---
name: basecamp
description: |
  Interact with Basecamp via the Basecamp CLI. Full API coverage: projects, todos, cards,
  messages, files, schedule, check-ins, timeline, recordings, templates, webhooks,
  subscriptions, lineup, chat, pings, gauges, assignments, notifications, bookmarks,
  bubble-up, drafts, notes, calendars, and accounts.
  Use for ANY Basecamp question or action.
triggers:
  # Direct invocations
  - basecamp
  - /basecamp
  # Resource actions
  - basecamp todos
  - basecamp project
  - basecamp cards
  - basecamp chat
  - basecamp campfire
  - basecamp messages
  - basecamp file
  - basecamp document
  - basecamp bookmarks
  - basecamp bubble-up
  - basecamp drafts
  - basecamp notes
  - basecamp calendars
  - basecamp schedule
  - basecamp checkin
  - basecamp check-in
  - basecamp timeline
  - basecamp template
  - basecamp webhook
  - basecamp gauge
  - basecamp assignment
  - basecamp notification
  - basecamp account
  # Common actions
  - link to basecamp
  - track in basecamp
  - post to basecamp
  - comment on basecamp
  - complete todo
  - mark done
  - create todo
  - move card
  - download file
  # Search and discovery
  - search basecamp
  - find in basecamp
  - look up basecamp
  - check basecamp
  - list basecamp
  - show basecamp
  - get from basecamp
  - fetch from basecamp
  # Questions
  - can I basecamp
  - how do I basecamp
  - what's in basecamp
  - what basecamp
  - does basecamp
  # My work
  - my todos
  - my tasks
  - my schedule
  - my basecamp
  - assigned to me
  - my assignments
  - my notifications
  - overdue todos
  - upcoming events
  - project gauge
  - project progress
  # URLs
  - 3.basecamp.com
  - basecampapi.com
  - https://3.basecamp.com/
invocable: true
argument-hint: "[action] [args...]"
---

# /basecamp - Basecamp Workflow Command

Full CLI coverage: 195 tracked in-scope endpoints across todos, cards, messages, files, schedule, check-ins, timeline, recordings, templates, webhooks, subscriptions, lineup, chat, pings, gauges, assignments, notifications, the account event feed, and accounts.

## Agent Invariants

**MUST follow these rules:**

1. **Choose the right output mode** — `--jq` when you need to filter/extract data; `--json` for full JSON; `--md` when presenting results to a human (see Output Modes below). **Never pipe to external `jq` — use `--jq` instead.**
2. **Parse URLs first** with `basecamp url parse "<url>"` to extract IDs
3. **Comments are flat** - reply to parent recording, not to comments
4. **Check context** via `.basecamp/config.json` before assuming project
5. **Content fields accept Markdown, and most accept @mentions** — the CLI converts these rich-text fields from Markdown to HTML: message bodies, document bodies, comment content, todo descriptions, card bodies, schedule entry descriptions, upload descriptions, check-in answers and notes. Two rich-text fields are sent as written, so give them HTML: todolist descriptions and gauge needle descriptions. Chat is different again: `chat post` sends plain text unless you pass `--content-type text/html` or the line carries a mention. Use Markdown formatting (lists, bold, links, code blocks, tables) for rich content. @mentions resolve in message bodies, comment content, card bodies, schedule descriptions and chat lines — not in todo descriptions, documents, uploads, check-ins or notes. Four mention syntaxes are available (prefer deterministic for agents):
   - **`[@Name](mention:SGID)`** — zero API calls, embeds SGID directly (preferred for agents)
   - **`[@Name](person:ID)`** — one API call, resolves person ID to SGID via pingable set
   - **`@sgid:VALUE`** — inline SGID embed for pipeline composability
   - **`@Name` / `@First.Last`** — fuzzy name resolution (may be ambiguous)

   Raw HTML is also accepted, but it is all-or-nothing per field: a tag the CLI detects as HTML (`<p>`, `<ul>`, `<strong>`, `<a>`, `<img>`, `<table>` and the other common formatting tags) outside a backtick code span or backtick fence (a `~~~` fence does not hide it) skips Markdown conversion for the whole field, so any Markdown alongside it — `![alt](/local/path)` included — is sent literally. The HTML itself goes through as written, except that an empty separator paragraph is inserted between directly adjacent `<p>` blocks so they render with spacing; a local path in a raw `<img src>` is still uploaded and replaced with an attachment in every converted field except notes, which take no attachments. Titles (a todo's content argument, card and message titles) are plain text and never converted.

   **Table boundary:** GFM tables round-trip: they render in message/comment
   bodies, display converts them back to pipe tables, and the TUI in-place
   editors open simple grids for editing. Only **complex** tables — merged
   cells (colspan/rowspan), captions, extra header rows, nested tables,
   attachments/images or block content inside cells, multi-paragraph or
   multi-line cells, or a table inside a blockquote or list — refuse to open,
   since a GFM pipe table can't represent those shapes (edit them on Basecamp
   web, or replace the
   whole field via `messages update` / `comments update` / `todos update
   --description`, which take fresh content and are unaffected). Complex
   tables still **display** best-effort, flattened to a plain grid.

   **Multiline / non-ASCII content:** do not rely on bash ANSI-C quoting (`$'...\n...'`) — it is a bash/zsh extension. Under a POSIX `/bin/sh` (dash, busybox-ash, common in sandboxes) the `$` is passed through literally and posts a stray leading `$`, and `\n` stays a literal backslash-n. Pipe the content via stdin instead, using `-` as the content argument:
   ```bash
   printf '%s\n' '海报 mockup 方向稿：' '' '<bc-attachment ...>' | basecamp comments create <recording_id> - --in <project> --json
   ```
   `-` means "read from stdin" on every content input: content-kind positionals
   (`comments create/update`, `messages create [body]`, `cards create [body]`,
   `todos create`, `docs documents create [content]`, `chat post/update`, `boost create`,
   `checkins answer create/update`, `notes set`) and content flags (`--data` on
   `api post/put`, `--body`, `--content`, `--description`, `--comment` on
   `todos sweep`, `--file` on `notes set`). Each command's `--agent` help lists
   its stdin inputs. Rules:
   - A pipe is **never consumed implicitly** — without `-` it is ignored (or, where
     content is required and missing, the error teaches `-`).
   - Only one input can read stdin per invocation.
   - A literal `-` anywhere else (a title, a name, a path) **errors when stdin is
     piped**. Escape a positional after the `--` separator
     (`basecamp projects create -- -`); a flag value has no in-line escape — run
     the command without piped stdin. `basecamp help` and shell completion are
     exempt: they write nothing to Basecamp, and completion legitimately
     receives `-` as the word being completed.
   - `-` with nothing piped (interactive TTY) errors immediately instead of
     hanging; use a pipe, a heredoc (`basecamp comments create <id> - <<'EOF'`),
     or `--edit` where offered.
   - Trailing newlines are trimmed from stdin content, so `printf 'x\n' | ... -`
     posts `x` (this keeps `boost create -` inside its 16-rune limit).
  - Universal `-` support (and the stray-`-` guard) shipped in **v0.10.0**. Older
    CLIs do not support it consistently: `comments create/update` read stdin,
    while unsupported inputs may treat `-` as literal content or fail. For
    example, `messages create "Title" -` posts a body of `-`, which Markdown
    renders as an empty bullet list. When the CLI version is unknown, check
    `basecamp --version` first, or pass the content portably as
    `"$(cat file.md)"` and verify the posted `content` when it matters.
6. **Project scope is mandatory for most commands** — via `--in <project>` or `.basecamp/config.json`. Cross-project exceptions: `basecamp reports assigned` for assigned work, `basecamp assignments` for structured assignment views, `basecamp reports overdue` for overdue todos, `basecamp reports schedule` for upcoming schedule across all projects, `basecamp recordings <type>` for browsing by type, `basecamp notifications` for notifications, `basecamp gauges list` for account-wide gauges, `basecamp events poll` and `basecamp inbox` for the account event feed, and the seven list commands covered in item 7.
7. **Account-wide listing.** `basecamp todos list --all-projects --json` lists across every project; the same flag does the same on `cards list`, `messages list`, `comments list`, `files list`, `forwards list`, and `checkins answers`. It overrides a configured project, and with no project in scope those commands already list account-wide rather than prompting. Flags that name something inside a single project are rejected there rather than silently ignored.
   Account-wide listings return **the first 100 items by default** — account-wide "all" is the whole account, not one project's worth. Use `--limit N` to raise the cap (it walks pages until N are collected) or `--all` for everything. `--page N` fetches exactly one page, but only on the paginated listings.
   The two overdue variants — `basecamp todos list --all-projects --overdue` and `basecamp cards list --all-projects --overdue` — come from unpaginated endpoints. They accept `--limit` and `--all` but **reject `--page`**, so do not generate `--page` against them.

### Output Modes

**Choosing a mode:**

| Goal | Flag | Format |
|------|------|--------|
| Filter/extract JSON data | `--jq '<expr>'` | Built-in jq filter (no external jq needed). Implies `--json`; filter runs on the envelope. |
| Filter in agent mode | `--agent --jq '<expr>'` | Filter runs on data-only payload (no envelope), matching `--agent` contract. |
| Full JSON output | `--json` | JSON envelope: `{ok, data, summary, breadcrumbs, meta}`; errors: `{ok:false, error, code, retryable, hint, meta}` |
| Show results to a user | `--md` / `-m` | GFM tables, task lists, structured Markdown |
| Automation / scripting | `--agent` | Success: raw JSON data (no envelope); errors: `{ok:false,...}` object; no interactive prompts |

Always pass `--json` or `--md` explicitly — auto-detection depends on config and may not produce the format you expect. Use `--md` when composing reports, summarizing data, or displaying results inline. `--agent` is for headless integration scripts.

**Avoiding interactive prompts.** The flags `--agent`/`--json`/`--quiet`/`--ids-only`/`--count` and the environment variable `BASECAMP_NONINTERACTIVE=1` suppress interactive selection prompts. `--md` does **not** — if a required target is ambiguous (e.g. a project with multiple todosets and no `--todoset`), and the CLI is attached to a terminal, it will show a blocking picker. When you need Markdown output *and* no prompts, either pass the flag that names whatever is ambiguous (`--todoset <id>` for the todoset case above, or `--in <project>` / `--list <id>` when the project or list is ambiguous) or set `BASECAMP_NONINTERACTIVE=1` in the environment. `BASECAMP_NONINTERACTIVE` disables all prompts (they become actionable errors instead) without changing the output format — an escape hatch for agents running under a PTY.

**Other modes:** `--quiet` (success: raw JSON, no envelope; errors: `{ok:false,...}`), `--ids-only`, `--count`, `--stats` (session statistics), `--styled` (force ANSI), `-v` / `-vv` (verbose/trace), `--jq '<expr>'` (built-in jq filter — see below).

### CLI Introspection

Navigate unfamiliar commands with `--agent --help` — returns structured JSON describing any command:

```bash
basecamp todos --agent --help
```

```json
{"command":"todos","path":"basecamp todos","short":"...","long":"...","usage":"...","notes":["..."],
 "subcommands":[{"name":"sweep","short":"...","path":"basecamp todos sweep"}],
 "flags":[{"name":"assignee","type":"string","default":"","usage":"..."}],
 "inherited_flags":[{"name":"json","shorthand":"j","type":"bool","default":"false","usage":"..."}]}
```

Walk the tree: start at `basecamp --agent --help` for top-level commands, then drill into any subcommand. Commands carry domain-specific agent hints (e.g., "`--assignee` filters the account-wide listing only; within a project, fetch all and filter client-side").

**Note:** a subcommand's `inherited_flags` is deliberately short — the CLI curates it down to `--account`, `--json`, `--md`, `--project`, `--quiet` (and drops `--project` where the command takes `<id|url>`). Every other global flag (`--jq`, `--agent`, `--styled`, `--verbose`, `--profile`, ...) is listed only at the root (`basecamp --agent --help`) but still applies on every subcommand (a command that cannot honor one refuses it with an explicit error — `version` rejects `--jq`); its absence from a subcommand's help does not mean it is unsupported.

### Pagination

```bash
basecamp <cmd> --limit 50   # Cap results (default varies by resource)
basecamp <cmd> --all        # Fetch all (may be slow for large datasets)
basecamp <cmd> --page 1     # First page only, no auto-pagination
```

`--all` and `--limit` are mutually exclusive. `--page` cannot combine with either.

### Smart Defaults

- `--assignee me` resolves to current user
- `--due tomorrow` / `--due +3` / `--due "next week"` — natural date parsing, **when setting a due date** (`todos create`, `todos update`, `cards create`, and so on)
- `--due` on a **listing** is a different flag and does not take dates: it accepts only `with`, `without`, or `overdue`, and only account-wide. `basecamp todos list --due tomorrow` is rejected. For date-based listing use `--overdue`, `--no-due-date`, or `basecamp assignments due <scope>`
- Project from `.basecamp/config.json` if `--in` not specified
- Multiple identities use named profiles: `basecamp profile create <name>`, then select one with global `--profile <name>` or `BASECAMP_PROFILE=<name>`.

## Quick Reference

> **Note:** Most queries require project scope (via `--in <project>` or `.basecamp/config.json`). Cross-project exceptions: `basecamp reports assigned`, `basecamp assignments`, `basecamp reports overdue`, `basecamp reports schedule`, `basecamp recordings <type>`, `basecamp notifications`, `basecamp gauges list`, `basecamp events poll`, `basecamp inbox`.
>
> Seven list commands also list account-wide: `basecamp todos list --all-projects --json`, and likewise `cards list`, `messages list`, `comments list`, `files list`, `forwards list`, and `checkins answers`.

| Task | Command |
|------|---------|
| List projects | `basecamp projects list --json` |
| My todos (in project) | `basecamp todos list --assignee me --in <project> --json` |
| My todos (cross-project) | `basecamp reports assigned --json` (defaults to "me") |
| My schedule (cross-project) | `basecamp reports schedule --json` (upcoming events across all projects) |
| All todos (cross-project) | `basecamp todos list --all-projects --json` (grouped by project) |
| Overdue todos (in project) | `basecamp todos list --overdue --in <project> --json` |
| Overdue todos (cross-project) | `basecamp todos list --all-projects --overdue --json` (flat, oldest first) or `basecamp reports overdue --json` (bucketed by lateness) |
| All cards (cross-project) | `basecamp cards list --all-projects --json` (grouped by project) |
| Someone's todos (cross-project) | `basecamp todos list --all-projects --assignee "Ann" --json` (server-side filter) |
| Two people's todos (cross-project) | `basecamp todos list --all-projects --assignee ann --assignee bob --json` (matches either) |
| Someone's cards (cross-project) | `basecamp cards list --all-projects --assignee "Ann" --json` |
| Todos with no due date set (cross-project) | `basecamp todos list --all-projects --due without --json` |
| My bookmarks | `basecamp bookmarks list --json` |
| Bookmark something | `basecamp bookmarks add <id-or-url> --json` |
| Is it bookmarked? | `basecamp bookmarks check <id-or-url> --json` (always exits 0) |
| Bubble a recording up | `basecamp bubble-up add <id-or-url> --json` |
| Schedule a bubble-up | `basecamp bubble-up add <id-or-url> --at tomorrow --json` |
| Pop a bubble-up | `basecamp bubble-up remove <id-or-url> --json` |
| My unpublished drafts | `basecamp drafts list --json` |
| Read my personal note | `basecamp notes show --json` |
| Replace my personal note | `basecamp notes set "<content>" --json` |
| Check-ins I owe answers to | `basecamp checkins reminders --json` |
| Add to Up Next | `basecamp assignments prioritize <id> --json` |
| Recolor a calendar | `basecamp calendars update <id-or-url> --color blue --json` |
| Todo outside any list | `basecamp todos create "<content>" --loose --in <project> --json` |
| Assign todo | `basecamp assign <id> [id...] --to <person> --in <project> --json` |
| Assign card | `basecamp assign <id> [id...] --card --to <person> --in <project> --json` |
| Assign card step | `basecamp assign <id> [id...] --step --to <person> --in <project> --json` |
| Create todo | `basecamp todos create "Task" --in <project> --list <list> --json` |
| Create todolist | `basecamp todolists create "Name" --in <project> --json` |
| Complete todo | `basecamp todos complete <id> --json` |
| List cards | `basecamp cards list --in <project> --json` |
| Create card | `basecamp cards create "Title" --in <project> --json` |
| Complete card | `basecamp cards done <id|url> --in <project> --json` |
| Move card | `basecamp cards move <id> --to <column> [--position N] --in <project> --json` |
| Move card to on-hold | `basecamp cards move <id> --on-hold --in <project> --json` |
| Move card to another project | `basecamp cards move <id> --to-wormhole <wormhole_id> --in <project> --json` (async teleport) |
| Post message | `basecamp messages create "Title" "Body" --in <project> --json` |
| Post with @mention | `basecamp messages create "Title" "Hey @First.Last, ..." --in <project> --json` |
| Post silently | `basecamp messages create "Title" "Body" --no-subscribe --in <project> --json` |
| Post to chat | `basecamp chat post "Message" --in <project> --json` |
| List pings | `basecamp notifications --json --jq '.data.reads[]? | select(.section == "pings")'` |
| Read ping thread | `basecamp api get "/buckets/<circle_id>/chats/<chat_id>/lines.json" --agent` |
| Post to ping thread | `basecamp api post "/buckets/<circle_id>/chats/<chat_id>/lines.json" --data '{"content":"<p>message</p>"}' --json` |
| Add comment | `basecamp comments create <recording_id> "Text" --in <project> --json` |
| Inspect comment / reply atoms | `basecamp comments show <url> --json` → `reply_target` + `mention` in `.data` |
| List attachments | `basecamp attachments list <id\|url> --json` |
| Download attachments | `basecamp attachments download <id> --out /tmp/` |
| Show + download | `basecamp todos show <id> --download-attachments --json` |
| Stream attachment to stdout | `basecamp attachments download <id> --file <name> --out -` |
| Change history for an item | `basecamp events <id\|url> --json` (when a card moved columns, when a todo was completed) |
| Account-wide activity feed (resumable) | `basecamp events poll --since now --json` (then resume with `--position`) |
| Items that addressed me (agents only) | `basecamp inbox --since now --json` |
| Search | `basecamp search "query" --json` |
| Parse URL | `basecamp url parse "<url>" --json` |
| Upload file | `basecamp files uploads create <file> [--vault <folder_id>] --in <project> --json` |
| Download file | `basecamp files download <id> --in <project>` |
| Stream file to stdout | `basecamp files download <id> --out - --in <project>` |
| Download storage URL | `basecamp files download "https://storage.3.basecamp.com/.../download/report.pdf"` |
| My assignments | `basecamp assignments --json` (priorities + non-priorities) |
| Overdue assignments | `basecamp assignments due overdue --json` |
| Completed assignments | `basecamp assignments completed --json` |
| Notifications | `basecamp notifications --json` |
| Mark notification read | `basecamp notifications read <id> --json` |
| All bubble-ups (BC5) | `basecamp notifications bubbleups --json` |
| Gauges (account-wide) | `basecamp gauges list --json` |
| Gauge needles | `basecamp gauges needles --in <project> --json` |
| Create needle | `basecamp gauges create --position 75 --color green --in <project> --json` |
| Account details | `basecamp accounts show --json` |

## URL Parsing

**Parse URLs before acting on them — unless you're handing the URL to a command
that accepts a URL directly** (`show`, `comments show`, `comments thread`,
`attachments list`/`attachments download`), which extract the IDs for you. Only `comments show` and
`comments thread` verify the URL's host and account before any fetch. For other
URL-accepting commands, only pass URLs from a trusted Basecamp host:
`basecamp url parse` extracts IDs but does **not** validate the URL's origin, so
parsing an attacker-controlled path yields trusted-looking IDs.

```bash
basecamp url parse "https://3.basecamp.com/2914079/buckets/41746046/messages/9478142982#__recording_9488783598" --json
```

Returns: `account_id`, `project_id`, `type`, `recording_id`, `comment_id` (from fragment).

**URL patterns:**
- `/buckets/27/messages/123` - Message 123 in project 27
- `/buckets/27/messages/123#__recording_456` - Comment 456 on message 123
- `/buckets/27/card_tables/cards/789` - Card 789
- `/buckets/27/card_tables/columns/456` - Column 456 (for creating cards)
- `/buckets/27/todos/101` - Todo 101
- `/buckets/27/uploads/202` - Upload/file 202
- `/buckets/27/documents/303` - Document 303
- `/buckets/27/schedule_entries/404` - Schedule entry 404

**Replying to comments:**
```bash
# Comments are flat - reply to the parent recording_id, not the comment_id
basecamp url parse "https://...messages/123#__recording_456" --json
# Returns recording_id: 123 (parent), comment_id: 456 (fragment) - comment on 123, not 456
basecamp comments create 123 "Reply" --in <project>

# Or get the whole reply-ready context deterministically in one call:
basecamp comments thread "https://...messages/123#__recording_456" --json
# .data.reply_target.recording_id  → where to post the reply
# .data.reply_target.account_id    → the account that reply belongs to (build a fully-qualified command)
# .data.focus.author.mention.syntax → paste-ready [@Name](mention:SGID)
# .data.comments                   → surrounding discussion (default window of 41)
# --all returns every fetched comment; --window N sets the window size
# When the account came from the URL (none configured), the reply breadcrumb carries --account
```

## Decision Trees

### Finding Content

```
Need to find something?
├── Know the type + project? → basecamp <type> list --in <project> --json
│   (some groups have default list behavior; use --agent --help if unsure)
├── My assigned work? → basecamp assignments --json (priorities + non-priorities)
│   Or: basecamp reports assigned --json (traditional view, defaults to "me")
├── My overdue assignments? → basecamp assignments due overdue --json
├── My notifications? → basecamp notifications --json
├── Upcoming schedule? → basecamp reports schedule --json (cross-project)
├── Overdue across projects? → basecamp reports overdue --json
├── Browse by type cross-project? → basecamp recordings <type> --json
│   (types: todos, messages, documents, comments, cards, uploads)
│   Note: Defaults to active status; use --status archived for archived items
│   ⚠ No assignee data — cannot filter by person; use reports assigned instead
├── Full-text search? → basecamp search "query" --json
├── Have a comment URL, or a notification link targeting a comment? → basecamp comments thread <url> --json
└── Have a URL? → basecamp url parse "<url>" --json
```

### Modifying Content

```
Want to change something?
├── Have URL? → basecamp url parse "<url>" → use extracted IDs
├── Have ID? → basecamp <resource> update <id> --field value
├── Change status? → basecamp recordings trash|archive|restore <id>
├── Complete todo? → basecamp todos complete <id>
├── Complete card? → basecamp cards done <id|url> --in <project>
└── Reply to a comment? → basecamp comments show <url> --jq '.data | {reply_target, mention}'
    (one call, cheap atoms — the mention is machine-only, so use --jq/--json, not plain show)
    or basecamp comments thread <url> when you need the surrounding discussion;
    then basecamp comments create <reply_target.recording_id> <text>
```

## Common Workflows

### Link Code to Basecamp Todo

```bash
# Get commit info and comment on todo (use printf %q for safe quoting)
COMMIT=$(git rev-parse --short HEAD)
MSG=$(git log -1 --format=%s)
basecamp comments create <todo_id> "Commit $COMMIT: $(printf '%s' "$MSG")" --in <project>

# Complete when done
basecamp todos complete <todo_id>
```

### Track PR in Basecamp

```bash
# Create todo for PR work
basecamp todos create "Review PR #42" --in <project> --assignee me --due tomorrow

# When merged
basecamp todos complete <todo_id>
basecamp chat post "Merged PR #42" --in <project>
```

### Bulk Process Overdue Todos

```bash
# Preview overdue todos
basecamp todos sweep --overdue --dry-run --in <project>

# Complete all with comment
basecamp todos sweep --overdue --complete --comment "Cleaning up" --in <project>
```

### Mentioning people (preferred — deterministic)

```bash
# 1. Look up the person
basecamp people pingable --jq '.data[] | select(.name == "Jane Smith")'
# => {"id": 42000, "attachable_sgid": "BAh7CEkiCG...", "name": "Jane Smith"}

# 2. Use SGID in Markdown mention syntax (zero API calls during post)
basecamp comments create 123 "Hey [@Jane Smith](mention:BAh7CEkiCG...), check this" --in <project>

# Or use person ID (one lookup during post)
basecamp comments create 123 "Hey [@Jane Smith](person:42000), check this" --in <project>
```

### Mentioning people (interactive — may be ambiguous)

```bash
# Fuzzy matching: use @First.Last to reduce ambiguity
basecamp comments create <id> "@Jane.Smith, please review this" --in <project>
basecamp messages create "Update" "cc @Jane, @Alex" --in <project>
basecamp chat post "@Jane, done!" --in <project>

# Ambiguous names return an error with suggestions
# Use @First.Last for disambiguation
```

### Move Card Through Workflow

```bash
# List columns to get IDs
basecamp cards columns --in <project> --json

# Complete a card (moves it to the Done column automatically)
basecamp cards done <card_id> --in <project>

# Move card to column
basecamp cards move <card_id> --to <column_id> --in <project>

# Move card to specific position in column (1-indexed)
basecamp cards move <card_id> --to <column_id> --position 1 --in <project>

# Move card to on-hold section of its current column
basecamp cards move <card_id> --on-hold --in <project>

# Move card to on-hold section of a specific column (numeric ID)
basecamp cards move <card_id> --to <column_id> --on-hold --in <project>

# Move card to on-hold section of a named column (requires --card-table)
basecamp cards move <card_id> --to "Column Name" --on-hold --card-table <table_id> --in <project>
```

### Download File from Basecamp

```bash
basecamp files download <upload_id> --in <project> --out ./downloads

# Download attachment from a storage URL (no --in needed)
basecamp files download "https://storage.3.basecamp.com/123/blobs/abc/download/report.pdf"

# Stream to stdout (for piping)
basecamp files download <upload_id> --out - --in <project>
```

### Working with Attachments (Multimodal Agent Workflow)

Messages, todos, cards, and documents may contain images and file attachments
(mockups, screenshots, annotated designs). Show commands surface these as
field-scoped collections — `content_attachments` and/or `description_attachments`
— keyed by which rich-text attribute contained them. The notice field hints at
the download command.

**Step 1: Fetch the recording and check for attachments**
```bash
basecamp todos show <id> --json
# Response includes description_attachments when attachments are present
# Messages/documents use content_attachments; cards may have both
# The notice field hints: "3 attachment(s) — download: basecamp attachments download <id>"
```

**Step 2 (one-shot): Download attachments with the show command**
```bash
# --download-attachments fetches + downloads in one shot
basecamp todos show <id> --download-attachments --json
# content_attachments/description_attachments entries now include "path" pointing to local files
# Downloads to OS temp dir by default, or specify: --download-attachments /tmp/att
```

**Step 2 (two-step alternative): Download separately**
```bash
# Download all at once (shows progress on stderr)
basecamp attachments download <id> --out /tmp/attachments
```

**Step 3: View images with your native file-read tool**
For multimodal LLMs (Claude, Gemini), use your file-read tool on the `path`
from the response to view downloaded images directly — no browser needed.
This surfaces visual context (mockups, screenshots, annotated designs) that
is often the most important part of a Basecamp todo or message.

```bash
# Stream a single image to stdout for piping
basecamp attachments download <id> --file mockup.png --out -

# Select by index when names collide
basecamp attachments download <id> --index 2 --out -
```

**Key pattern:** When a show command response contains `content_attachments`
or `description_attachments`, always download and view them — visual context is
often more important than the text content. Use `--download-attachments` for
one-shot fetch+download, or follow the breadcrumb hint for two-step control.

## Resource Reference

### Projects

```bash
basecamp projects list --json               # List all
basecamp projects show <id> --json          # Show details
basecamp projects create "Name" --json      # Create
basecamp projects update <id> --name "New"  # Update
basecamp projects trash <id>                # Move to trash (recoverable)
```

**Archiving a project:** the CLI does not have a dedicated archive command, but the
underlying status endpoint can be hit via raw API. Same path works for restoring
to active or moving to trashed.

```bash
basecamp api put "projects/<id>/status/archived" -d '{}' --json   # Archive
basecamp api put "projects/<id>/status/active" -d '{}' --json     # Unarchive
basecamp api put "projects/<id>/status/trashed" -d '{}' --json    # Trash (same as `projects trash`)
```

Verify with `basecamp projects show <id> --jq '.data.status'`.

### Todos

```bash
basecamp todos list --in <project> --json               # List in project
basecamp todos list --assignee me --in <project>        # My todos
basecamp todos list --overdue --in <project>            # Overdue only
basecamp todos list --status completed --in <project>   # Completed
basecamp todos list --list <todolist_id> --in <project> # In specific list
basecamp todos create "Task" --in <project> --list <list> --assignee me --due tomorrow
basecamp todos complete <id> [id...]                    # Complete (multiple OK)
basecamp todos uncomplete <id>                          # Reopen
basecamp assign <id> [id...] --to <person> --in <project>       # Assign to-do (multiple OK)
basecamp unassign <id> [id...] --from <person> --in <project>   # Remove to-do assignee (multiple OK)
basecamp assign <id> [id...] --card --to <person> --in <project>   # Assign card
basecamp unassign <id> [id...] --card --from <person> --in <project> # Remove card assignee
basecamp assign <id> [id...] --step --to <person> --in <project>   # Assign card step
basecamp unassign <id> [id...] --step --from <person> --in <project> # Remove step assignee
basecamp todos position <id> --to 1                     # Move to top
basecamp todos position <id> --to 1 --list <id|name|url> # Move to different list
basecamp todos sweep --overdue --complete --comment "Done" --in <project>
basecamp todos create "Task" --in <project> --list <list> --notify-on-completion "Jane,Bob"  # Notify when done
basecamp todos update <id> --notify-on-completion "Jane"  # Set who's notified on completion
basecamp todos update <id> --no-notify-on-completion      # Clear completion notifications
```

**Flags:** `--assignee` (repeatable; server-side account-wide, client-side within a project; also on `cards list` account-wide, but not on messages), `--status` (completed/incomplete/archived/trashed), `--overdue`, `--list`, `--due` (**listing filter: `with`/`without`/`overdue` only, account-wide only** — not a date; see Smart Defaults), `--limit`, `--all`

**Completion subscribers** ("When done, notify…"): set with
`--notify-on-completion <names or IDs, comma-separated>` on `todos create` and
`todos update`; clear with `--no-notify-on-completion` on `todos update`.
Plain updates (title, due date, etc.) preserve existing completion subscribers.

**Todo Subtasks (checklist steps):** Basecamp to-do subtasks are stored as
`Kanban::Step` records, even when their parent is a normal `Todo`. The regular
`basecamp todos show` response may not include them; use
`basecamp recordings list --in <project> --type Kanban::Step` and filter by
`parent.id` to list/check subtasks for a todo.

```bash
# Create a subtask under a todo.
# Use the numeric project ID and todo ID in this card-style path.
basecamp api post /buckets/<project_id>/card_tables/cards/<parent_todo_id>/steps.json \
  --data '{"title":"Subtask title"}' \
  --json

# Read or edit a subtask
basecamp api get /buckets/<project_id>/card_tables/steps/<step_id>.json --json
basecamp api put /buckets/<project_id>/card_tables/steps/<step_id>.json \
  --data '{"title":"Updated subtask title"}' \
  --json

# List subtasks for a todo
PARENT_TODO_ID=<parent_todo_id> \
basecamp recordings list --in <project> --type Kanban::Step --all \
  --jq '.data[] | select(.parent.id==(env.PARENT_TODO_ID | tonumber)) | {id,title,status,parent:.parent.id,url}'

# Assign or set a due date. Send only what you're changing — omitted fields are
# left alone. `assignee_ids` replaces the whole list, so name everyone who stays.
basecamp api put /buckets/<project_id>/card_tables/steps/<step_id>.json \
  --data '{"assignee_ids":[<person_id>,<existing_person_id>],"due_on":"<YYYY-MM-DD>"}' \
  --json

# Complete or reopen a subtask
basecamp api put /buckets/<project_id>/card_tables/steps/<step_id>/completions.json \
  --data '{"completion":"on"}' \
  --json
basecamp api put /buckets/<project_id>/card_tables/steps/<step_id>/completions.json \
  --data '{"completion":"off"}' \
  --json

# Trash a subtask from the todo UI by trashing the step record (Kanban::Step)
basecamp recordings trash <step_id> --in <project> --json
```

Key points: replace numeric placeholders such as `<project_id>`,
`<parent_todo_id>`, and `<person_id>` before running the examples. Bucket-scoped
API paths require a numeric project/bucket ID; `--in <project>` can still accept
a project name where CLI commands support name resolution. For creating todo
subtasks, Basecamp accepts the parent todo ID in the
`/buckets/<project_id>/card_tables/cards/<parent_todo_id>/steps.json` path. To
list subtasks under a todo, use
`basecamp recordings list --in <project> --type Kanban::Step` with the
`parent.id` filter shown above.

Completed subtasks have `completed: true` and a `completion` object with
`created_at` and `creator`. Open subtasks have `completed: false` and no
`completion` object. Trashed subtasks may still be readable directly with
`status: "trashed"` and `inherits_status: false`, but they no longer appear in
the todo UI.

In testing with todo-backed steps, these bucket-scoped direct `GET` requests
returned `not_found`:
`/buckets/<project_id>/card_tables/cards/<parent_todo_id>/steps.json`,
`/buckets/<project_id>/card_tables/cards/<parent_todo_id>.json`, and
`/buckets/<project_id>/todos/<parent_todo_id>/steps.json`. To inspect trashed
subtasks, add `--status trashed`; archived parents may require
`--status archived`.

**Raw step updates are partial.** `PUT .../card_tables/steps/<id>.json` leaves
every parameter you omit unchanged, so send only the fields you are changing.
Echoing back a `title` you did not mean to change is not merely redundant — it
reverts anyone who edited the title between your read and your write. To clear a
value, say so explicitly: `"due_on": null` clears the due date, `"assignee_ids":
[]` removes everyone. `assignee_ids` always replaces the whole list rather than
adding to it, so name every person who should remain assigned.

(This is bc3#12521. Before it, an omitted field *was* cleared and a title-less
update was rejected, which is why older guidance said to resend the title. Todo
subtasks and card steps share one endpoint and one contract — `PUT
card_tables/steps/:id` routes to the same controller for both.)

The generic
`basecamp assign <step_id> --step ...` command is intended for card steps and
may fail with `Bad Request` for todo-backed steps, so prefer `assignee_ids` on
the raw step update endpoint for todo subtasks.

### Todolists

Todolists are containers for todos. Create a todolist before adding todos.

```bash
basecamp todolists list --in <project> --json              # List todolists
basecamp todolists show <id> --in <project>                # Show details
basecamp todolists create "Name" --in <project> --json     # Create
basecamp todolists create "Name" --description "Desc" --in <project>
basecamp todolists create "Name" --visible-to-clients --in <project>  # Visible to clients
basecamp todolists update <id> --name "New" --in <project> # Update
basecamp todolists position <id> --to 1                     # Reorder one list (1 = top)
basecamp todolists position <id> <id> <id>                  # Order incomplete lists, top→bottom
```

Bulk `position` sets the visible order in one command: pass incomplete lists from
the same todoset, top to bottom. It always places them at the top.

### Cards (Kanban)

**Note:** `--assignee` on `cards list` is **account-wide only** — pass `--all-projects` (or have no project in scope) and it becomes a real server-side filter. Within a single project cards have no assignee filter: fetch all and filter client-side. `--due with|without|overdue` is account-wide only on cards too. If a project has multiple card tables, you must specify `--card-table <id>`. When you get an "Ambiguous card table" error, the hint shows available table IDs and names.

```bash
basecamp cards list --in <project> --json             # All cards
basecamp cards list --card-table <id> --in <project>  # Specific table (required if multiple)
basecamp cards list --column <id> --in <project>      # Cards in column
basecamp cards columns --in <project> --json          # List columns (needs --card-table if multiple)
basecamp cards show <id> --in <project>               # Card details
basecamp cards create "Title" "<p>Body</p>" --in <project> --column <id>
basecamp cards update <id> --title "New" --due tomorrow --assignee me
basecamp cards done <id|url> --in <project>           # Move to the Done column automatically
basecamp cards move <id> --to <column_id>             # Move to column (numeric ID)
basecamp cards move <id> --to "Done" --card-table <table_id>  # Move by name (needs table)
basecamp cards move <id> --to "Done" --position 1 --card-table <table_id>  # Move to position
basecamp cards move <id> --on-hold                    # Move to on-hold of current column
basecamp cards move <id> --to <column_id> --on-hold   # Move to on-hold of target column
```

**Cross-project card move (wormholes):** the only way to move a card to another
project is to teleport it through a *wormhole* — a portal on the card table that
sends cards to a preconfigured column on another project's card table (max 4 per
table). The teleport is **asynchronous and mints a new card id**: after the move
is accepted, the server copies the card into the destination and deletes the
original, so the **original id 404s** — do not reuse it.

```bash
basecamp cards wormholes list --in <project>          # Discover wormholes (id, destination, linked)
basecamp cards wormholes create --to-column <id|url> --in <project>   # Link to a column on another table (≤4)
basecamp cards wormholes update <id> --to-column <id|url> --in <project>
basecamp cards wormholes delete <id> --in <project>
basecamp cards move <card_id> --to-wormhole <wormhole_id> --in <project>          # Teleport (async)
basecamp cards move <card_id> --to-wormhole <destination_column_url> --in <project>  # Match by destination column
```

`--to-wormhole` is mutually exclusive with `--to`/`--on-hold`/`--position`. Pass
a numeric wormhole id to route directly, or a destination-column URL to match it
against the source table's wormholes.

**Archived/trashed cards:** `cards list` only returns active cards. For archived or trashed cards, use `basecamp recordings cards --status archived --in <project>` or `--status trashed`.

**Identifying completed cards:** Cards in Done columns have `parent.type: "Kanban::DoneColumn"` and `completed: true`. Use this to identify completed cards that haven't been archived.

**When a card moved columns:** don't read `updated_at` — it changes on any
modification. Use the event history instead: `basecamp events <card_id> --json`
records an `adopted` event for every column move, and a card crossing into or
out of a Done column pairs that with `completed`/`uncompleted`. See
[Events](#events-change-history).

**Card Steps (checklists):**
```bash
basecamp cards steps <card_id> --in <project>     # List steps
basecamp cards step create "Step" --card <id> --in <project>
basecamp cards step complete <step_id> --in <project>
basecamp cards step uncomplete <step_id>
```

**Column management:**
```bash
basecamp cards column show <id> --in <project>
basecamp cards column create "Name" --in <project>
basecamp cards column update <id> --title "New"
basecamp cards column move <id> --position 2
basecamp cards column color <id> --color blue
basecamp cards column on-hold <id>                # Enable on-hold section
basecamp cards column watch <id>                  # Subscribe to column
```

### Messages

```bash
basecamp messages list --in <project> --json  # List messages
basecamp messages show <id> --in <project>    # Show message
basecamp messages create "Title" "Body" --in <project>
basecamp messages create "Draft" "WIP" --draft --in <project>  # Create draft
basecamp messages publish <id>               # Publish a draft
basecamp messages update <id> --title "New" --body "Updated"
basecamp messages pin <id> --in <project>     # Pin to top
basecamp messages unpin <id>                  # Unpin
```

**Archived/trashed messages:** `messages list` only returns active messages. For archived or trashed messages, use `basecamp recordings messages --status archived --in <project>` or `--status trashed`.

**Flags:** `--draft` (create as draft), `--no-subscribe` (silent, no notifications), `--subscribe "people"` (comma-separated names, emails, IDs, or "me"; mutually exclusive with `--no-subscribe`), `--message-board <id>` (if multiple boards), `--visible-to-clients` (make visible to clients on the project; omit for the server default)

```bash
basecamp messages create "Bot update" "Done" --no-subscribe --in <project>
basecamp messages create "FYI" "Note" --subscribe "Alice,bob@x.com" --in <project>
basecamp messages create "For the client" "..." --visible-to-clients --in <project>
```

**Client visibility at create time:** `messages create`, `todolists create`,
`schedule create`, `checkins question create`, and `tools create` accept
`--visible-to-clients` to post a client-visible recording in one call (for
`tools create`, only chat and kanban_board tool types honor it — other types
inherit the project default). Omitting the flag uses the
server default, which is context-dependent: **team-only when you post as a team
member**, but a **client-authenticated caller always creates client-visible
records** (an explicit `--visible-to-clients=false` is overridden server-side for
client callers). Passing `--visible-to-clients` posts client-visible in every
case. To change visibility on an already-created recording, use
`recordings visibility <id> --visible`.

### Comments

```bash
basecamp comments list <recording_id> --in <project> --json
basecamp comments show <comment-id|comment-url> --json            # Now returns reply_target + paste-ready mention (JSON)
basecamp comments thread <comment-id|comment-url> --json          # Reply-ready: parent + focus + discussion + @mention
basecamp comments thread <comment-id> --all --json                # Every fetched comment instead of a window
basecamp comments thread <comment-id> --window 11 --json          # Focus-centered window of 11
basecamp comments create <recording_id> "Text" --in <project>
basecamp comments create <recording_id> "@Jane.Smith, looks good!" --in <project>  # With @mention
basecamp comments update <id> "Updated" --in <project>
```

**Cheap atoms vs. deep context (choose by need):**
- `comments show <url> --jq '.data | {reply_target, mention}'` — one API call. Returns
  `reply_target` (`recording_id` — where a reply is posted, comments are flat — plus
  `account_id`) and a paste-ready author `mention` (JSON only; human output shows a reply
  breadcrumb). Use this for the exact-comment reply atoms.
- `comments thread <url>` — two extra calls. Adds the full parent recording, the
  surrounding discussion (windowed, truncation-honest), and focus attachments. Use this
  when the surrounding discussion matters.

### Files & Documents

```bash
basecamp files list --in <project> --json               # List all (folders, files, docs)
basecamp files list --vault <folder_id> --in <project>  # List folder contents
basecamp files list --all-projects --json               # Across every project (first 100)
basecamp files list --all-projects --limit 500          # Walk pages until 500 collected
basecamp files list --all-projects --page 2             # Exactly page 2
basecamp files list --all-projects --all                # Every page (slow on big accounts)
basecamp files show <id> --in <project>                 # Show item (auto-detects type)
basecamp files versions <upload_id> --json              # Every version of an uploaded file
basecamp files versions <upload_id> --limit 5 --json    # Cap results (default: all)
basecamp files replace <upload_id> <file>               # Replace the file, keep the ID/URL/comments
basecamp files replace <upload_id> <file> --description "v2 notes"  # Also set a new description
basecamp files download <id> --in <project>             # Download file
basecamp files download <id> --out ./dir                # Download to specific dir
basecamp files download "https://storage.../download/f" # Download from storage URL
basecamp files uploads create <file> --in <project>      # Upload file to root
basecamp files uploads create <file> --vault <folder_id> --in <project>  # Upload to folder
basecamp files uploads create <file> --visible-to-clients --in <project>  # Client-visible (root folder only)
basecamp files folder create "Folder" --in <project>
basecamp files doc create "Doc" "Body" --in <project>
basecamp files doc create "Draft" --draft --in <project>
basecamp files doc create "Notes" "..." --no-subscribe --in <project>
basecamp files doc create "For client" "..." --visible-to-clients --in <project>  # Client-visible (root folder only)
basecamp files update <document_id> --title "New" --content "Updated"
basecamp files update <document_id> --title "New" --in <project>      # Preserves existing document content
basecamp files update <document_id> --content "Updated" --in <project> # Preserves existing document title
```

**Document update semantics:** `basecamp files update <document_id>` is safe for partial updates in the CLI: when you pass only `--title` or only `--content`, the CLI first fetches the current document and preserves the untouched field.

**Client visibility at create time:** `doc create` and `uploads create` accept
`--visible-to-clients`, but the server only honors it in the project's **root
Docs & Files folder**. Targeting a nested folder (`--vault`/`--folder`) with the
flag is a hard error raised before anything is uploaded — a nested item inherits
its folder's visibility, and that can't be changed per-item afterward (the
visibility endpoint rejects nested docs/uploads). To make a nested item
client-visible, create it in the root folder, or change the eligible top-level
ancestor that controls the folder's visibility first. Omitting the flag uses the
server default; as with Messages, a **client-authenticated caller always creates
client-visible records** regardless. `recordings visibility` is **not** a
remediation for nested docs/uploads.

**Upload versions:** replacing a file keeps the earlier copies under the same
upload ID, so `basecamp files versions <upload_id>` is how you see the history of
one file. A file that was never replaced returns its single current version, not
an error. Only `--page 1` is accepted; use `--all` to walk every page.
`basecamp files replace <upload_id> <file>` publishes a new version in place —
the upload keeps its ID, URL and comments, nobody is notified, and the
description carries forward unless `--description` is given. Use it instead of
`uploads create` when shipping a new build of the same file.

**Subcommands:** `folders`, `uploads`, `documents` (each with pagination flags)

### Schedule

For upcoming events across all projects, use `basecamp reports schedule --json`.

```bash
basecamp schedule info --in <project> --json       # Schedule info
basecamp schedule entries --in <project> --json   # List entries
basecamp schedule show <id> --in <project>        # Entry details
basecamp schedule show <id> --date 20240315       # Specific occurrence (recurring)
basecamp schedule create "Event" --starts-at "2024-03-15T09:00:00Z" --ends-at "2024-03-15T10:00:00Z" --in <project>
basecamp schedule create "Meeting" --all-day --notify --participants 1,2,3 --in <project>
basecamp schedule create "Sync" --starts-at "..." --ends-at "..." --no-subscribe --in <project>
basecamp schedule update <id> --summary "New title" --starts-at "..."
basecamp schedule settings --include-due --in <project>  # Include todos/cards due dates
```

**Flags:** `--all-day`, `--notify`, `--participants <ids>`, `--no-subscribe`, `--subscribe "people"` (mutually exclusive), `--status` (active/archived/trashed), `--visible-to-clients` (make visible to clients; omit for the server default)

### Check-ins

```bash
basecamp checkins --in <project> --json           # Questionnaire info
basecamp checkins questions --in <project>        # List questions
basecamp checkins question <id> --in <project>    # Question details
basecamp checkins answers <question_id> --in <project>  # List answers
basecamp checkins answers <question_id> --by me --in <project>  # My answers only
basecamp checkins answers <question_id> --by "Alice Smith" --in <project>  # Filter by person (name, email, or ID)
basecamp checkins answer <id> --in <project>      # Answer details
basecamp checkins question create "What did you work on?" --in <project>
basecamp checkins question update <id> "New question" --frequency every_week
basecamp checkins answer create <question-id> "My answer" --in <project>  # Defaults to today
basecamp checkins answer update <id> "Updated" --in <project>
```

**Schedule options:** `--frequency` (every_day, every_week, every_other_week, every_month, on_certain_days), `--days 1,2,3,4,5` (0=Sun), `--time "5:00pm"`

**Client visibility:** `checkins question create` accepts `--visible-to-clients` to make the question visible to clients (omit for the server default; see the note under Messages for the context-dependent rule).

**Managing a question:**

```bash
basecamp checkins question pause <id> --json      # Stop asking it
basecamp checkins question resume <id> --json     # Start asking it again
basecamp checkins question answerers <id> --json  # Who answers it
basecamp checkins question notify <id> --on-answer --json
basecamp checkins question notify <id> --no-on-answer --json
basecamp checkins question notify <id> --digest-include-unanswered --json
```

`notify` changes **your own** settings, and each one is left alone unless you
name it — so `--on-answer` does not silently reset the digest setting. The
`--no-...` spellings send an explicit false; passing neither setting is refused
rather than sent as an empty update.

**Your pending reminders** (account-wide, no `--in`):

```bash
basecamp checkins reminders --json
basecamp checkins reminders --limit 10 --json
```

`reminders` and `answerers` take `--limit` but deliberately **no `--page`**: the
API does not honor a page number on these, so the flag would accept a value it
could not act on.

### Timeline

```bash
basecamp timeline --json                          # Account-wide activity
basecamp timeline --in <project> --json           # Project activity
basecamp timeline me --json                       # Your activity
basecamp timeline --person <id> --json            # Person's activity
```

Use `--limit N` to cap results or `--all` to fetch everything (default: 100 events).

### Events (change history)

`basecamp timeline` reports activity across a project or account. For the audit
trail of one specific item — todo, card, message, document — use `basecamp
events`:

```bash
basecamp events <id|url> --json                   # Change history for one item
basecamp events <id> --limit 25 --json            # Cap results (default 100)
basecamp events <id> --all --json                 # Fetch everything
```

Common `action` values: `created`, `completed`/`uncompleted`,
`assignment_changed`, `content_changed`, `archived`/`unarchived`,
`commented_on`, and — for cards — `adopted`, which is recorded every time a card
moves to another column. That makes `events` the way to answer "when did this
card move?" or "when was this actually finished?", neither of which `updated_at`
can tell you.

`--page` accepts only `1`; use `--all` to walk every page.

### Event feed (account-wide, resumable)

`basecamp events <id>` is one recording's history. The account-wide **event
feed** is a different resource, and the way an agent hears about activity it
did not cause:

```bash
basecamp events poll --since now --json            # Enter at the present
basecamp events poll --position "$POSITION" --json # Resume from a held position
basecamp events poll --since 0 --all --json        # Replay served history
basecamp inbox --since now --json                  # Addressed items (agents only)
basecamp events ticket --json                      # Mint a live-stream ticket
```

Each page is an envelope: `events` (or `items`), a durable `position`, and a
`next` continuation URL while the current walk has more to serve. Persist
`position` only after processing the page, then pass it back with
`--position`. The response's `notice` spells the whole resume command out,
filters included. `--all` walks `next` to the end of the current walk — it is
not a live tail.

The feed is a notification lane, not an audit log:

- Deduplicate by event id — polls repeat what the live stream delivered, and
  an event can appear up to ~30 seconds after it happens.
- Deduplicate inbox items by `addressing_id`, never by event id: one event
  addresses you once per reason and each reason is its own item.
- Events are thin pointers. Refetch the recording (`basecamp show
  <recording_id>`) before acting on it.
- An agent that acts on what it hears should pass `--exclude-performers self`,
  which drops its own performances without hiding other agents' activity.

The two lanes have different filter sets, and they are not interchangeable:

- `events poll`: `--types`, `--buckets`, `--creators`, `--performers`,
  `--exclude-performers`, `--actor-types`.
- `inbox`: `--reasons`, `--types`, `--buckets`. The other four are not flags
  here.

Each takes a comma-separated list. These are the only narrowing available:
both commands accept the global `--in <project>` flag but ignore it, because
these are account endpoints. Narrow to a project with `--buckets`.

A position is bound to the filter set it was minted for, so a resume has to
carry the same filters. Changing them exits `1` naming both filter digests,
and the fix is to re-enter with `--since`. A position that is no longer
servable exits `2`, and the hint carries the whole re-entry command, filters
included.

`basecamp inbox` is served to agent principals only for now; other principals
get exit `4`. `basecamp events ticket` redacts the ticket and its URL unless
`--show-secret` is passed — both are bearers, so never log either, and mint a
fresh one per connection attempt.

### Recordings (Cross-project)

Use `basecamp recordings <type>` for cross-project type browsing. **For assigned todos, prefer `basecamp reports assigned`** — recordings do not include assignee data and cannot be filtered by person.

```bash
basecamp recordings todos --json                  # All todos across projects
basecamp recordings todos --all --json            # All todos (paginate through all)
basecamp recordings messages --in <project>       # Messages in project
basecamp recordings documents --status archived   # Archived docs
basecamp recordings cards --sort created_at --direction asc
basecamp recordings cards --status archived --all --json  # Include archived cards
```

**Types:** `todos`, `messages`, `documents`, `comments`, `cards`, `uploads`

**Status filtering:** By default, only `active` recordings are returned. Use `--status archived` or `--status trashed` to query other statuses. You may need separate queries to get complete data (e.g., active + archived).

**Status management:**
```bash
basecamp recordings trash <id> --in <project>     # Move to trash
basecamp recordings archive <id> --in <project>   # Archive
basecamp recordings restore <id> --in <project>   # Restore to active
basecamp recordings visibility <id> --visible --in <project>  # Show to clients
basecamp recordings visibility <id> --hidden      # Hide from clients
```

### Templates

```bash
basecamp templates list --json                    # List project templates
basecamp templates show <id> --json               # Project template details
basecamp templates create "Template Name"         # Create empty project template
basecamp templates update <id> --name "New Name"
basecamp templates delete <id>                    # Trash project template
basecamp templates construct <id> --name "New Project"  # Create project (async)
basecamp templates construct <id> --name "New Project" --start-date 2026-09-01  # Anchor template dates to that week
basecamp templates construction <template_id> <construction_id>  # Check project status

basecamp templates library --json                 # List active to-do list templates
basecamp templates copy <template_id> --in <project>  # Start copying into To-dos
basecamp templates copy-status <copy_id>          # Check copy status
```

**Asynchronous results:** `construct` returns a construction ID; poll `construction`
until `status="completed"` to get the project. A template's dates are relative to its
first week, and template weeks start on Sunday: `--start-date` (YYYY-MM-DD or natural,
e.g. `next monday`) anchors them to the Sunday on or before that date; without it they
anchor to the week of construction. `copy` returns a copy ID; poll
`copy-status` through `pending` and `processing` until it is `completed` or `failed`.

A copy can report the people who need access to the destination project. Show those
people to the user and rerun with `--confirm-adding-p