dauber 0.1.8

Command-line interface for the Canvas LMS API

dauber

A command-line interface for the Canvas LMS API. Manage courses, assignments, grading, and content from the terminal.

dauber also serves as a backend for Claude Code skill commands, enabling AI-assisted grading workflows via --format json.

Installation

Requirements: Python 3.11+ and uv.

git clone https://github.com/francojc/dauber.git
cd dauber

Install dauber to ~/.local/bin/ so it's available system-wide:

uv tool install -e .

The -e (editable) flag means changes to the source take effect immediately without reinstalling. To update after a git pull:

uv tool install -e . --force

If you prefer to run from the project directory without a global install, use uv run dauber instead (after uv sync).

Set two environment variables for Canvas API access:

export CANVAS_API_KEY="your-canvas-api-token"
export CANVAS_BASE_URL="https://your-institution.instructure.com"

The /api/v1 path is appended automatically. Generate an API token from your Canvas account settings under "Approved Integrations."

Verify the install and connection:

dauber --version
dauber --test

Quick start

# List your courses
dauber courses list

# List assignments for a course (accepts codes or numeric IDs)
dauber assignments list --course IS505

# View submissions for an assignment
dauber grading submissions --course IS505 42

# Set up course-level config
dauber config init

All commands accept --format (-f) with four output modes:

Mode	Use case
table	Aligned columns (default)
json	Machine-readable, for piping
plain	Simple key-value pairs
csv	Header + rows, pipe to file/tools

Configuration

dauber uses two TOML config files. Local values override global.

Global (instructor defaults shared across courses):

$XDG_CONFIG_HOME/dauber/config.toml    # default: ~/.config/dauber/config.toml

Set up interactively or write defaults without prompting:

dauber config global              # interactive prompts
dauber config global --defaults   # write starter config

Local (per-course settings in the repo root):

./dauber/config.toml

dauber config init                # interactive prompts, pre-fills from global

View merged config:

dauber config show                # shows each value with [global] or [local] source

Commands

courses

dauber courses list [--concluded]
dauber courses show [--course COURSE]
dauber courses enrollments [--course COURSE]

List, inspect, and view enrollments for your courses. The --course option accepts course codes (e.g., IS505) or numeric Canvas IDs. When omitted, falls back to canvas_course_id in your config file.

assignments

dauber assignments list [--course COURSE]
dauber assignments show [--course COURSE] <assignment-id>
dauber assignments create [--course COURSE] <name> [--points N] [--due ISO] [--publish]
dauber assignments update [--course COURSE] <assignment-id> [--name ...] [--points N]

Create, update, list, and inspect assignments. For interactive guided creation with defaults, validation, and rubric handoff, see /assignments:create.

rubrics

dauber rubrics list [--course COURSE]
dauber rubrics show [--course COURSE] <rubric-id>
dauber rubrics create [--course COURSE] --file <path>
dauber rubrics import [--course COURSE] --csv <path>
dauber rubrics attach [--course COURSE] <rubric-id> <assignment-id> [--use-for-grading]

List, inspect, create, and attach rubrics. show looks up a rubric by its direct ID. create reads a JSON file with title and criteria fields. import reads a Canvas-format CSV file (the wide-format template exported from Canvas). attach associates an existing rubric with an assignment; pass --use-for-grading to map rubric scores to the assignment grade. For guided format selection (CSV/JSON/interactive) and automatic create → attach sequencing, see /rubrics:create.

Example JSON for create:

{
  "title": "Essay Rubric",
  "criteria": [
    {
      "description": "Thesis",
      "points": 25,
      "ratings": [
        {"description": "Excellent", "points": 25},
        {"description": "Needs work", "points": 10},
        {"description": "Missing", "points": 0}
      ]
    }
  ]
}

grading

dauber grading submissions [--course COURSE] <assignment-id> [--anonymize]
dauber grading show [--course COURSE] <assignment-id> <user-id> [--anonymize]
dauber grading submit [--course COURSE] <assignment-id> <user-id> <grade> [--comment ...]
dauber grading submit-rubric [--course COURSE] <assignment-id> <user-id> <file> [--comment ...]

View submissions, inspect individual student work, and post grades. submit-rubric reads rubric criterion scores from a JSON file. For distribution stats and missing-submission flags across a cohort, see /grading:overview.

assess

dauber assess setup [--course COURSE] <assignment-id> [--exclude-graded] [--anonymize]
dauber assess load <file>
dauber assess update <file> <user-id> [--rubric-json ...] [--approved]
dauber assess submit <file> [--course COURSE] <assignment-id> [--confirm]

Full rubric-based assessment workflow: fetch assignment data into a local JSON file, update individual scores, and submit approved grades back to Canvas. Submit runs in dry-run mode by default; pass --confirm to post grades. These commands are building blocks; for the full AI grading pipeline use /assess:setup → /assess:ai-pass → /assess:refine → /assess:submit.

modules

dauber modules list [--course COURSE] [--items] [--search ...]
dauber modules show [--course COURSE] <module-id>
dauber modules create [--course COURSE] <name> [--position N] [--publish]
dauber modules update [--course COURSE] <module-id> [--name ...] [--publish/--unpublish]
dauber modules delete [--course COURSE] <module-id>

pages

dauber pages list [--course COURSE] [--search ...] [--sort title|created_at|updated_at]
dauber pages show [--course COURSE] <page-url>
dauber pages create [--course COURSE] <title> [--body ...] [--publish]
dauber pages update [--course COURSE] <page-url> [--title ...] [--body ...]
dauber pages delete [--course COURSE] <page-url>

Pages are identified by their URL slug (e.g., syllabus-spring-2026). To publish a local Markdown file with automatic HTML conversion and module placement, see /content:publish.

discussions

dauber discussions list [--course COURSE] [--announcements]
dauber discussions show [--course COURSE] <topic-id>
dauber discussions create [--course COURSE] <title> [--message ...] [--announcement] [--publish]
dauber discussions update [--course COURSE] <topic-id> [--title ...] [--message ...]

Pass --announcements to list only announcements. Use --announcement when creating to post an announcement rather than a discussion topic. For AI-drafted announcement text with tone matching and an approval loop, see /discuss:announce.

config

dauber config init [--base .]
dauber config global [--defaults]
dauber config show

See Configuration above for details.

commands

dauber commands install [--overwrite] [--local]
dauber commands install --pi [--overwrite]
dauber commands install --pi --global [--overwrite]

Installs bundled skill commands for either Claude Code or Pi.

Invocation	Target
dauber commands install	~/.claude/commands/ (Claude, global)
dauber commands install --local	./.claude/commands/ (Claude, project)
dauber commands install --pi	./.pi/skills/ (Pi, project)
dauber commands install --pi --global	~/.pi/agent/skills/ (Pi, global)

Pass --overwrite to replace existing files. See Skill commands below.

Global options

dauber --version             # show version
dauber --test                # test Canvas API connection
dauber --config              # show API URL and token (masked)
dauber --format json <cmd>   # JSON output for any command
dauber --install-completion  # install shell tab-completion

Anonymizing student data

Commands that return student information support --anonymize to strip personally identifiable information. When enabled, user_name and user_email are replaced with empty strings. The numeric user_id is retained for grade submission round-tripping.

Affected commands: assess setup, grading submissions, grading show.

dauber assess setup --course IS505 42 --anonymize --format json
dauber grading submissions --course IS505 42 --anonymize

This is opt-in. Without --anonymize, output includes full names and emails as returned by the Canvas API.

Skill commands

Some dauber commands are self-contained: you know the inputs, you run the command, you get the result. courses list, assignments show, modules create, and most read/write operations fall into this category — no skill needed.

Other commands are most useful as building blocks. The skill commands below orchestrate sequences of dauber calls, inject AI reasoning (drafting, scoring, normalizing), and manage state across steps. --format json is what makes this work: skills parse structured output from one command and feed it into the next.

Install all skills with:

# Claude Code (global)
dauber commands install

# Pi Agent Skills (project-local)
dauber commands install --pi

For Claude Code, this copies Markdown command files into ~/.claude/commands/. Once installed, invoke them from Claude Code (e.g., /assess:setup, /course:overview).

For Pi, this copies SKILL.md files into ./.pi/skills/. Once installed, the skills are available to the Pi coding agent from that project tree. Use --pi --global to install to ~/.pi/agent/skills/ for user-wide availability.

Skill	CLI commands used	What it does
/assess:setup → /assess:ai-pass → /assess:refine → /assess:submit	assess setup/load/update/submit	Full AI grading pipeline: fetch submissions → AI-evaluate against rubric → normalize scores → post to Canvas
/assignments:create	assignments create	Interactive parameter collection with defaults, validation, and rubric handoff
/rubrics:create	rubrics create/import/attach	Guides format choice (CSV/JSON/interactive), sequences create → attach
/discuss:announce	discussions create --announcement	AI-drafted announcement text with tone/formality matching and approval loop
/content:publish	pages create	Markdown → HTML conversion, existing-page detection, module placement
/grading:overview	grading submissions	Distribution stats (mean, median, quartiles) and missing-submission flags across the cohort
/course:overview	courses show, assignments list, modules list	Unified dashboard: enrollment, upcoming deadlines, content counts
/course:setup	--test, courses show/enrollments, config init	First-time course initialization with guided config creation

Writing custom skills

Any script or Claude Code skill can call dauber. A minimal example that lists ungraded submissions:

dauber grading submissions --course IS505 42 --format json \
  | jq '[.[] | select(.grade == null)]'

Skills that need course context (feedback language, formality, etc.) read from ./dauber/config.toml. Run dauber config init to set it up.

Development

uv sync                         # install dependencies
uv run dauber --help             # verify install
uv run pytest tests/            # run tests
uv run ruff check src/ tests/   # lint
uv run ruff format src/ tests/  # format

Architecture

CLI (Typer) -> services (async) -> core (HTTP client, config, cache)

• Core -- CanvasClient (httpx async, pagination, 429 retry), Config (pydantic-settings), CourseCache (bidirectional code/ID mapping), config file helpers (TOML).
• Services -- Async functions per Canvas entity. Accept a CanvasClient, return dicts/lists, raise CanvasError on failure.
• CLI -- Typer commands that bridge async via decorator, format output through format_output(), and exit with appropriate codes.

Tests

Tests are organized in two layers:

• tests/services/ -- mock at the CanvasClient transport level
• tests/cli/ -- mock at the service function level, use typer.testing.CliRunner

License

See LICENSE.