Jmcomic

JMComic Skill

This skill enables you to interact with JMComic (18comic), a popular manga platform, to search, browse, and download manga content.

When to Use This Skill

Activate this skill when the user wants to:

• Search for manga by keyword or category
• Browse popular manga rankings (daily, weekly, monthly)
• Read album-specific or site-wide comments and nested replies, including spoiler flags
• List favorite folders, browse saved albums, add an album to favorites, or delete an album from favorites
• Download entire albums or specific chapters (Returns structured dict with status, paths, and metadata)
• Get detailed information about a manga album
• Configure download settings (paths, concurrency, proxies)
• Post-process downloaded content (Zip, PDF, LongImage) with native parameters or dir_rule
• Download the latest Android APK published by hect0x7/JMComic-APK
• Run jms --help to obtain the current upstream options, then invoke jms directly for an existing or newly downloaded manga directory

For APK download, local-reader startup, LAN safety, and download-to-read continuation rules, read references/ecosystem.md before acting.

📥 Download Tools Return Structured Data

Both download_album and download_photo return structured dictionaries:

download_album(album_id: str, ctx: Context = None) returns:

{
    "status": "success" | "failed",
    "album_id": str,
    "title": str,
    "download_path": str,  # Absolute path to download directory
    "duration": float | None,  # Total elapsed time in seconds
    "image_paths": list[str],  # Actual downloaded or cached image paths
    "export_files": dict[str, list[str]],  # Plugin outputs grouped by suffix
    "task_id": str,        # Dedicated ID for this download invocation
    "log_path": str,       # Absolute path to this task's log file
    "error": str | None
}

download_photo(photo_id: str, ctx: Context = None) returns:

{
    "status": "success" | "failed",
    "photo_id": str,
    "image_count": int,    # Number of actual image paths in this result
    "download_path": str,  # Absolute path to download directory
    "duration": float | None,  # Total elapsed time in seconds
    "image_paths": list[str],  # Actual downloaded or cached image paths
    "export_files": dict[str, list[str]],  # Plugin outputs grouped by suffix
    "task_id": str,        # Dedicated ID for this download invocation
    "log_path": str,       # Absolute path to this task's log file
    "error": str | None
}

Real-time Progress Tracking: Both methods accept an optional ctx: Context parameter (automatically injected by FastMCP). When provided, progress updates are sent via MCP notifications in real-time, allowing AI agents to monitor download progress.

Persistent Task Logs: Every download invocation creates an isolated log file and returns its task_id and log_path, including failed downloads. Logs default to ~/.jmcomic-ai/logs; set JM_TASK_LOG_DIR to use another directory.

Artifact Handoff: Use image_paths and export_files for packaging, sending, or cleanup instead of rescanning the configured directory. download_path comes from the completed upstream download result rather than a pre-download path prediction.

File-only Logging: Regular jmcomic, jmcomic_ai, and MCP runtime logs share ~/.jmcomic-ai/jmcomic_ai.log and are not emitted to stdout or stderr. Set JM_LOG_PATH to override the global log file. MCP protocol messages and explicit CLI result output are unaffected.

Upstream Rich Terminal Progress: jmcomic 2.7.7 includes the optional download_progress plugin with log_file and terminal_log_lines. Recommend it only when the user explicitly wants Rich progress in an interactive terminal. Do not enable it for jmai MCP/stdio downloads, which already use MCP Context notifications and isolated task logs.

Album Comments

Use get_album_comments(album_id: str, page: int = 1) to read one page of comments. This is a read-only tool; it does not post comments or replies.

{
    "album_id": str,
    "page": int,
    "page_size": int,
    "total": int | None,
    "page_count": int | None,
    "comment_count": int,
    "comments": [
        {
            "comment_id": str | None,
            "album_id": str | None,
            "user_id": str | None,
            "parent_comment_id": str | None,
            "content": str,
            "username": str,
            "nickname": str,
            "is_spoiler": bool,
            "created_at": object,
            "likes": int | None,
            "replies": [...]  # Recursive nested replies
        }
    ]
}

Use get_forum_comments(page: int = 1) for the latest site-wide comments. It returns the same comment objects and pagination fields without an album_id filter; each comment's own album_id identifies its source album. This is also read-only.

Favorites

All favorite tools require an authenticated client. Call login in the same MCP session or configure valid cookies. For HTML queries authenticated only by Cookie, pass username; after login, it can be omitted. The API client ignores username and always queries the current account. Separate script processes need authentication in their configuration; they do not inherit an MCP session's login.

get_favorite_folders(username: str = "") returns:

{"folders": [{"id": str, "name": str}]}

The directory can be empty. Use folder_id="0" to browse all favorites even if this ID is absent from the returned directory.

browse_favorite_albums(folder_id: str = "0", page: int = 1, order_by: str = "favorite_time", username: str = "") returns:

{
    "albums": [{"id": str, "title": str, "tags": list, "cover_url": str}],
    "total_count": int,
    "page": int,
    "folder_id": str,
    "error": str  # Present only for invalid page, folder ID, or sort parameters
}

Pagination starts at 1; folder_id="0" means all favorites. An empty result retains pagination and folder metadata. order_by accepts favorite_time (default) and update_time; the favorites list has its own upstream sort vocabulary, which differs from browse_albums. The service does not sort a page locally. Invalid arguments return an empty list with error; authentication and request failures propagate as tool errors, not empty collections.

add_favorite_album(album_id: str) returns:

{
    "status": "success" | "error",
    "album_id": str,
    "title": str,   # Album title
    "message": str
}

album_id accepts a numeric ID, a JM-prefixed ID, or an album URL and is normalized when parsing succeeds. The album is saved to the account's default favorites placement. Adding an album that is already saved returns status="error" and leaves the saved state unchanged.

delete_favorite_album(album_id: str) returns the same fields and takes the same album_id forms. Deleting an album that is not saved returns status="error" and leaves the saved state unchanged.

Both tools behave identically whichever client implementation the configuration selects, and both return the album title. Name every affected album as title (ID); when title is empty or absent, reuse an earlier search, browse, or detail result, or fetch the name with get_album_detail(album_id) or python scripts/album_info.py --id ID.

Core Capabilities

🛠️ Post-Processing

This skill supports advanced post-processing of downloaded manga. It returns structured data including the output path of the generated file.

• 📦 Zip Compression: Pack an entire album or individual chapters into a ZIP file.
• 📄 PDF Conversion: Merge all images of an album into a single PDF document.
• 🖼️ Long Image Merging: Combine all pages of a chapter into one continuous long image.

post_process(album_id: str, process_type: str, params: dict = None) returns:

{
    "status": "success" | "error",
    "process_type": str,  # Process type used
    "album_id": str,  # Album ID processed
    "output_path": str,  # Absolute path to generated file/directory (empty string on error)
    "output_paths": list[str],  # All files actually registered by the upstream plugin
    "is_directory": bool,  # True if output is a directory (e.g., photo-level zip), False on error
    "message": str  # Success/error message
}

All fields are always present. On error, output_path is empty, output_paths is empty, and is_directory is False.

Output Control: Use dir_rule (a {"rule": "DSL_STRING", "base_dir": "BASE_PATH"} dict) for custom output paths. If omitted, files are saved in the configured default directory. The DSL supports Bd (base_dir), Axxx/Pxxx album/photo attributes, and {attr} Python format placeholders. When looping over albums into one base_dir, include {Aid} or {Atitle} to avoid overwrites.

For the full set of ZIP/PDF/LongImg × album/photo dir_rule examples, see references/post_process.md.

This skill provides command-line utilities for JMComic operations. All utilities are Python scripts located in the scripts/ directory and should be executed using Python.

Data Structure Notes

Most search and browsing tools (e.g., search_album, browse_albums) return a consistent structure that supports pagination:

{
  "albums": [ ... ],
  "total_count": 1234
}

total_count provides the total number of items available across all pages, allowing you to calculate the number of remaining pages and decide if further searching is needed.

Important: Browse Albums Data Limitations

browse_albums is the unified tool for browsing albums by category, time_range, and order_by (ranking + category browsing in one interface). Its response is lightweight — each album has only id, title, tags, and cover_url, and no stats (likes/views/author). To get those, call get_album_detail(album_id) per album.

order_by accepts: latest (default), likes, views, pictures, score, comments. For the full enumeration, use-case recipes (rankings, category browsing, combined queries), and the "top 10 with details" workflow, see references/browse_albums.md.

Configuration Reference

For detailed configuration options, refer to:

• references/reference.md: Human-readable configuration guide
• assets/option_schema.json: JSON Schema for validation

Common configuration examples:

# Change download directory
dir_rule:
  base_dir: "/path/to/downloads"
  rule: "Bd / Ptitle"

# Adjust concurrency
download:
  threading:
    image: 30  # Max concurrent image downloads
    photo: 5   # Max concurrent chapter downloads

# Set proxy
client:
  async_impl: async_api  # Used by jmcomic native async APIs
  cache: level_option    # Reuse metadata across clients from this option
  postman:
    meta_data:
      proxies:
        http: "http://proxy.example.com:8080"
        https: "https://proxy.example.com:8080"

# Or use system proxy
client:
  postman:
    meta_data:
      proxies: system

# Configure login cookies
client:
  postman:
    meta_data:
      cookies:
        AVS: "your_avs_cookie_value"

# Use plugins
plugins:
  after_album:
    - plugin: zip
      kwargs:
        level: photo
        suffix: zip
        delete_original_file: true