Replaces the single-line `NanoClaw` wordmark printed by
nanoclaw.sh with a multi-line splash frame: the lobster mascot
rendered as truecolor braille, drifting bubbles on either side,
the figlet wordmark below (Nano in bold, Claw in cyan bold),
three taglines — "Small.", "Runs on your machine.", "Yours to
modify." — and a navy seafloor line.
The frame is pre-rendered into `assets/setup-splash.txt` (built
from `assets/nanoclaw-icon.png` via chafa for the lobster +
figlet for the wordmark). nanoclaw.sh just streams the literal
bytes — no runtime dependency on chafa, figlet, or
ImageMagick.
Total height: 30 lines. Visible width: ~40 columns (fits any
terminal). Truecolor ANSI codes are used directly; terminals
without truecolor support will see a degraded but still
readable frame.
Also removes the standalone "Small. Runs on your machine.
Yours to modify." tagline line that nanoclaw.sh used to print
above the bootstrap spinner — those taglines now appear inside
the splash, so showing them again would duplicate.
The wordmark-suppression flow downstream (`setup:auto` honoring
`NANOCLAW_BOOTSTRAPPED=1`) is unchanged: the splash prints once
in nanoclaw.sh, setup:auto's `printIntro()` sees the flag and
keeps the clack `p.intro` line clean ("Let's get you set up.").
On GUI devices the URL was previously rendered dim inside the
instructional `note(...)` card, then `confirmThenOpen` printed
its prompt below: read the card, see the URL, then a separate
"Press Enter to open the X" prompt with no link near it. Two
visual moments for what's really one decision.
This PR pulls the URL out of the card on GUI devices and
relocates it directly under the action line of the confirm
prompt, separated only by a dim "If browser does not appear,
please visit: <url>" line:
│
◆ Press Enter to open the Developer Portal
│ If browser does not appear, please visit: … (dim)
│ ● Yes / ○ No
│
Action and fallback live as one prompt block — the user sees
both at the same time, no need to scroll back up to grab the
URL if the auto-open misses.
Headless behavior is unchanged: `formatNoteLink` still emits
"Get started: <url>" inside the card on headless devices (per
#2146), and `confirmThenOpen` still no-ops on headless (per
#2145). The only thing that changed for headless is the leading
`\n` in the helper output, which acts as a visual separator from
the steps above.
Five call sites adjusted (Discord ×3, Slack ×1, Telegram ×1) to
use `.filter((line) => line !== null)` so the now-nullable
`formatNoteLink` cleanly drops out of GUI-rendered cards.
When a card's auto-open is gated on `confirmThenOpen`, the URL also
appears inside the surrounding `note(...)` as a copy-paste fallback —
rendered dim because on a GUI device the auto-open is doing the
heavy lifting and the printed URL is just an incidental backup.
On headless devices the auto-open doesn't run (per #2145), so the
URL inside the note is the user's *only* path forward. A dim URL
reads as "incidental reference" exactly when it should be reading
as "this is the action."
Adds `formatNoteLink(url)` to setup/lib/browser.ts:
- GUI device → `k.dim(url)` (unchanged from today)
- Headless device → `Get started: <url>` at full strength
Replaces five call sites (Discord ×3, Slack ×1, Telegram ×1).
Single helper, atomic switch via the same `isHeadless()` plumbing
introduced in #2145, so the headless behavior across all five
flows stays in sync.
Wires the existing `isHeadless()` from setup/platform.ts into
`confirmThenOpen`. When the helper detects a headless device
(Linux without `DISPLAY`/`WAYLAND_DISPLAY`), both the
"Press Enter to open your browser" prompt and the actual
`openUrl(...)` call are skipped — there's no browser to launch
and the user can't usefully press Enter to summon one.
Why this is enough — the surrounding flow already supports the
headless path implicitly:
- Every `confirmThenOpen` call site sits beneath a `note(...)`
that prints the URL and the steps the user needs to take.
The URL is already visible to copy-paste onto another
device.
- Every site is followed by an explicit confirmation prompt
("Got your bot token?", "Done with the X?", etc.) that
naturally serves as the headless user's "I finished the
thing on my other device" signal.
So the headless branch becomes: read the note, do the thing,
answer the next prompt — without a useless "Press Enter to
open your browser" detour in between.
Coverage rationale (~95% accurate for the cases that actually
cause user confusion today):
- Linux + no `DISPLAY`/`WAYLAND_DISPLAY` → headless. Catches:
• Raspberry Pi headless installs
• Bare-metal Linux servers
• SSH'd into Linux without X11 forwarding
• CI environments on Linux
• Linux containers (which have no display)
- macOS → never headless. Even SSH'd Macs can usually still
open URLs through the local user's session, so treating
them as GUI-capable is the right default.
- Windows → never headless (effectively always GUI in
practice).
The remaining ~5% are edge cases (someone manually unset
`DISPLAY` on a desktop Linux session, etc.) that almost never
happen accidentally and recover gracefully — the URL is still
visible in the surrounding note.
Six call sites in channel adapters (Discord ×3, Slack ×1,
Telegram ×1, Teams ×1) all change behavior atomically through
the single helper. No per-site copy changes needed; consistency
is enforced by the central wiring.
Adds a `fmtDuration(ms)` helper in `setup/lib/theme.ts` that returns
`47s` under a minute and `1m 34s` from 60s onward, then routes every
elapsed-time spinner suffix in the setup flow through it. Replaces
the inline `Math.round((Date.now() - start) / 1000)` + `(${elapsed}s)`
pattern at every site.
Format is consistent past 60s — `1m 0s` over `1m` — so the live
spinner doesn't change shape at every whole-minute crossing.
Sites updated: setup/auto.ts, setup/lib/{runner,tz-from-claude,
claude-assist}.ts, and setup/channels/{signal,whatsapp,telegram,
discord,slack}.ts. Pre-allocated suffix budgets in `fitToWidth`
calls bumped from `' (999s)'` to `' (99m 59s)'` so long-running
steps don't blow past the reserved width.
The spinner label exceeded terminal width, breaking clack's cursor-up
redraw and causing each animation tick to print a new line instead of
updating in-place. Wrap with fitToWidth() like other setup spinners.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Move the MIME/type-to-extension maps and derivation helpers out of
session-manager.ts into a dedicated attachment-naming module — keeps
session-manager focused on session lifecycle and gives the helpers
a natural home for unit tests alongside the existing attachment-safety
module.
Two small fixes alongside the extraction:
- extForMime now guards `typeof mime !== 'string'` before .split, so a
buggy bridge passing `mimeType: { ... }` (object) no longer crashes
the inbound write loop.
- deriveAttachmentName computes Date.now() once per call instead of
twice, and tightens the explicit-name check to a string-and-truthy
guard so non-string values fall through to derivation.
Adds attachment-naming.test.ts with 11 cases covering MIME normalization
(case + parameters), Telegram type fallback, the non-string defensive
guard, and the bare-timestamp fallback.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Removing the "Not logged in · Please run /login" detection and
substitution from this PR — narrowing scope to just the OneCLI
gateway transient-retry change. The login-message handling will be
addressed separately.
Reverts:
- AgentProvider.isAuthRequired / authRequiredMessage
- ClaudeProvider auth-required regex, classifier, and remediation text
- poll-loop writeAuthRequiredMessage helper + call sites
- claude.test.ts (auth-only test file)
OneCLI/wakeContainer changes (the remaining content of the PR) are
unaffected.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Closes#1820.
The container agent-runner sets CLAUDE_CODE_AUTO_COMPACT_WINDOW
unconditionally on the container process env, with no way to override
it per-deployment without editing source. Read process.env first and
fall back to the existing 165000 literal when unset.
Default behavior is unchanged for installs that do not set the env
var. Operators running 1M-context models or emergency-tuning a live
deployment can now raise or lower the threshold from the host env.
Adds a paired `authRequiredMessage()` method to AgentProvider so
per-provider auth-failure remediation can differ. Claude returns the
Anthropic/`claude` instruction; future providers (Codex, OpenCode, …)
can return their own remediation text. The poll-loop calls
`provider.authRequiredMessage?.()` and falls back to a generic message
if a provider implements `isAuthRequired` without supplying its own
remediation.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
When a channel bridge passes an attachment without an explicit `name`,
extractAttachmentFiles fell back to `attachment-<ts>` with no extension.
Agents could not tell whether the file was a JPEG, PDF, or audio clip,
and tools keyed on extension (image viewers, exiftool, etc.) misbehaved.
Two cases are now covered:
1. Channels that set `mimeType` but no `name` (Discord/Slack documents,
Telegram document uploads). A small MIME-to-extension table covers
the common content types — image/*, audio/*, video/*, pdf, zip,
txt, json. Unknown MIMEs fall back to the unsuffixed name.
2. Channels that set `att.type` but no `mimeType` (Telegram photos,
stickers, voice, animations). The chat-sdk bridge sets a coarse
media-class (`photo` / `sticker` / `voice` / `video` /
`animation`) which is reliable enough to derive a canonical
extension. Telegram GIFs are MP4 under the hood.
The existing isSafeAttachmentName security guard is preserved — the
derived name still passes through it before disk I/O. The new lookup
tables emit static values from internal maps and cannot construct a
path-traversal payload; attacker-controlled att.name continues to flow
through the same validator.
- wakeContainer now never throws — returns Promise<boolean>, catches
internally. Closes the regression risk for the 5 awaited callers in
agent-to-agent, interactive, and approvals/response-handler that the
previous version left unwrapped. Router uses the boolean to stop the
typing indicator on transient failure; host-sweep just awaits.
- Tighten AUTH_REQUIRED_RE: anchor to start-of-string with the specific
`·` (U+00B7) separator the CLI uses, so an agent that quotes the
banner mid-sentence in a normal reply doesn't trip the classifier.
- Log a one-line note from writeAuthRequiredMessage so substitutions
are visible when debugging "user got the credentials message but I
don't see why."
- Add unit tests for ClaudeProvider.isAuthRequired covering both banner
variants, trailing content, mid-sentence quoting, leading-prose
quoting, alternate separators, and unrelated text.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Two related fixes for the case where credentials aren't usable:
1. Replace Claude Code's "Not logged in / Invalid API key · Please run
/login" output with a host-aware message. The user can't run /login
from chat, so the raw text is unhelpful. Provider gains an optional
isAuthRequired() classifier; the poll-loop substitutes the message
on both result-text and error paths.
2. Treat OneCLI gateway failure as a transient hard error instead of
spawning a credential-less container. The catch in container-runner
now propagates; router and host-sweep wrap wakeContainer to log and
leave the inbound row pending so the next 60s sweep tick retries.
Router also stops the typing indicator on failure.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Replace the hardcoded Approve/Ignore card with a multi-step flow:
- Single agent: "Connect to [name]" / "Connect new agent" / "Reject"
- Multiple agents: "Choose existing agent" (follow-up list) / "Connect new agent" / "Reject"
- "Connect new agent" prompts for a free-text name via DM, creates immediately on reply
- Add setMessageInterceptor router hook for capturing free-text replies
- Add resolveChannelName optional method to ChannelAdapter interface
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Wraps the word "assistant" in `accentGreen` (#3fba50, added in #2103)
across the six channel adapters that ask "What should your assistant
be called?" — Discord, iMessage, Signal, Slack, Telegram, WhatsApp.
Mirrors the green emphasis on "you" in the display-name prompt: the
green word names the subject of the question (assistant vs operator)
so the operator parses it at a glance.
github-actions[bot]