fb

A delightful, scriptable command line for Facebook. One binary that resolves a Page, profile, or group to a rich record, streams its recent feed, and pulls posts, comments, photos, videos, and events into clean structured data you can pipe anywhere, with no login and no browser.

fb page nasa

NAME  CATEGORY                 LIKES      FOLLOWERS  VERIFIED  URL
NASA  Aerospace company        24.1M      25.0M      true      https://www.facebook.com/nasa

Full documentation: facebook-cli.tamnd.com.

Why

Pulling data out of Facebook usually means a headless browser, a brittle pile of selectors, or the Graph API with its app review and tokens. fb takes a different route: it reads the public pages Facebook serves to search engines, parses the server-rendered HTML into typed records, and renders them in the output format you ask for. One static binary, no login, no browser, no API key.

Because it reads the public crawler surface, fb sees what a search engine sees: full public Pages, profiles, groups, and posts, the most recent feed rather than the entire history, and a post's preview comments rather than its whole thread. Private content stays private, and fb is explicit about the wall rather than silently returning nothing.

Install

go install github.com/tamnd/facebook-cli/cmd/fb@latest

Or grab a prebuilt binary from the releases page. The binary is pure Go with no runtime dependencies.

Build from source:

git clone https://github.com/tamnd/facebook-cli
cd facebook-cli
make build      # produces ./bin/fb

Quick start

fb page nasa                       # a Page's full profile
fb page nasa --posts --limit 20    # its twenty most recent posts
fb post <url> --comments           # a post and its comment thread
fb id <anything>                   # classify any Facebook id or URL

How it reads Facebook

fb reads anonymously, as a web crawler, with no login and no cookie. It asks Facebook for the same server-rendered pages a search engine gets and parses what comes back.

fb whoami        # reports the access mode and user agent

This works on any public Page, profile, group, or post. When a target is private, or behind a login wall, fb exits 4 with a one-line hint so scripts can tell that apart from a real error. The trade-off is depth: a feed exposes the most recent posts rather than the full history, and a post carries a few preview comments rather than its whole thread.

How it works

fb resolves any handle, id, or URL to a typed identity first (fb id shows exactly what it sees), then fetches the matching no-JavaScript page from mbasic.facebook.com and parses the HTML into records with goquery. Feeds and comment threads page through Facebook's "see more" links, so a --limit walks as many pages as it needs and stops cleanly. Responses are cached on disk by URL, so re-running a command is instant and polite to Facebook.

Every record is a plain struct with JSON tags, so -o json gives you the full shape and --fields narrows it. Nothing is invented: a field that Facebook does not surface anonymously simply stays empty rather than being guessed.

Commands

Command	What it does
page	A Page, fully resolved; --posts/--about/--photos/--videos/--events
profile	A person's public profile; --posts, --about, --photos
group	A group and its feed; --posts
post	One or more posts; --comments, --replies, --reactions
comments	Every comment and reply on a post
reactions	The reaction breakdown, and --list every reactor
photos / photo	A handle's photos, or one photo with full metadata
videos / video	A Page's videos and reels, or one with --streams
events / event	A Page's public events, or one event in full
search	Search pages, profiles, groups, posts, photos, videos, events
feed	Stream the feed of any handle (page/profile/group)
id	Classify any Facebook id or URL, no network needed
seed	Expand a root into a stream of URLs for crawling
crawl	Fetch a stream of URLs into full records (and optionally a DB)
db	Query the local SQLite store
whoami	Report how fb is accessing Facebook
config	Show resolved configuration and paths
cache	Inspect and clear the on-disk cache
completion	Generate a shell completion script
version	Print version, commit, and build date

Run fb <command> --help for the full flag list on any command.

Recipes

Pull a Page's last 50 posts as JSON Lines:

fb page nasa --posts --limit 50 -o jsonl

Get a post and its whole comment thread, replies expanded:

fb post <url> --comments --replies -o jsonl

See who loved a post:

fb reactions <url> --list --type love -o table

Collect every photo URL on a Page:

fb photos nasa --limit 200 -o url

Resolve a short link to a canonical id:

fb id "https://fb.watch/xxxxx" -o json

Build a dataset: expand a Page feed into URLs, then crawl each into SQLite:

fb seed page nasa --limit 100 | fb crawl --db nasa.db --comments
fb db --db nasa.db query "select owner_name, count(*) from posts group by 1"

Output formats

Every command renders through the same formatter. Pick a format with -o, or let fb choose: a table when writing to a terminal, JSON Lines when piped.

fb page nasa --posts -o table   # aligned columns for reading
fb page nasa --posts -o jsonl   # one JSON object per line, for piping
fb page nasa --posts -o json    # a single JSON array
fb page nasa --posts -o csv     # spreadsheet friendly (tsv too)
fb page nasa --posts -o yaml    # YAML documents
fb page nasa --posts -o url     # just the permalink

Narrow the columns with --fields, or template each row:

fb page nasa --posts --fields permalink,reactions_count,comments_count
fb page nasa --posts --template '{{.Permalink}} {{.ReactionsCount}}'

--raw prints the upstream HTML untouched, for when you want to parse it yourself.

Configuration

fb keeps its cache and data under the standard XDG paths (~/.cache/fb and ~/.local/share/fb by default; honor XDG_CACHE_HOME and XDG_DATA_HOME to move them). See the resolved paths and settings any time:

fb config show
fb config path

Useful global flags (all have sensible defaults):

Flag	Meaning
-o, --output	Output format (default auto)
-n, --limit	Maximum records (0 means unlimited)
--since / --until	Stop or skip feed items by date
--rate	Minimum delay between requests, to stay polite (default 2s)
--surface	mbasic, mobile, or auto
-j, --workers	Concurrency for fan-out commands
--no-cache	Bypass the on-disk cache

Exit codes

Code	Meaning
0	Success
1	Generic error
2	Usage error
3	Content not found or unavailable
4	Login wall: the content is not public
5	Rate limited
6	Network error

Development

make test    # run the test suite
make vet     # go vet
make build   # build ./bin/fb
make smoke   # run every command and assert it works or walls cleanly

The code is layered. cli/ is the command tree built on Cobra. fb/ is the library it sits on: the HTTP client, cache, the server-rendered HTML parsers for each record type, and the SQLite store. pkg/fbid/ is a standalone URL/id classifier with no other dependencies, importable on its own.

License

Apache-2.0. See LICENSE.