filemaid

A local, AI-powered file organizer for macOS. It watches your Desktop and Downloads, classifies files with a local Ollama LLM, moves them into categorized archives, applies Finder tags, and quarantines uncertain items for review. It also cleans up stale development artifacts like Docker images, npm/cargo/pip caches, Homebrew packages, and Xcode DerivedData.

⚠️ Experimental: This project is under active development and may not work reliably in all environments. File movements, classifications, and cleanups can have side effects. Please review the code before running it on important data. Bug reports, issues, and pull requests are welcome to improve behavior.

Feedback: If something breaks or behaves unexpectedly, file an issue or open a pull request.

Changelog: See CHANGELOG.md for a version-by-version summary of new features and breaking changes.

Features

Automatic classification — files are classified by a local LLM using filename, extension, content snippets, image analysis, and directory context when --depth is used.
Tag + comment in place, move when you say so — by default files are classified, tagged, and commented in place. Pass --move (or set move_files: true) to relocate them into categorized archive folders or send review items to ~/.filemaid/review/.
Review-before-delete — uncertain files are flagged for review; with --move they go to ~/.filemaid/review/. Deletions only happen for explicitly safe patterns or duplicates.
Directory-aware deep classification — --depth descends into directories, detects project markers, and classifies directories read-only with keep|review|trash|archive recommendations.
LLM retry and review retry — transient Ollama failures are retried automatically, and filemaid review --retry re-processes review-queue items that failed due to transient errors.
History and undo — filemaid history shows recent runs; filemaid undo rolls back the most recent run.
Smart rename — optionally renames files based on LLM-suggested names when the suggested name quality meets your threshold, with duplicate and near-duplicate detection.
Smart Folders + Finder hub — automatically builds a ~/Documents/Filemaid hub with per-category Smart Folders, Archive/Review aliases, and a Finder sidebar pin.
Private & offline — no cloud services; everything runs locally via Ollama.
Single Go binary — one self-contained binary; only Cobra is used for the CLI.

Requirements

macOS 13+ on Apple Silicon (uses launchctl, xattr, mdimport, osascript)

Homebrew — package manager for macOS. Install it with:

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

Ollama running locally with a vision-capable model (install with brew install ollama) 0.30.0+

Quick Start

Install the latest release with the install script (no Go required):

curl -fsSL https://raw.githubusercontent.com/logicminds/filemaid/main/install.sh | bash

The script downloads the Apple Silicon binary from the GitHub releases page, verifies its checksum, and installs it to ~/.local/bin/filemaid. Only macOS on Apple Silicon (arm64) is supported; the script reports an error on any other platform.

To install to a different directory, set INSTALL_DIR:

curl -fsSL https://raw.githubusercontent.com/logicminds/filemaid/main/install.sh | INSTALL_DIR=/usr/local/bin bash

Or download the release manually:

Go to the latest release.
Download filemaid-darwin-arm64 and filemaid-darwin-arm64.sha256.

Verify the checksum:

shasum -a 256 -c filemaid-darwin-arm64.sha256

Move the binary to a directory on your PATH, for example ~/.local/bin/filemaid, and make it executable:
```
chmod +x ~/.local/bin/filemaid
```

After installing, run setup:

filemaid setup

setup checks for Ollama, detects your Mac's RAM, recommends a model, and runs a short configuration interview:

24 GB+ RAM → filemaid-gemma4-26b
16 GB+ RAM → filemaid-gemma4-12b
less RAM → filemaid-metadata

Press Enter to accept each recommendation or default, or type a custom value when prompted.

To skip the model prompt, pass --model:

filemaid setup --model filemaid-gemma4-12b

To skip the configuration interview and use the shipped defaults, pass --no-interactive:

filemaid setup --no-interactive

To install the background launchd agents (disabled by default):

filemaid setup --agents

To skip the scheduled scan agent when using --agents:

filemaid setup --agents --no-scan

To see the macOS Shortcuts folder-automation steps:

filemaid setup --shortcuts

Or install the latest release directly with go install (requires Go):

go install github.com/logicminds/filemaid/cmd/filemaid@latest

Make sure $(go env GOPATH)/bin is on your PATH to run the installed binary as filemaid.

Or build locally from source (requires Go):

git clone https://github.com/logicminds/filemaid.git ~/Projects/filemaid
cd ~/Projects/filemaid
go build -o bin/filemaid ./cmd/filemaid
./bin/filemaid setup

After setup you will have:

~/.local/bin/filemaid — installed command-line binary
~/.config/filemaid/config.json — user configuration
~/.local/share/filemaid/ — logs and SQLite database
~/.filemaid/review/ — quarantine folder
~/Documents/Filemaid/ — Finder hub with Smart Folders, aliases, and sidebar pin (when smart_folders is enabled)
~/Library/LaunchAgents/biz.logicminds.filemaid.*.plist — background agents (only when setup --agents is used)
Custom Ollama models (filemaid-gemma4-26b, filemaid-gemma4-12b, filemaid-metadata) — created automatically if Ollama is installed

For instant per-file processing without background agents, use the Shortcuts folder automation. Run filemaid setup --shortcuts to see the steps.

Usage

# Process files manually (human-readable list output is default)
# Without --move, files are classified/renamed in place; pass --move to relocate them.
filemaid process ~/Desktop/Screenshot*.png ~/Downloads/receipt.pdf

# Process files and move them to classified archive folders
filemaid process --move ~/Desktop/Screenshot*.png ~/Downloads/receipt.pdf

# Process files as a table
filemaid process --format table ~/Desktop/Screenshot*.png

# Get process results as JSON
filemaid process --json ~/Desktop/Screenshot*.png

# Suppress JSON log lines on stderr
filemaid process --quiet ~/Desktop/Screenshot*.png ~/Downloads/receipt.pdf

# Scan watch directories
filemaid scan

# Scan and move files to their archive categories
filemaid scan --move

# Scan a single directory
filemaid scan --dir ~/Downloads

# Process with smart rename enabled for this run (uses rename_level from config)
filemaid process --rename ~/Desktop/*.pdf

# Rename with an inline quality threshold (1=most aggressive, 5=most conservative)
filemaid process --rename=3 ~/Desktop/*.pdf

# Force processing even if a file looks like a duplicate or similar to history
filemaid process --force ~/Desktop/*.png

# Preview renames and destinations without moving files
filemaid process --dry-run ~/Desktop/*.png

# Scan with move preview
filemaid scan --move --dry-run
# Scan directories up to depth 2 (read-only directory recommendations)
filemaid scan --depth=2

# Process a directory tree with directory context
filemaid process --depth=1 ~/Downloads/project-folder

# Retry transient LLM failures from the review queue
filemaid review --retry

# Show the last run summary
filemaid history --last

# Undo the most recent run
filemaid undo

# Run cleaners in dry-run mode
filemaid cleanup --dry-run

# Run cleaners for real (table output is default)
filemaid cleanup

# Get cleaner results as JSON
filemaid cleanup --format json

# Run cleaners and see estimated space that would be freed
filemaid cleanup --dry-run --format table

# View the review queue
filemaid review

# Open the review queue in Finder
filemaid review --open

# Approve or reject a review item by its relative path
filemaid review --approve "Screenshots/old-screenshot.png"
filemaid review --reject "Documents/unwanted-receipt.pdf"

# Show recent processing history
filemaid history

# Show the last run with a count summary
filemaid history --last

# Regenerate the Filemaid hub (Smart Folders, aliases, sidebar pin)
filemaid smart-folders

# Tail logs
filemaid logs --tail 50

# Show resolved configuration
filemaid config

# Install background launchd agents
filemaid setup --agents

# Output macOS Shortcuts folder-automation steps
filemaid setup --shortcuts

If you are running from a local clone, use ./bin/filemaid instead of filemaid.

Shortcuts Setup

Shortcuts folder automations are the recommended trigger: they run instantly when a file lands in a folder, do not require Full Disk Access, and avoid leaving a background agent running.

Run the following command to print the exact steps for your binary path:

filemaid setup --shortcuts

For instant per-file processing, add a Shortcuts folder automation:

Open Shortcuts → Automations → Personal Automation → + → Folder.
Select Desktop, choose Run immediately.

Add Run Shell Script:

Shell: /bin/zsh
Pass input: As arguments

Command:

export PATH="$HOME/.local/bin:$(go env GOPATH)/bin:/usr/local/bin:/opt/homebrew/bin:$PATH"
filemaid process "$@"

If you are running from a local clone, use the binary path instead:

"$HOME/Projects/filemaid/bin/filemaid" process "$@"

Repeat for Downloads.

Shortcuts runs in your user session and does not require Full Disk Access.

How Classification Works

When a file is processed, filemaid sends its name, extension, size, modification time, and (for supported images and text files) a content snippet to the local Ollama model. With --depth, files inside descended directories also include ancestor path, depth, and detected project-marker context. The model returns one of the categories listed in config.json, along with a concise subcategory for images, suggested Finder tags, and an action (move, delete, or review).

For photos, the subcategory describes the main subject or scene (for example cat, dog, baby, wedding, or car). For screenshots, it describes the app or context (for example Safari, Terminal, Slack, browser, or lock-screen). The subcategory is added as a Finder tag when tags is enabled.

The LLM's reason for the classification is written to the file's Finder comment when comments is enabled, so you can see why a file was organized the way it was from the Finder Get Info panel.

Set subcategorize_images to false to disable the extra image detail and only receive the top-level category.

If you pass --move, the file is relocated: known categories go to their folder under categories, and uncertain/review items go to ~/.filemaid/review/.
Without --move, files are classified in place. Known categories and uncertain items both stay where they are, with Finder tags and comments applied (and a rename if rename is enabled and the LLM suggests one).
If the response cannot be parsed or the model is unreachable, filemaid falls back to a review-style classification.
The model is only allowed to choose from categories you define. Adding a new category in config.json is enough for the model to classify files into it.

To add a new category, add it to the categories map in ~/.config/filemaid/config.json:

{
  "categories": {
    "Presentations": "~/Documents/Archive/Presentations"
  }
}

The next time filemaid runs, the model may classify matching files into ~/Documents/Archive/Presentations.

Filemaid Hub and Smart Folders

When smart_folders is enabled (the default), filemaid builds a Finder hub at ~/Documents/Filemaid every time files are processed or scanned. The hub contains:

Smart Folders — one .savedSearch per configured category and per unique Finder tag filemaid has ever applied. These are live Spotlight searches scoped to files tagged filemaid.
Archive alias — a Finder alias pointing to the root of your categorized archive.
Review alias — a Finder alias pointing to ~/.filemaid/review.
Finder sidebar pin — the hub folder is added to Finder's Favorites for quick access.

The hub is recreated automatically after each process and scan run. To regenerate it manually:

filemaid smart-folders

Smart Folders rely on Spotlight indexing. If they appear empty, see the FAQ for troubleshooting steps.

To disable the hub, set smart_folders to false in ~/.config/filemaid/config.json. To change the hub location, edit smart_folders_dir.

Configuration

filemaid setup runs an interactive interview that asks for watch directories, archive location, Finder tags, dev cleaners, review-queue retention, and safe-delete patterns. Press Enter at each prompt to accept the default. To skip the interview and use the shipped defaults, run:

filemaid setup --no-interactive

The generated configuration is written to ~/.config/filemaid/config.json and can be edited at any time.

Full config example

{
  "ollama_url": "http://localhost:11434",
  "model": "filemaid-gemma4-26b",
  "image_model": "filemaid-gemma4-26b",
  "text_model": "filemaid-metadata",
  "watch_dirs": ["~/Desktop", "~/Downloads"],
  "allowed_dirs": ["~/Desktop", "~/Downloads", "~/Documents/Archive", "~/.filemaid/review"],
  "allowed_cleaners": ["docker", "npm", "cargo", "pip", "brew", "xcode", "review"],
  "project_markers": [".git", "node_modules", ".venv", "vendor", ".terraform", "build"],
  "review_dir": "~/.filemaid/review",
  "log_path": "~/.local/share/filemaid/filemaid.log",
  "db_path": "~/.local/share/filemaid/filemaid.db",
  "tags": true,
  "comments": true,
  "subcategorize_images": true,
  "smart_folders": true,
  "smart_folders_dir": "~/Documents/Filemaid",
  "min_age_hours": 0,
  "request_timeout": "5m",
  "llm_retry_attempts": 2,
  "llm_retry_base_delay": "2s",
  "categories": {
    "Screenshots": "~/Documents/Archive/Screenshots",
    "Documents": "~/Documents/Archive/Documents",
    "Receipts": "~/Documents/Archive/Receipts",
    "Images": "~/Documents/Archive/Images",
    "Installers": "~/Documents/Archive/Installers",
    "Code": "~/Documents/Archive/Code",
    "Archives": "~/Documents/Archive/Archives",
    "Media": "~/Documents/Archive/Media",
    "Unknown": "~/.filemaid/review"
  },
  "safe_delete_patterns": [],
  "age_rules": [
    {"pattern": "~/Downloads/*.dmg", "days": 30, "action": "review"}
  ],
  "dev_cleanup": {
    "docker": {"enabled": true, "mode": "safe"},
    "npm":    {"enabled": true, "mode": "safe"},
    "cargo":  {"enabled": true, "mode": "safe"},
    "pip":    {"enabled": true, "mode": "safe"},
    "brew":   {"enabled": true, "mode": "safe"},
    "xcode":  {"enabled": true, "mode": "safe"}
  },
  "review_cleanup": {
    "enabled": true,
    "mode": "safe",
    "max_age_days": 30
  },
  "_rename_note": "Set rename=true to let the LLM suggest better filenames. rename_level (0-5) is the minimum quality threshold a suggested new name must meet before filemaid applies it. 1 = most aggressive (even weak suggestions), 2 = aggressive, 3 = moderate, 4 = conservative, 5 = most conservative (only excellent suggestions). 0 accepts any suggestion. Invalid characters are removed, extensions are preserved, and collisions get a counter suffix.",
  "rename": false,
  "rename_level": 2,
  "rename_max_length": 120,
  "rename_min_length": 20,
  "rename_invalid_chars": "<>:\"/\\\\|?*",
  "rename_image_similarity_threshold": 0.95,
  "rename_av_similarity_threshold": 0.90,
  "_ffmpeg_note": "rename_use_ffmpeg enables audio/video fingerprinting via ffmpeg. This improves duplicate/near-duplicate detection but can be slow for large media libraries.",
  "rename_use_ffmpeg": false,
  "move_files": false,
  "external_tools": {
    "ffmpeg": "ffmpeg"
  },
  "process_workers": 1,
  "max_image_dimension": 1024
}

Config keys

Key	Purpose
`ollama_url`	URL of the local Ollama server.
`model`	Default Ollama model tag. Used for all files unless `image_model` or `text_model` is set.
`image_model`	Model used for image files. Falls back to `model` when empty.
`text_model`	Model used for non-image files. Falls back to `model` when empty.
`watch_dirs`	Directories scanned by `filemaid scan`.
`allowed_dirs`	Files outside these directories are ignored; also gates destination paths.
`allowed_cleaners`	Which dev cleaners may run. Must include `"review"` to enable review-queue cleanup.
`review_dir`	Quarantine folder for uncertain files.
`log_path`	Path to the main application log.
`db_path`	Path to the SQLite history database.
`tags`	Whether to apply Finder tags to organized files.
`comments`	Whether to write the classification reason as a Finder comment.
`subcategorize_images`	When `true`, images and screenshots receive a subject/app subcategory that is also added as a Finder tag.
`smart_folders`	When `true`, build the Filemaid hub with Smart Folders, aliases, and sidebar pin.
`smart_folders_dir`	Directory for the Filemaid hub.
`min_age_hours`	Minimum file age before processing (0 = process immediately).
`request_timeout`	Per-request timeout for Ollama calls (e.g. `120s`, `2m`).
`llm_retry_attempts`	How many times to retry transient Ollama failures before giving up.
`llm_retry_base_delay`	Base delay between retries; backoff doubles each attempt (e.g. `2s`).
`project_markers`	Directory names that identify a project root when scanning with `--depth`. User values are added to the built-in defaults.
`categories`	Destination folders for each classification. The model may only return categories defined here.
`move_files`	When `true`, `process`/`scan` move files to their classified category. Defaults to `false`; use `--move` to enable per-run.
`safe_delete_patterns`	Glob patterns for files allowed to be deleted without review.
`age_rules`	Patterns + age that force a specific action, e.g. old `.dmg` installers become `review`.
`dev_cleanup`	Per-cleaner enable/disable and mode (`safe` is the only mode currently).
`review_cleanup`	Enable and set retention for the review-queue cleaner. Set `max_age_days` to `0` to disable.
`rename`	When `true`, the LLM may suggest better filenames.
`rename_level`	Minimum quality threshold (0-5) a suggested new name must meet. 1 = most aggressive, 5 = most conservative.
`rename_max_length`	Maximum length for a renamed file.
`rename_min_length`	Minimum length below which a rename is not applied.
`rename_invalid_chars`	Characters stripped from suggested names.
`rename_image_similarity_threshold`	Perceptual-hash similarity (0-1) above which images are sent to review.
`rename_av_similarity_threshold`	Audio/video similarity (0-1) above which files are sent to review.
`rename_use_ffmpeg`	Enable ffmpeg-based audio/video fingerprinting. Slow for large libraries; requires ffmpeg on PATH.
`max_dir_sample_entries`	Maximum child entries sampled when gathering directory metadata for `--depth` classification.
`max_dir_sample_bytes`	Maximum bytes read from sampled directory children for `--depth` classification.
`external_tools`	Paths to optional tools (`ffmpeg`).
`process_workers`	Concurrency for classification/hashing/apply (minimum 1).
`max_image_dimension`	Largest dimension for image payloads sent to the vision model (minimum 64).

Changes take effect the next time filemaid process, filemaid scan, or filemaid cleanup runs.

Rename

When rename is enabled, filemaid asks the LLM to suggest a better filename and rate the quality of that suggestion from 1 (weak) to 5 (excellent). rename_level is the minimum quality threshold the suggestion must meet before it is applied; 1 is the most aggressive (renames even on weak suggestions) and 5 is the most conservative (only excellent suggestions). If the suggested name's rating is at least rename_level, the file is renamed while preserving its extension.

Enable renaming for a single run with --rename. With no value it uses rename_level from config; pass a number inline to override the threshold for that run:

# Use config rename_level
filemaid process --rename ~/Desktop/*.pdf

# Override threshold for this run only
filemaid process --rename=3 ~/Desktop/*.pdf

Suggested names are sanitized: characters matching rename_invalid_chars are stripped, the length is clamped between rename_min_length and rename_max_length, and collisions are resolved with a counter suffix (e.g., document-2.pdf).

Duplicate or near-duplicate files are routed to the review queue instead of being renamed:

Exact SHA-256 duplicates are always sent to review.
Images are compared by perceptual hash; similarity above rename_image_similarity_threshold goes to review.
Audio/video files are compared by ffmpeg-extracted signatures when rename_use_ffmpeg is true and ffmpeg is available; otherwise a structural fallback is used. Similarity above rename_av_similarity_threshold goes to review.

Performance note: enabling rename_use_ffmpeg can be slow for large media libraries because each audio/video file is decoded and fingerprinted. Only enable it if you need robust duplicate detection for media.

Use --dry-run to preview suggested names without moving or renaming files:

filemaid process --rename --dry-run ~/Desktop/*.pdf

Rename level	Typical behavior
1	Apply almost any suggested name.
2	Apply reasonable names (default).
3	Apply only clearly better names.
4	Apply only high-confidence names.
5	Apply only very high-confidence names.

Custom Ollama Models

filemaid ships with three Ollama Modelfiles embedded in the binary. They bundle a system prompt, low temperature, and output constraints so the model returns the JSON shape filemaid expects.

Model	Base	Use case
`filemaid-gemma4-26b`	`gemma4:26b-a4b-it-qat`	Default high-quality vision model for images + text
`filemaid-gemma4-12b`	`gemma4:12b-it-qat`	Faster fallback with vision
`filemaid-metadata`	`qwen2.5:7b`	Non-vision/text-only model; classifies from filename and metadata only

filemaid setup creates all three models automatically if Ollama is installed. Switch models by editing ~/.config/filemaid/config.json:

{
  "model": "filemaid-gemma4-12b"
}

Use filemaid-metadata when running on a machine without a vision-capable model or when you only want filename/text classification.

Architecture

filemaid process <paths>
  -> internal/config.Load()    load config + defaults
  -> internal/state.Open()     open SQLite history
  -> internal/cli process
       -> internal/llm.Classify()  → Decision
       -> internal/actions.Apply()
       -> internal/state.Record()
       -> internal/hub.Build()     → Smart Folders, aliases, sidebar pin

With --depth, directories are summarized and classified read-only via internal/directory.Gather and internal/llm.ClassifyDirectory → DirectoryDecision. filemaid undo rolls back the most recent run using internal/state history records.

internal/cli/ — Cobra root command and subcommands (process, scan, cleanup, review, history, undo, logs, config, setup, smart-folders, uninstall).
internal/llm/ — Ollama classifier and Decision value object.
internal/actions/ — Applies decisions: whitelist, duplicates, rename, moves, tags, trash, review.
internal/state/ — SQLite history, duplicate detection, and cached decisions.
internal/config/ — Config loading with defaults and ~ expansion.
internal/smartfolder/ — macOS .savedSearch (Smart Folder) generation.
internal/directory/ — Bounded directory metadata gathering and project-marker detection for --depth.
internal/fingerprint/ — Perceptual-image and audio/video fingerprinting for duplicate detection.
internal/hub/ — Builds the Filemaid hub: Smart Folders, archive/review aliases, and Finder sidebar pin.
internal/cleaners/ — Plugin registry for dev-artifact cleanup.
internal/setup/ — Installation and uninstallation of binary, config, and launchd agents.
cmd/filemaid/main.go — CLI entry point.

Recent performance improvements cache decisions by file hash and skip LLM classification for duplicates that cannot be safely deleted, so already-seen files are applied instantly.

Scheduling

Background agents are disabled by default. To install them, run setup with --agents:

filemaid setup --agents

Agent	Schedule	Logs
`biz.logicminds.filemaid.scan`	Every 15 minutes (optional)	`~/.local/share/filemaid/scan.log`
`biz.logicminds.filemaid.cleanup`	06:00, 12:00, 18:00, 23:00	`~/.local/share/filemaid/cleanup.log`

Use --no-scan with --agents to install only the cleanup agent. If you do not install agents, run filemaid scan and filemaid cleanup manually, or use the Shortcuts folder automations above for instant per-file processing.

Uninstall

filemaid uninstall

This removes the LaunchAgents and the ~/.local/bin/filemaid binary. It does not remove your config, logs, database, review queue, or the Filemaid hub in ~/Documents/Filemaid.

Troubleshooting

Missing requirements

filemaid setup checks for Ollama before installing. If Ollama is missing, it prints:

Ollama is not installed or not on PATH.
Install Ollama with Homebrew:
  brew install ollama
Or download it from https://ollama.com/

If Ollama is installed but not running, start it with ollama serve and run setup again.

Install Homebrew

If you do not have Homebrew yet:

/bin/bash -c "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)"

Then install the required tools:

brew install go ollama

Acceptance Test Checklist

Before releasing a rename-related change, run through the following:

Scan agent cannot read Desktop/Downloads

The background scan agent may need Full Disk Access for the filemaid binary:

System Settings → Privacy & Security → Full Disk Access
Click +, press Cmd+Shift+G, and enter the path to the binary (e.g. ~/.local/bin/filemaid).

Or skip this entirely by using the Shortcuts folder automations above.

Ollama not responding

Ensure Ollama is running:

ollama list

If a custom model is missing, recreate it from the Modelfile:

cd ~/Projects/filemaid
ollama create -f modelfiles/Modelfile.filemaid-gemma4-26b filemaid-gemma4-26b

To use the base model directly instead, pull it and update model in ~/.config/filemaid/config.json:

ollama pull gemma4:26b-a4b-it-qat

If Ollama is unreachable, files are sent to review instead of erroring.

Development

Run the same checks the CI runs:

make build   # go build -o bin/filemaid ./cmd/filemaid
make test    # go test ./...
make fmt     # go fmt ./...
make lint    # go vet ./...
make coverage # go test -coverprofile=coverage.out ./...

CI enforces:

go test ./... passes on Go 1.25 and 1.26.4.
Overall test coverage stays above 80%.
go vet ./... and gofmt -l . are clean.
JSON configs and generated launchd plists are valid.

The default config.json and all Ollama Modelfiles are embedded into the binary with //go:embed, so go install github.com/logicminds/filemaid/cmd/filemaid@latest produces a fully portable installer: filemaid setup writes the config and models at runtime.

License

MIT

Name		Name	Last commit message	Last commit date
Latest commit History 124 Commits
.agents/skills/beads		.agents/skills/beads
.beads		.beads
.claude		.claude
.codex		.codex
.github/workflows		.github/workflows
cmd/filemaid		cmd/filemaid
docs		docs
internal		internal
public/images		public/images
testdata/acceptance		testdata/acceptance
todos		todos
.gitignore		.gitignore
AGENTS.md		AGENTS.md
CHANGELOG.md		CHANGELOG.md
CLAUDE.md		CLAUDE.md
DEVELOPMENT.md		DEVELOPMENT.md
FAQ.md		FAQ.md
Makefile		Makefile
PRD.md		PRD.md
README.md		README.md
architecture.md		architecture.md
compound-engineering.local.md		compound-engineering.local.md
config.json		config.json
go.mod		go.mod
go.sum		go.sum
install.sh		install.sh

Navigation Menu

Repository files navigation

filemaid

Features

Requirements

Quick Start

Usage

Shortcuts Setup

How Classification Works

Filemaid Hub and Smart Folders

Configuration

Full config example

Config keys

Rename

Custom Ollama Models

Architecture

Scheduling

Uninstall

Troubleshooting

Missing requirements

Install Homebrew

Acceptance Test Checklist

Scan agent cannot read Desktop/Downloads

Ollama not responding

Development

License

About

Releases

Packages

Contributors

Languages

Folders and files

Latest commit

History

Repository files navigation

filemaid

Features

Requirements

Quick Start

Usage

Shortcuts Setup

How Classification Works

Filemaid Hub and Smart Folders

Configuration

Full config example

Config keys

Rename

Custom Ollama Models

Architecture

Scheduling

Uninstall

Troubleshooting

Missing requirements

Install Homebrew

Acceptance Test Checklist

Scan agent cannot read Desktop/Downloads

Ollama not responding

Development

License

About

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages