Commit graph

32 commits

Author SHA1 Message Date
38e4797ee9 Merge remote-tracking branch 'origin/main' into feat/files-sftp
# Conflicts:
#	cmd/agentbbs/main.go
#	internal/auth/auth.go
#	internal/store/store.go
2026-06-23 10:42:32 +00:00
006235ce92
Feat/mail all members (#55)
* feat(mail): give every verified member a free @bbs.profullstack.com mailbox

Email was built but paid-only (Founding Lifetime gate) and never wired to a
running backend. Make it a free benefit of membership and split the address
domain from the mail-server host.

- internal/mailu: Mailu admin-API client; EnsureUser idempotently provisions a
  mailbox via the loopback admin REST API (token = mailu.env API_TOKEN).
- main.go: auto-provision <name>@<mailDomain> at join@ verification and on first
  Mail open; un-gate the Mail hub entry + mail@ (membership/email-verified, not
  Premium); address domain (AGENTBBS_MAIL_ADDR_DOMAIN, default the BBS host) is
  now distinct from the mail server host (AGENTBBS_MAIL_DOMAIN) and the webmail
  URL. Drop the forwardemail alias path (Mailu now owns delivery for everyone).
- mailbox: gate on membership (a registered handle) instead of Paid;
  ErrNotPaid -> ErrNotMember.
- join@ copy: list email under free membership; premium now pitches custom
  domains + Tor only.
- setup.sh / docs/mail.md / deploy/mailu: address-domain vs server-host split,
  Mailu API token, MX for the address domain, local-relay SMTP for verify codes.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* chore(mailu): pin Docker network subnet to match SUBNET; ignore runtime state

The base compose declares no network, so Docker assigns the default bridge an
arbitrary subnet that won't match mailu.env SUBNET — breaking Mailu's internal
service auth/relay. Add a docker-compose.override.yml.example that pins the
default network to 192.168.203.0/24, and gitignore the live override + Mailu
runtime state (mailu.env, certs/, data/).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* feat(mail): plaintext loopback IMAP so the gateway bypasses Mailu's front

Mailu's front (nginx mail proxy) pre-authenticates against Mailu's user DB before
proxying to Dovecot, which rejects the Dovecot master-user login <addr>*gateway.
The gateway must reach Dovecot directly. The imap container has no TLS cert (only
the front does), so the bypass is plaintext over loopback — the master password
never leaves the host.

- mailbox: IMAPConfig.Plaintext dials with DialInsecure (loopback only).
- main.go: mailClientFor sets Plaintext from AGENTBBS_MAIL_IMAP_PLAINTEXT.
- override.example: add the unbound resolver (admin needs DNSSEC), webmail image
  fix (2024.06 uses mailu/webmail), and publish Dovecot 143 on 127.0.0.1:14143.
- docs/mail.md: document the front-bypass, the dovecot.conf master passdb (Mailu
  includes that exact filename), and the 644 master-users perms (640 = temp_fail).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* deploy(mailu): wire gateway IMAP to the loopback Dovecot path in setup.sh

setup.sh §9e set AGENTBBS_MAIL_IMAP_ADDR to the front's :993, which the front's
auth proxy rejects for the master-user login (and would clobber the working
loopback wiring on every self-update). Point it at 127.0.0.1:14143 +
AGENTBBS_MAIL_IMAP_PLAINTEXT=1 instead, matching the override + docs.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* feat(mail): give free members a webmail password at join@

The gateway opens mailboxes via the Dovecot master user (no member password),
but webmail (Roundcube) needs the member to have a password. join@ now sets a
fresh, readable webmail password via the Mailu API and shows it with the webmail
URL + login, so free members can use webmail at mail.profullstack.com.

- mailu: SetPassword (PATCH /user/<email> raw_password) + test.
- main.go: setWebmailPassword + readablePassword; join@ displays url/login/password.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-23 03:38:31 -07:00
6dc94bd784 feat(files): SFTP member storage — private workspaces + shared public area + mgmt TUI
Implements M4 (Files). A fully virtual Go SFTP server (pkg/sftp + crypto/ssh,
no OS users) wired as an "sftp" subsystem on the existing :22 wish listener, so
members reach their files with their login key:

    sftp files@bbs.profullstack.com     # scp/rsync ride the same endpoint

Identity is the SSH key (the username is conventional/ignored). Two areas per
session: a private, quota-limited /me workspace and a single shared public file
area /public (old-school BBS file area; world-read, members-only write by
default, operator-moderated). This reverses the old NG1 "no sharing" boundary in
favour of one sanctioned, inspectable sharing surface (PRD §9.3 amended).

internal/files:
- backend.go  service, layout, quota/usage, live-session registry, operator API
- fs.go       per-session virtual FS; resolve() is the single security
              chokepoint (area confinement + symlink-escape guard) + pkg/sftp
              request handlers
- server.go   subsystem handler: key auth -> member session -> request server,
              with byte metering and force-disconnect
- tui.go      in-BBS member browser (hub plugin "Files")
- admin.go    operator management TUI: sessions, workspaces/quotas, public area

Operator console: ssh sftp@<host> (allowlist-gated; sftpadmin@/filesadmin@
aliases) — list/disconnect sessions, set per-user quotas, revoke SFTP access,
toggle public write, moderate the public area.

store: files_access (per-user quota override + revoked) and files_settings
(public-write mode) tables + methods. main.go wiring guarded by AGENTBBS_FILES
(+ AGENTBBS_FILES_QUOTA_MB, default 1 GiB). Route names reserved.

Tests (incl -race): path traversal/confinement, symlink-escape rejection,
public-write ACL, quota enforcement, usage accounting, and an end-to-end run
against a real SFTP client. Docs: docs/files.md; PRD §5.3/§5.3.1/§9.3 + README
updated.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-23 09:35:04 +00:00
de5517c000
fix(pods): never downgrade AGENTBBS_POD_IMAGE on a transient build failure (#54)
The deploy's rootless-podman context intermittently fails (pre-pull/build/even
image-exists), which made setup.sh upsert AGENTBBS_POD_IMAGE back to the base
ubuntu and silently strip Claude Code/Codex from pods. Only upsert when we
actually have localhost/agentbbs-pod:latest; otherwise leave the configured
value untouched (the agentbbs daemon uses the local image from its own session).

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-23 02:31:27 -07:00
94ce79374c feat(pods): SSH agent forwarding → git push from the pod with your key
Code in your pod and push to git.profullstack.com using the SAME SSH key you
signed in with — nothing is copied into the pod. When a member attaches with
agent forwarding (ssh -A), agentbbs listens on a fresh unix socket in a
per-user agent dir bind-mounted at /run/agentbbs-agent and proxies it back over
the session; the pod shell gets SSH_AUTH_SOCK pointed at it. The pod image's
ssh_config sends git@git.profullstack.com to Forgejo's SSH server (:2222), so
`git clone git@git.profullstack.com:you/repo.git` just works.

- pods.go: agentDir + startAgent (per-session socket, cleaned up on exit);
  Attach injects SSH_AUTH_SOCK when ssh.AgentRequested; ensure() bind-mounts the
  agent dir and self-heals idle pods missing it. No main.go change needed —
  charmbracelet/ssh sets AgentRequested from the session request loop.
- pods/Containerfile: /etc/ssh/ssh_config.d entry (port 2222, user git,
  accept-new) so the conventional git@ URL reaches Forgejo.
- setup.sh: keep using an already-built pod image if a later rebuild
  transient-fails, so a flaky deploy never downgrades pods to the base image.

Build/vet/test/gofmt clean. Image rebuilt on the host; `ssh -G
git.profullstack.com` resolves to port 2222 / user git. End-to-end push needs a
live `ssh -A` session (validate after deploy).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-23 09:16:29 +00:00
281d4b55fd
Merge pull request #52 from profullstack/feat/landing-page
feat(web): rich landing page — BBS history + all commands
2026-06-23 02:05:34 -07:00
2330d9dcb7
Merge pull request #51 from profullstack/feat/pod-ai-tools
feat(pods): Claude Code + Codex in member pods
2026-06-23 02:05:19 -07:00
f6ce174cae feat(web): rich landing page — what a BBS is + every command
Replace the stub site root with a full page: a short "what's a BBS?" history
(1980s dial-up boards, SysOps, door games, FidoNet/Usenet) and the complete
command list — join/bbs/NAME/pod/mail/news/irc/game/domain over SSH — plus the
web services (AgentGit profiles, IRC, member homepages) and the git
"membership is your account" push flow. Regenerated on every run (templated
from $DOMAIN/$GIT_DOMAIN/$IRC_DOMAIN) so it stays current as features land.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-23 08:59:14 +00:00
c8edd2eed2 feat(pods): Claude Code + Codex in member pods (custom image)
Members can now code in their pod: build a custom pod image (FROM the base
Ubuntu) that ships git, openssh-client, Node.js 22, and the Claude Code
(`claude`) and Codex (`codex`) CLIs. BYO key — no credentials are baked in; a
member exports their own ANTHROPIC_API_KEY / OPENAI_API_KEY (or uses the tools'
login flow), stored in their persisted home.

- pods/Containerfile: the image (also drops a BYO-key + git-push login hint).
- setup.sh: build it on the host (rootless podman, layer-cached), switch
  AGENTBBS_POD_IMAGE to localhost/agentbbs-pod:latest (upserted for existing
  installs), keeping the base image if the build fails.
- pods.go: image-aware self-heal — an idle pod on an out-of-date image is
  recreated (home volume kept) so the new tooling rolls out without a manual
  rebuild and without disturbing active sessions.

Verified: image builds on the host; inside it node v22, git, ssh, `claude
--version` (2.1.186) and `codex --version` (0.142.0) all run.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-23 08:48:38 +00:00
f210b4296b feat(agentgit): register members' SSH keys + enable git push over SSH
Make git.profullstack.com a real, key-authenticated git host for every BBS
member ("BBS membership is the git account", SSH-key auth end to end):

- forgejo.EnsureKey: register a member's SSH public key on their Forgejo
  account (idempotent, ignores the key comment). So the key they sign in to
  the BBS with is also their git push key.
- provisionGit now takes the session public key and registers it after
  ensuring the account; called on email verification AND (newly) on every
  member login, so members who predate AgentGit — or whose key wasn't
  registered yet — are backfilled automatically and off the hot path.
- setup.sh:
  - admin token scopes write:admin,read:user,write:user (the old write:admin
    alone failed userExists' /users lookup, so provisioning never worked).
  - REQUIRE_SIGNIN_VIEW=false so member profiles + public repos are viewable
    at git.profullstack.com/<name> (private repos stay private; accounts are
    still created only by agentbbs).
  - Enable Forgejo's built-in SSH server (port 2222, BUILTIN_SSH_SERVER_USER=git)
    and open the firewall, so members push to git@git.profullstack.com:2222.

Verified live: all members provisioned, git.profullstack.com/chovy serves the
profile, and a push over ssh://git@host:2222 with a registered key succeeds.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-23 08:39:37 +00:00
e0a267e343
Fix/pod public html bind mount (#30)
* feat(pods/admin): root admin alias, default-caps cleanup, pod rebuild script

- auth: add `root` as an admin-console route alias (alongside admin/sysop);
  still gated by $AGENTBBS_ADMINS — the name confers nothing on its own.
- pods: drop the now-redundant tuneApt apt-sandbox hack. Rootless podman keeps
  its default capability set, so apt/chown/su work without disabling the
  download sandbox.
- scripts/rebuild-pods.sh: recreate all member pods (keeps home volumes) so
  they pick up the current container profile on next `ssh pod@`.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* feat(arcade,ui): 80s arcade classics + shared menu theme

Arcade games (PRD §5.1), generalizing the sandboxed-PTY DOOM path into an
external-game registry: Space Invaders (nInvaders), Pac-Man (pacman4console),
Tetris (tint/vitetris), Moon Patrol (moon-buggy). Binaries resolve from
assets/bin, PATH, then /usr/games, so a distro install or a hand-built binary
lights each game up; missing games are skipped with a discovery hint. Installed
on the host via `scripts/fetch-assets.sh --arcade` (apt), wired into setup.sh
behind FETCH_ARCADE (default on).

UI: new shared ui.Theme.MenuItem widget (accent cursor + badge, description
shown only for the focused row) adopted by the hub and arcade menus; the hub
groups rows under Features/Sessions headers and the arcade under
DOOM/ARCADE/BUILT-IN.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-15 08:06:32 -07:00
d763d732a3
fix(deploy): build Go binaries on the runner, ship them, SKIP_BUILD on box (#27)
The deploy SSHed into the ~458MB droplet and ran `go build` there. The Go
linker's peak memory OOM-killed the build — and with it the sshd serving the
deploy session — surfacing as "Connection closed by remote host" (exit 255).
It was flaky because it tracked momentary memory pressure from the co-resident
ergo/forgejo/tor/podman/agentbbs processes (run #25 passed, #26 failed on
near-identical code).

Build both binaries on the 16GB GitHub runner instead (pure-Go, modernc
sqlite, so CGO_ENABLED=0 static cross-build), scp them to the droplet, and run
setup.sh with SKIP_BUILD=1 so the box never compiles. Arch is detected from
the droplet so amd64/arm64 both work. setup.sh now also skips the Go toolchain
download when SKIP_BUILD=1.

Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-15 07:09:57 -07:00
336ff00fa3
Merge pull request #15 from profullstack/feat/irc-store-auth
feat(irc): gate IRC on the BBS user store; remove ssh irc@; external clients only
2026-06-14 20:23:54 -07:00
cb09300357 Self-host Mailu mail stack at mail.profullstack.com
Members get a real <name>@mail.profullstack.com mailbox, served by a
co-located Mailu (Postfix+Dovecot+Roundcube+rspamd) Docker stack. Coexists
with the host Caddy: Mailu owns the mail ports; Caddy fronts the loopback
webmail and supplies the TLS cert (TLS_FLAVOR=mail), the same cert-copy
pattern as the Ergo/IRC and NNTP services.

- deploy/mailu/: docker-compose.yml, mailu.env.example, refresh-certs.sh
  (copy Caddy's mail cert into Mailu on renewal), provision-mailbox.sh
  (member mailbox + Dovecot gateway master user), README.
- setup.sh: MAIL flag + mail.${DOMAIN#*.} Caddy site + §9e (cert timer,
  mail-port firewall, conditional compose bring-up, AGENTBBS_MAIL_* env).
- docs/mail.md: architecture, DNS (MX/SPF/DKIM/DMARC/PTR), gateway
  master-user setup, env, provisioning, webmail-only policy.

Apex profullstack.com stays corporate; member mail is only on mail.*.
Infra is inspection-verified (bash -n, YAML lint); deploy pending.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-14 16:33:25 +00:00
243ef58e59 feat(irc): gate on BBS user store (loopback /irc-auth), drop ssh irc@
Rework the IRC access model to the user-level source of truth — the
bbs.profullstack.com user store — per the agreed design:

- Auth via the store, not OS users. New loopback endpoint /irc-auth (on the
  /verify server) answers {member,premium} from the store via UserByName. Ergo's
  auth-script now curls it instead of `getent passwd`; the OS-user (tilde.town)
  reconcile in setup.sh is reverted. require-sasl on, registration off.
- Remove the in-BBS `ssh irc@` route: delete internal/irc (in-process client +
  TUI), handleIRC/runIRC, the hub IRC menu entry, and the dispatch case. Members
  connect with an external client (or web) at irc.profullstack.com:6697 / wss.
  IRCNames kept only to reserve the "irc" name (IsIRCName removed).
- Keep external clients; keep the irc.profullstack.com hostname + cert work.

Premium channel creation: /irc-auth already returns each account's premium
status, but enforcement is NOT yet wired (the old irc@ /create was removed with
the route; Ergo can't gate creation per-account natively). operator-only-creation
stays off for now; server-side premium gating (a ChanServ-style bot) is a
documented follow-up. See docs/irc.md.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-14 16:24:38 +00:00
5eb1e96480 feat(irc): irc.profullstack.com host, OS-user (tilde.town) gate, premium channels
Hostname: serve the network as irc.profullstack.com (new IRC_DOMAIN var,
default irc.<root-of-DOMAIN>). Caddy serves an irc.profullstack.com site so it
gets a Let's Encrypt cert; ergo-refresh-certs copies that into Ergo for 6697.
Needs an A record irc.profullstack.com -> the box (self-signed until it resolves).

Members are OS users (tilde.town model): setup.sh reconciles a real OS account
per member dir (root-side, on each deploy + the 15-min timer; nologin shell, so
identity-only — no shell access). The IRC auth-script now gates on
`getent passwd` with uid>=1000 instead of the member dir, so "OS user" == member.

Premium channels: free members may /join; creating channels is a premium perk.
The ssh irc@ client gains /create #name (premium-gated via ensurePremium): it
joins the fresh channel and registers it with ChanServ as the member's founder.
v1 gate is route-level (operator-only-creation left off); external-client
creation hardening is a follow-up. See docs/irc.md.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-14 15:23:44 +00:00
3b6b9a4a78 feat: members-only Usenet (NNTP) + Forgejo git provisioning + founding-lifetime $99
WIP feature branch: NNTPS news server, per-member Forgejo accounts on email
confirm, and founding-lifetime pricing tier.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-14 14:26:23 +00:00
3d0d483239 deploy: inject AGENTBBS_QRYPT_ISSUER_KEY to enable qrypt invite minting
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-14 11:46:45 +00:00
97da723c5c feat(qrypt): qrypt.chat anonymous-invite issuer
AgentBBS becomes a trusted Ed25519 issuer for qrypt.chat anonymous
accounts. Members mint a signed, single-use qci1 token (per the shared
invite contract) that qrypt.chat verifies and redeems.

- internal/qryptinvite: Mint / GenerateIssuerKey / ParsePrivateKey +
  Config (AGENTBBS_QRYPT_* env) with unit tests (independent verify,
  payload assertions, jti uniqueness, tamper rejection, seed/full key).
- store: qrypt_invites table + QryptInviteCount / RecordQryptInvite
  (per-member quota, enforced in a tx; ErrQuotaExceeded) + test.
- plugins/qryptinvite: hub plugin (members only) — checks quota, mints,
  records, prints token + redeem URL.
- cmd/agentbbs: `qrypt-invite <user>` and `qrypt-issuer-keygen`
  subcommands wired into dispatch.
- setup.sh env template + docs/qrypt-invites.md.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-14 11:35:05 +00:00
8adafaf515 feat(irc): members-only Ergo IRC network co-located on the BBS
Provision a self-hosted Ergo IRC network (irc.${DOMAIN}) in setup.sh §9b:
single Go binary on its own ports/user, reusing Caddy's Let's Encrypt cert
for 6697 TLS (refreshed by a timer; self-signed fallback on first boot),
loopback 6667 + a loopback WebSocket fronted by Caddy at wss://${DOMAIN}/irc.

Access is MEMBERS-ONLY: every client must authenticate with SASL, self-service
registration is off, and an auth-script (deploy/ergo/auth-script.sh, installed
as /usr/local/bin/ergo-auth-member) approves a login only if the account name
maps to an existing AgentBBS member home dir under <data>/users/. Passphrase is
ignored — membership (the filesystem dir) is the credential. require-sasl has
no IP exemption so WebSocket clients (which reach Ergo via Caddy from 127.0.0.1)
can't bypass the gate; accounts are auto-created on first successful auth.

Public attack surface is TLS-only (ufw opens 6697; 6667 is loopback). Toggle
with IRC=0. See docs/irc.md.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-14 11:26:07 +00:00
4be87440d5 tor: premium tor-url / tor-irc / tor routes over Tor
Add three premium-gated SSH routes:
  ssh tor-url@host <url>        one-shot HTTP(S) GET over Tor (host-side,
                               curl via SOCKS, 30s/2MB caps, http/https only)
  ssh -t tor-irc@host <server>  interactive IRC over Tor in the member's pod
  ssh -t tor@host <command...>  run any command over Tor (torsocks) in the pod

tor-url runs host-side and constrained; tor/tor-irc run inside the member's
isolated pod (new pods.Exec) so arbitrary/interactive commands are sandboxed,
never on the host. internal/tor wraps curl/torsocks/irssi. All gated by
ensurePremium; names reserved. setup.sh installs + enables tor (SOCKS
127.0.0.1:9050) and torsocks.

Note: tor-url is host-side and self-contained. tor/tor-irc still need the pod
image to carry torsocks+irssi and reach the Tor SOCKS — follow-up.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-14 10:54:21 +00:00
31c00cd65a setup.sh: persistent Go cache + HOME so timer deploys build instead of OOM/erroring
The self-update timer runs setup.sh as a systemd oneshot with no $HOME, so
`go build` aborted with "module cache not found: neither GOMODCACHE nor GOPATH
is set" before compiling — auto-deploys silently pulled new code but never
rebuilt or restarted. On top of that, a cold cache made the 458MB droplet
OOM-kill the compiler.

Set explicit, root-owned, persistent Go caches (/var/cache/agentbbs/{go,go-build})
plus HOME and GOMAXPROCS=1 in setup.sh, and add HOME=/root to the generated
agentbbs-update.service. A warm persistent cache means incremental redeploys
recompile almost nothing, keeping peak memory within RAM+swap; combined with
the existing `go build -p=1` the compiler no longer gets OOM-killed.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-14 10:31:29 +00:00
ceaf055e0e setup.sh: restart agentbbs on (re)deploy so rebuilt binaries actually load
enable --now leaves an already-running service untouched, so a redeploy's new
binary never loaded. Use enable + restart.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-14 09:28:31 +00:00
31c444829d setup.sh: fix wildcard Caddy block — use host label placeholder (host_regexp isn't a built-in matcher)
caddy validate failed: 'module not registered: http.matchers.host_regexp'.
Drop the regexp; map the subdomain via {http.request.host.labels.N} where N is
the DOMAIN label count. The on-demand ask endpoint still gates cert issuance to
registered members.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-14 09:23:03 +00:00
ae8b1c25f4 Free per-user homepages at <name>.<host> (wildcard subdomains)
- sites.AskHandler now approves <name>.AGENTBBS_HOST on-demand-TLS when <name>
  is a registered member (so only real users mint subdomain certs).
- Caddyfile: *.${DOMAIN} block serving users/<name>/public_html with on_demand
  TLS; unknown users 404. Needs a wildcard DNS record *.${DOMAIN} -> host.
- Test: TestAskUserSubdomain.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-14 09:20:34 +00:00
74c292f341 setup.sh: move socket-activated sshd (Ubuntu 24.04) to the admin port
On Ubuntu 22.10+/24.04 sshd is socket-activated via ssh.socket, which owns the
listen port and ignores sshd_config's Port — so the admin-port move silently
failed (safety check aborted before freeing :22). Override ssh.socket's
ListenStream to ADMIN_SSH_PORT when present; fall back to restarting the service
on classic setups.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-14 09:15:37 +00:00
2600b1e3b6 setup.sh: support tiny droplets — SKIP_BUILD (prebuilt binaries), auto-swap, -p=1, unzip
512MB droplets OOM-kill the Go build (livekit/redis/modernc deps). Add:
- ensure_swap: adds a 3G swapfile on low-RAM hosts (build + runtime headroom)
- SKIP_BUILD=1: use prebuilt /usr/local/bin/{agentbbs,ascii-live} (copied from
  CI or a dev box) instead of compiling on the host
- go build -p=1 to cap peak memory when building on-box
- install unzip (Freedoom fetch needs it)

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-14 09:08:30 +00:00
7fd10c3436 payments: call CoinPay REST API directly (drop coinpay CLI dependency)
CreatePremiumCharge/VerifyPremium now POST /payments/create and GET
/payments/:id against the CoinPay API (Bearer COINPAY_API_KEY, business_id =
AGENTBBS_COINPAY_MERCHANT_ID), so the droplet needs no coinpay CLI installed.
The created payment id is stored (store: User.PremiumPayID + premium_pay_id col
+ SetPremiumPayment); ensurePremium verifies that id on a later connect and
grants premium on confirmed/forwarded status. Removed the CLI command-template
env knobs; added AGENTBBS_COINPAY_API_URL. Tests: httptest-backed payments_test
+ store SetPremiumPayment test. Build/vet/gofmt/test green.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-14 08:27:33 +00:00
5c0feda8d4 Wire CoinPay merchant id + deploy secret injection
- payments: AGENTBBS_COINPAY_MERCHANT_ID -> --business-id on the premium
  create/pay commands (coinpay CLI reads COINPAY_API_KEY from env for auth).
- deploy.yml: forward COINPAY_API_KEY + COINPAY_MERCHANT_ID (GitHub secrets,
  masked in logs) to the remote setup.sh.
- setup.sh: idempotent upsert_env writes those secrets into agentbbs.env
  (COINPAY_API_KEY, AGENTBBS_COINPAY_MERCHANT_ID, COINPAY_BUSINESS_ID),
  preserving the rest. No secret values are committed.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-14 08:19:48 +00:00
3230807421 Autonomous deploy + free-pod/Premium-email membership
Deploy automation (idempotent, runs on every deploy):
- .github/workflows/deploy.yml: push to main/master (or dispatch) SSHes to the
  droplet and re-runs setup.sh; deploys the pushed branch; smoke-tests :22.
- scripts/self-update.sh + agentbbs-update.timer: autonomous backstop that
  redeploys only when origin advances.
- setup.sh hardened: flock, fetch+reset (survives force-push), fixed the
  always-skipped arcade asset fetch path.

Membership model:
- Free, email-verified members get their own Docker pod (pod@ paywall removed)
  and a /~name homepage (seeded at join@).
- join@ is now interactive: email -> emailed 6-digit code -> enter code.
- Premium ($10 one-time, lifetime via CoinPay) grants a personal
  <name>@host email (new internal/forwardemail; forwardemail.net aliases) and
  custom domains (domain@ gated to Premium).
- ensurePremium() silently verifies/grants/provisions on hub login, join@, and
  domain@. New-signup details emailed to AGENTBBS_SIGNUP_NOTIFY (subject "bbs").

Store: User.Premium + premium/premium_ref cols, ConfirmEmailCode, GrantPremium.
Tests: store_premium_test.go, forwardemail_test.go. Build/vet/gofmt/test green.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
2026-06-14 07:58:51 +00:00
85024afcf6 setup.sh: wire join@ email verification through Caddy
Proxy /verify on the main site to the agentbbs loopback endpoint and
surface AGENTBBS_HTTP_ADDR + the SMTP knobs in the generated env file
(links are logged when SMTP is unconfigured; existing env files keep
working via the in-binary 127.0.0.1:8088 default).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-11 15:54:27 +00:00
f3f8e70996 bbs.profullstack.com launch kit: provisioner, custom domains, email verify, ascii-live
- setup.sh: idempotent one-shot droplet provisioner — agentbbs on :22
  (admin OpenSSH moved to :2202), rootless podman, Caddy front end for
  https://bbs.profullstack.com with tilde-style /~user homepages
- internal/sites + domain@ SSH route: self-service custom domains
  (ssh domain@host add example.com) backed by a symlink farm and an
  on-demand-TLS ask endpoint so Caddy only issues certs for mapped hosts
- internal/mail + join@ email verification: optional email at signup,
  confirmation link served by a loopback /verify endpoint behind Caddy
- internal/source + cmd/ascii-live: live video → terminal ASCII groundwork
  (docs/ascii-live.md)
- store: additive sqlite migrations (email/verify columns, domains table)

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
2026-06-11 15:53:19 +00:00