Optimize GitHub and GitLab repository processing with API-based tag resolution and shallow cloning, providing 3-50x speedup for population operations.

Notices

  • CLI: Change populate command to use positional arguments for source and target instead of –source and –target flags. This is a breaking change for existing scripts.

  • CLI: Split command-line interface into user-facing (agentsmgr) and maintainer-facing (agentsmgr-maintain) tools. The validate command is now available via agentsmgr-maintain.

agentsmgr 1.0a0 (2025-10-12)

Enhancements

  • CLI: Add Git ref specification support with @ref syntax for targeting specific branches, tags, or commits, with automatic latest tag fallback.

  • CLI: Add Git source support with github:, gitlab:, and git+https: URL schemes for pulling agent configurations from remote repositories.

  • CLI: Add agentsmgr command-line tool with detect, populate, and validate subcommands for managing AI agent configurations.

  • CLI: Add automatic memory file symlink creation for coder-specific filenames (CLAUDE.md, AGENTS.md) pointing to shared project conventions.

  • CLI: Add multi-target support with per-user and per-project targeting modes, including intelligent symlink management for seamless AI tool integration.

  • Comprehensive slash command and agent configuration library for Python development, releases, architecture documentation, and project management.

  • Hybrid distribution architecture combines Copier templates for base configuration with agentsmgr CLI for dynamic content generation from structured data sources.

  • Multi-AI tool support for Claude Code, Opencode, and future AI development environments with extensible configuration management.

  • Plugin architecture with extensible source handlers (git, local) and renderers (Claude, Opencode, Codex) using decorator-based registration system.

  • Tag-based release system enables rapid configuration distribution using agents-N versioning scheme for atomic, consistent deployment.