Since v2026.8.31 (#99007 gateway-owned room authority, #99047 replica takeover, #99099 gateway-side turn driver) a room whose members all live on one gateway continues without any Desktop attached. The Bot Mode page carries the cross-machine and replication beats but never states this durability, so readers still assume the Desktop drives the room. Adds one bullet under Groups and group chats; groups.capabilities.driver flag per tui_gateway/methods_groups.py.
hermes mcp login --flow device (RFC 8628, #104891, v2026.9.7) is documented on reference/mcp-config-reference but the user-guide MCP page, where people land when the loopback callback cannot reach their browser, still lists only Desktop relay, paste-back, SSH forward and proxied redirect_uri. Adds one bullet with a link to the reference section.
The code-execution page describes project vs strict mode, the minimal environment and the RPC channel, but never mentions that execute_code calls share a persistent per-session kernel (tools/code_kernel.py), that a timeout kills the kernel, that reset=true exists, or that stdout beyond 50 KB is spilled to cache/exec with the path returned (tools/code_execution_tool.py). Defaults from hermes_cli/config_defaults.py (kernel_idle_timeout 1800, max_session_kernels 4). Remote-backend kernel and fail-open per-call fallback from tools/code_kernel_remote.py.
The Automatic Cleanup section on the same page states that sessions.auto_prune is on by default (since #54189, default flipped to True in hermes_cli/config_defaults.py), while the tip box under the prune commands still said "auto-prune ships disabled. Enable it if...". A reader gets two answers from one page. Rewrites the tip to the shipped default and shows how to turn it off.
The delegation page and the delegation-patterns guide still stated the pre-v2026.8.31 defaults (50 iterations, 3 concurrent subagents). Shipped values: DEFAULT_MAX_ITERATIONS = 250 (tools/delegate_tool.py) and max_concurrent_children: 10 (hermes_cli/config_defaults.py). The per-task output_schema contract (one bounded correction retry, schema_valid / schema_errors on the result) was not documented anywhere on the page. The Max Iterations section also showed max_iterations as a per-call argument; delegate_task ignores caller-supplied values and reads delegation.max_iterations from config.
Preserve Ben Barclay's diagnosis and replace the staging-file approach with
SQLite-coordinated writes and a five-second backup callback deadline. A
main-file replacement can replay an abandoned destination WAL; immutable
source reads can miss committed source WAL. Neither raw copy nor replacement
is safe when the destination is locked.
Refuse unavailable auth databases without raw-copy fallback, retaining the
existing close-browser-and-retry flow. Keep two invariant tests for lock
refusal/recovery and source-versus-destination WAL contents. Convert existing
text masquerading as database fixtures into real SQLite fixtures.
Related: #105754
Related: #96659
A `browser_exec` call could park a thread in `sqlite3_sleep` permanently
while mirroring Chrome's auth DBs, holding the agent's turn open. The turn
never reaches its `finally`, so no `session.info running=false` settle is
emitted and the Desktop composer latches busy — every later message queues
and never sends. Captured live: one thread stuck 24+ minutes across two
dumps, turn accepted at 15:03 with no `tui turn finished` 46 minutes later.
Root cause is the DESTINATION, not the source. `Connection.backup()` retries
a busy destination internally and ignores the connection's busy timeout, so
`sqlite3.connect(dst, timeout=5)` cannot bound it. A destination left locked
by an earlier hung mirror therefore blocks the next mirror forever — and
because the tool-level 420s timeout abandons the thread without interrupting
a C-level lock wait, the lock is never released and every subsequent launch
re-hangs the same way. Self-perpetuating.
Two changes:
- Back up into a fresh `<dst>.new` and `os.replace()` it into place. No other
process can hold a file we just created, so there is nothing to contend on,
and the swap stays atomic. Measured against a live Chrome with a
deliberately locked destination: 0.0006s vs an indefinite hang.
- Drop the `mode=ro` (no `immutable=1`) source fallback. 8e746668ba added
`immutable=1` to fix exactly this hang but left `mode=ro` as a fallback,
keeping the unbounded path one exception away; sqlite's busy timeout does
not cover lock negotiation, so nothing bounds it. `immutable=1` is also the
semantically correct mode — a committed snapshot of a file another process
owns. The bounded plain-copy fallback is unchanged.
Tests: three regressions, all mutation-checked (fail on base, pass here).
The locked-destination test runs the copy on a worker with a join deadline so
the unfixed behaviour fails fast instead of hanging the suite. 199 passing
across the browser real-profile and CLI suites.
Salvage bounded row projection and status glyphs from PR #70899; hydrate the existing tree from session-scoped snapshots and open with Ctrl+T without disturbing drafts.
Co-authored-by: joaomarcos <joaomarcosdias444@gmail.com>
Project live children and background units for shared TUI/Desktop consumers; read a bounded tail from the existing runtime transcript. Reuse existing queued steering semantics rather than claiming delivery. Projection adapted from PR #70899 with exact live session and transport ownership.
Co-authored-by: joaomarcos <joaomarcosdias444@gmail.com>
witcheer