メニュー
Organize per-repo agentic project workspaces under .agents/projects/<project-name>. Use when initializing or maintaining project prompts, plans, research, and temporary artifacts.
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
SOC 職業分類に基づく
Set up QMD in a repository with repository scanning, collection planning, YAML-defined collections, and approval-gated execution. Use when I need local markdown indexing/search initialized or adjusted.
Search local markdown knowledge bases, notes, docs, and wikis with QMD. Use when users ask to find notes, retrieve documents, inspect a wiki, or answer from indexed markdown.
Execute deterministic story plans through dependency-aware workflow state transitions, PR/card lifecycle discipline, and critique-gated completion. Use when implementing an existing multi-story plan artifact.
Use when creating, editing, auditing, or restructuring agent instruction entrypoints (`SKILL.md`, `AGENTS.md`, `CLAUDE.md`, subagent/agent definitions). Consolidate duplicate or conflicting rules, improve skill description discoverability, and validate metadata, assets, and cross-file consistency.
Produce deterministic execution plans with story artifacts ready for workflow-driven delivery. Use when tasks are multi-step, high-risk, ambiguous, or span multiple files/services.
Create branches and commits from local diffs with correct commit message conventions. Use when asked to branch, split commits, write commit messages, or apply commitlint/Conventional Commits rules.
| name | agentic-projects |
| description | Organize per-repo agentic project workspaces under .agents/projects/<project-name>. Use when initializing or maintaining project prompts, plans, research, and temporary artifacts. |
| allowed-tools | Bash(bash ~/.agents/skills/agentic-projects/scripts/manage-agentic-project.sh) |
| metadata | {"scripts":"scripts/manage-agentic-project.sh"} |
Manual trigger only.
Run the colocated script to initialize or update:
.agents/projects/<project-name>/
PROMPT.md: canonical project prompt + working instructions for you.plans/<plan-name>.md: one file per plan iteration or variant. Use for persisted planning history and persisted subagents orchestration synchronization artifacts that should be versioned.research/<research-item-name>.md: one file per research artifact or finding.tmp/: ephemeral artifacts (large outputs, reusable temp scripts, intermediate data, non-persisted subagents orchestration synchronization artifacts). Delete contents when project is complete..gitignore: ignores tmp/* so temporary artifacts are not committed.--project <project-name>--prompt "<initial prompt>" or --prompt-file <path>--force only when explicitly overwriting an existing PROMPT.md.The script handles creation of missing directories/files and returns clear errors for invalid input or conflicting state.