A second (or later) invocation of a --persistent sandbox failed with:
cp: cannot create regular file '.../root/certs/openssl.cnf': Permission denied
`DEV_SANDBOX_ASSETS` is a Nix store path when the sandbox is invoked via
the `sandbox` wrapper (nix/sandbox.nix sets it to a store-copied
scripts/sandbox/ directory), and Nix store files are always mode 0444.
Plain `cp SRC DEST` opens an existing DEST for writing in place rather
than replacing it, so once one sandbox run has copied a read-only
openssl.cnf/proxy.py into a --persistent root, every subsequent run on
that same root fails trying to overwrite its own prior copy.
This silently broke every multi-invocation --persistent scenario,
including tests/install/install-update-e2e.sh's update and installer
routes (both run `install` a second time in the same persistent root).
Reproduced identically on unmodified main.
Fix: `cp -f`, which unlinks and recreates the destination instead of
writing through it, so the destination's mode never matters. Verified
by running `sandbox install --persistent -- --skip-setup --skip-browser`
twice against the same root -- second run failed before this change,
passes after.
`hermes --version` only proves the venv and entry point execute; it says
nothing about whether a chat turn actually completes end to end (config
loading, provider resolution, the model-call path, tool loop, response
rendering).
Add require_hermes_chat_round_trip(), which inside the same sandbox
invocation: starts the shared mock inference server
(scripts/mock-inference-server.mjs, already available via the tarred-in
repo checkout at /work/repo -- no extra fixture plumbing needed), points
a generated config.yaml/.env at it, and runs `hermes -z "hi"`, asserting
the mock's canned reply comes back. Runs after both the initial install
and each update route (hermes update / installer re-run).
`-z`/`--oneshot` is a comparatively recent CLI surface, so probe the
installed binary's --help before relying on it (oneshot_supported()),
same pattern as the existing update_supports()/installer_supports()
probes -- older sampled releases skip the round-trip with a stated
reason instead of failing.
The sandbox's stage2-run.sh forces HTTP_PROXY/HTTPS_PROXY at everything
so install.sh's real network calls ride the fake-internet proxy; the
inner script explicitly unsets them and sets NO_PROXY for localhost so
the plain HTTP call to our own mock server doesn't get routed into
proxy.py, which was never built to relay it.
The OpenAI-compatible mock server used by the desktop E2E suite
(apps/desktop/e2e/mock-server.ts) was hand-duplicated into
apps/desktop/scripts/dev-mock.mjs for local dev, which had already
drifted (dev-mock.mjs lacked scripted tool-call turns and had its own
copy of the SSE framing).
Extract the dependency-free core (models list, chat completions,
streaming/non-streaming, canned reply, prompt capture) into
scripts/mock-inference-server.mjs, runnable standalone via Node
(no npm install needed — only node:http/fs) or importable as a module.
It exposes an onCompletionRequest hook so callers can layer scripted
multi-turn tool-call sequences on top without forking the core.
apps/desktop/e2e/mock-server.ts now wraps the shared module, keeping
only the desktop-specific scripted turns (interim messages, sidebar
states, queue-stop, correction-switch, verification-stop, blocking
clarify) and all existing exports. apps/desktop/scripts/dev-mock.mjs
now imports the shared module directly instead of carrying its own
copy.
This makes the mock server reusable outside apps/desktop, e.g. from
tests/install/install-update-e2e.sh for a real CLI chat round-trip.
Timing out a bounded git probe must not leave helper descendants
(credential helpers, git-remote-https, hook children) running after the
probe fails open. bounded_git_probe now spawns the child in its own
process group on POSIX (process_group=0), and _kill_git_process_tree
signals the whole group with os.killpg — gated on the child actually
leading its own group (pgid == pid), so a shared-group spawn can never
take down unrelated processes. Windows keeps the existing taskkill /T /F
tree kill.
Proven live on main: a fake git that forks a 300s descendant left the
descendant running after the probe timeout; with the fix the descendant
dies with the launcher. Fast path and fail-open contract unchanged.
Port of openai/codex#36793 (Terminate timed-out Git process trees).
Wires tests/install/install-update-e2e.sh into CI as a reusable workflow plus a
caller that fans out over real releases, because that is the question users care
about: can someone on a version they actually installed get to this commit?
install-e2e-run.yml takes `route` and `install-ref`, so the combinations that
matter are expressible without duplicating runner setup. Each leg is independent
-- its own runner, its own sandbox, its own install, nothing shared or rewound.
The starting versions are chosen at runtime by scripts/sandbox/pick-release-tags.sh:
newest, oldest, and an evenly spaced spread between (5 by default). Choosing at
runtime rather than hardcoding keeps the matrix honest -- a pinned list stops
covering the newest release the day after it ships, and pins an "oldest" long
after anyone still runs it. Newest catches "did the last release break
updating?", oldest is the longest upgrade jump still possible, and the spread
samples the migrations in between (config-schema bumps, venv layout changes,
dependency floors). Tags are read from the checkout with `git tag --list`, not
`git ls-remote`: the job has the repository already, so this needs no network,
works offline and on a fork, and takes 8ms. The repo is derived from the
script's own resolved path rather than $PWD, so a copy cannot silently report a
different checkout's tags. The pick-releases job takes the checkout that suits
it -- blob:none filter, sparse-checkout of just that script, and fetch-tags,
since tags are the entire input and the default shallow checkout has none.
Triggers match the shape of the work:
* every 12 hours, so upstream drift (a new uv, a Node bump, a PyPI change)
surfaces on a schedule instead of in someone's review cycle;
* on release tags, the moment the set of versions users can update FROM
changes and the moment a broken updater would strand them;
* manually, with the route and the sample size as inputs.
Not on pull_request: a leg is ~9 minutes of real toolchain installation and the
matrix multiplies it. fail-fast is off so one broken release does not mask the
others, and max-parallel caps the fan-out so a run does not hammer the runners
or PyPI. The tag list is resolved once and shared by both route matrices, so the
two routes cover the same versions.
Artifact names include the sanitized install-ref, since a matrix runs the
reusable workflow several times per route and same-named artifacts collide; that
name is built in a step because Actions expressions have no string-replace
function. The name step runs with `if: always()`, since a failing leg is exactly
when its logs are wanted.
.gitignore covers .hermes-sandbox-e2e*/ rather than the bare directory: the
per-route sandbox trees (-update, -installer) fell outside it, so the sandbox
made the worktree dirty and dev-sandbox reacted by snapshotting the working copy
into a fresh fake-main commit on every invocation.
Nothing covered the update path, which is the worst thing to break: a broken
updater strands users on the version that cannot fix itself. `hermes update`
alone is ~2000 lines (hermes_cli/update_cmd.py) and had no end-to-end test.
tests/install/install-update-e2e.sh installs a genuine earlier Hermes through
the real one-liner (curl -fsSL https://…/install.sh | bash, served by
dev-sandbox's MITM proxy at the canonical URL, cloning "github.com" through the
upload-pack shim), which really installs uv, a managed Python, Node and the
venv. It then applies ONE update route and requires the checkout to land on this
commit with `hermes --version` still working -- so a pass means the venv and
entry point survived, not merely that git moved.
One route per run, each on a sandbox built from scratch. Sharing one install
across routes -- or rewinding with `git reset --hard` between them -- leaves the
second route running against a tree the first already updated (same venv, same
console script, same __pycache__), which is not the state any real user is in: a
route could pass only because its predecessor did the work, and a failure in the
first left the second exercising something undefined.
--install-ref chooses what to install first, so this covers "update from an
older release", not just from the tip. Installer flags are probed against the
target rather than assumed, because releases from months back predate flags
current Hermes takes for granted: --skip-browser is read out of that ref's own
install.sh, and `--yes` is asked of the installed `hermes update --help` (the
update subcommand has lived in main.py, subcommands/update.py and update_cmd.py
across the tags we sample, so a static parse rots silently -- and did). Without
those probes, old releases die on "Unknown option: --skip-browser" and
"unrecognized arguments: --yes" before doing any work.
Installer output is streamed through tee rather than captured: a real install of
uv, Python, Node and the venv IS the substance of this test, so it belongs in
the job log, not only in an artifact. pipefail keeps the installer's exit status
rather than tee's, so a failed install cannot look like a pass. The sandbox's
own proxy log is printed in full on failure, since a rejected TLS handshake
explains a failure that otherwise reads as a bare `curl: (35)`.
Deliberately reuses dev-sandbox rather than adding a second harness. An earlier
draft rewrote install.sh's hardcoded URLs with insteadOf and ran it against the
host; that tested the installer LESS faithfully (bash install.sh instead of the
real one-liner, host libs instead of a clean machine, ssh disabled to keep a
failed rewrite from reaching real GitHub) while duplicating a fake Internet we
already have.
Shell, not pytest, so scripts/run_tests.sh and run_tests_parallel.py stay
untouched: a pytest version needed an entry in the former's `env -i` credential
allowlist and a _SKIP_PARTS exclusion in the latter, and every meaningful line
was a command run inside the sandbox anyway.
Two guards, both earned during bring-up. It prefers the `sandbox` wrapper and
falls back to the raw script only when bwrap is on PATH (under Nix the wrapper
supplies the PATH and DEV_SANDBOX_* vars, so the bare script exits 127). And it
refuses to run on a dirty worktree: every dev-sandbox invocation re-derives fake
main from the working copy, so uncommitted changes move the update target
between the call that installs and the call that verifies -- a failure that
looks like a broken updater but is a moving reference.
allow you to simulate the whole official curl | bash installer,
and subsequent hermes updates.
Run development commands in a bubblewrap filesystem and network sandbox
with a local HTTPS MITM fixture server and a fake github
git-upload-pack transport.
Package the sandbox command and expose it from the nix devShell.
Stage the local installer at its canonical fake HTTPS URL and add a
persistent installation/update test path. Route root installs through
sandbox-owned filesystem locations and snapshot dirty source worktrees
into temporary fake commits so update tests can fast-forward without
changing the real checkout.
Includes a --install-ref sandbox installer mode that fetches any commit
(--from-main is a nice shorthand for local development) outside the
sealed sandbox, installs from that snapshot, and then promotes the fake
remote to the current worktree so update flows can be exercised with FF.
Notes on non-root sandboxes:
Giving a non-root sandbox a network is tricky.
slirp4netns joins the target userns and setuids to root before configuring the
netns, so the userns must map a uid 0; bwrap's --unshare-user maps exactly ONE
uid, so --uid 1000 leaves no root to become and slirp diedswith
`setns(CLONE_NEWNET): Operation not permitted`. Stage 1 builds the user+net
namespaces with `unshare` and two one-id ranges:
inner 0 -> a subuid, unused by the payload, present only so slirp can
become root
inner 1000 -> our real host uid
Mapping the payload to the *host* uid (not a second subuid) keeps everything the
sandbox writes owned by us, so `rm -rf` on a persistent sandbox still needs no
privileges. Stage 2 execs bwrap WITHOUT --unshare-user -- it only adds mount/pid
-- sidestepping bwrap's refusal to accept --uid outside a userns it created.
Costs a /etc/subuid range for the invoking user (we error with the exact line to
add) and util-linux `unshare`; `--root` needs neither.
Selecting an NVIDIA NIM model whose id reached config without the nvidia/
prefix produced a bare "HTTP 404: 404 page not found" — retried three times,
never naming the model. It reads exactly like an outage or an auth failure,
which is where the Discord thread spent its time before the id was spotted.
normalize_model_for_provider() had no branch for nvidia, so a bare id passed
straight through to the API. Repair it from the provider's curated catalogue:
a bare name that matches exactly one entry modulo the prefix gets it back.
That's a lookup, not a guess — build.nvidia.com also fronts local NIM
containers and third-party models, and anything absent from the catalogue is
left alone. Because the repair runs on every runtime setup, an already-broken
config self-heals on the next turn and prints what it changed.
If a bare id still reaches the wire, the 404 now explains itself. The
classifier consults the same catalogue: a prefix-less id the provider only
serves as vendor/model is a deterministic failure, so it classifies as
model_not_found instead of burning three retries on a retryable "unknown",
and the error trace names the id to use.
Fixes#78796
Selecting an NVIDIA NIM model whose id reached config without the nvidia/
prefix produced a bare "HTTP 404: 404 page not found" — retried three times,
never naming the model. It reads exactly like an outage or an auth failure,
which is where the Discord thread spent its time before the id was spotted.
normalize_model_for_provider() had no branch for nvidia, so a bare id passed
straight through to the API. Repair it from the provider's curated catalogue:
a bare name that matches exactly one entry modulo the prefix gets it back.
That's a lookup, not a guess — build.nvidia.com also fronts local NIM
containers and third-party models, and anything absent from the catalogue is
left alone. Because the repair runs on every runtime setup, an already-broken
config self-heals on the next turn and prints what it changed.
If a bare id still reaches the wire, the 404 now explains itself. The
classifier consults the same catalogue: a prefix-less id the provider only
serves as vendor/model is a deterministic failure, so it classifies as
model_not_found instead of burning three retries on a retryable "unknown",
and the error trace names the id to use.
Fixes#78796
shutil.make_archive writes PAX with fractional-mtime records, which macOS
Archive Utility rejects ("Error 94 - Bad message") on double-click. Write
the profile archive with tarfile in GNU format instead: integer mtimes,
longlink for deep paths, extracts under Finder, bsdtar, and gnutar alike.
Verified against /usr/bin/tar (bsdtar) with >100-char member paths.
'Move to project' submenu in the session actions menu (kebab and right-click,
via the shared MenuKit) listing every project with a folder except the current
owner. Picking one calls session.workspace.move at the project root, mirrors
the new cwd/branch/root into the $sessions cache, and refreshes the tree so
the row hops immediately.
A session created in the wrong directory needs its cwd corrected after the
fact. session.cwd.set only reaches live runtime sessions, so cold rows were
stuck. The new RPC targets the persisted row by session_key, validates the
folder, and REPLACES the git branch/root identity (update_session_cwd grows a
replace_git_meta flag) so the project tree's grouping follows the move instead
of pinning the session under the project it left via a stale git_repo_root.
A live idle agent bound to the row is re-anchored through the runtime path;
a mid-turn session refuses with 'session busy'. Runs on the RPC pool — the
git probes are subprocesses.
Chrome's grammar on every zone tab strip: Shift-click ranges from the
anchor, ⌥-click (Ctrl-click off-Mac — ⌘ stays close, ⌃ stays the macOS
context menu) toggles, plain click collapses back to one tab. Selected
tabs wear an accent wash; dragging any of them carries the block — the
ghost chip counts it — into a strip slot, a zone edge, or a Shift-span,
so three tabs land in a new zone as one gesture.
movePanes/reorderPanesInGroup/mergeZonesWithPane now take a block of pane
ids in strip order: the lead pane decides the drop geometry (slot, split,
span-merge) and the rest stack in behind it, with the pressed tab fronting
at the destination. The tab-selection store holds the block (Chrome
grammar: toggle, anchor range, collapse on plain click), drag-session
carries it — every dragged tab dims, the insertion slot skips the whole
block, and a landed drop spends the selection while a deny-area release
keeps it for a retry.
Adds a last_read_at watermark to the sessions table so surfaces (CLI,
TUI, desktop) can badge unread conversations. Read state derives from
the watermark vs latest activity, so new messages flip a conversation
back to unread with zero writes on the message path. NULL means never
tracked, so shipping the column doesn't badge pre-existing history.
set_session_read() stamps the whole compression lineage, matching the
archive/pin semantics; list_sessions_rich() rows carry a derived
`unread` key. DB layer only — no surface exposes it yet.
* fix(discord): reject empty outbound messages
* test(discord): cover empty final reply backfill state
Missed-message backfill decides what to replay from discord_messages, so
a dropped final reply must be recorded as failed by the new guard the
same way the exception path records one — otherwise the reply is both
never sent and never retried.
Co-authored-by: Jony <619963502@qq.com>
* chore: map 619963502@qq.com to zyz619963502zyz for PR #73449 salvage
---------
Co-authored-by: Jony <619963502@qq.com>
Export stages desktop.json (skin + mode, bundled user-theme definitions,
rail color, layout tree) into the CLI's own profile archive; import applies
it, so the receiver gets the whole look as a ready-to-use profile. Doors:
Export/Import profile... in Cmd-K, an import button beside the rail's +,
and Export in each profile square's context menu. New selectSavePath IPC
(native save dialog); credentials never leave the machine (CLI filter).
/export [profile] [-o output.tar.gz] bundles a profile into the shareable
archive; /import <archive> [--name <name>] adopts one as a new profile
(wrapper alias created when safe). Registry-driven, cli_only, so the CLI
and TUI both pick them up in autocomplete and help.
export_profile() accepts extra_files (root-relative filename -> text) so a
caller can stage companion files into the archive; the desktop uses it for
desktop.json, its appearance/interface overlay, now part of the default
profile's export allow-list.
New routes wrapping the existing hermes profile export/import machinery:
- POST /api/profiles/{name}/export (extra_files + optional output path)
- POST /api/profiles/import (returns the bundled desktop overlay)
- GET /api/profiles/{name}/desktop-overlay
Paths cross the API, not bytes - the desktop's native dialogs and its
local/pooled backends share a filesystem.
Three fakes pin mark_exhausted_and_rotate's signature explicitly and broke on
the new argument. They now assert it rather than just tolerate it — the xAI
spending-limit case is exactly the billing-403 this fixes, so it should be
pinning `failure_reason == "billing"`.
The sole-credential cooldown sized the bench from the raw HTTP status, but
403 is overloaded: error_classifier maps OpenRouter's "key limit exceeded"
and xAI's spending-limit block to FailoverReason.billing, while an edge
throttle with the same status is transient. Only 402 was excluded from the
short cooldown, so a spent account on a single key retried every 60 seconds
and re-failed forever.
Thread the classified reason from recover_with_credential_pool through
mark_exhausted_and_rotate to _exhausted_ttl. Billing keeps the full bench
regardless of status; everything else transient still recovers in 60s. The
verdict is stored on the entry (_EXTRA_KEYS, so it persists to auth.json) —
without that a restart would re-read a bare 403 and downgrade the bench.
Tests: sole billing-403 stays benched, survives reload, unclassified 403
still recovers; call-site coverage that the reason actually reaches the pool.
Three existing kwargs assertions updated for the new argument.
next_available_at() was computing the full 1-hour TTL for a sole
credential on a 429, contradicting the 60s cooldown in _available_entries.
The fallback restore gate (agent_runtime_helpers) uses next_available_at
to decide when to switch back from fallback to primary — so the agent
stayed on fallback for an hour instead of ~60s.
Add sole_credential computation in next_available_at mirroring
_available_entries, and a test verifying the short cooldown propagates.
A pool with only one usable (non-DEAD) credential has nothing to rotate to.
On a transient throttle (429 rate-limit, 403 edge-throttle, 5xx) the offending
key was benched for a full hour (EXHAUSTED_TTL_429/DEFAULT), so single-key /
no-fallback setups got an hour of hard failures for a throttle that resets in
seconds. The pool already special-cases 401 to recover quickly for single-key
setups; extend that to transient throttles when the credential is the sole
non-DEAD entry. 402 (billing/quota) keeps the full bench — a quick retry can't
help. Provider-supplied reset_at still overrides.
Adds tests covering sole 429/403 recovery, 402 full-bench, and multi-key
(no early recovery).
hermes debug share runs on the backend. A desktop app connected to a
remote, docker, or SSH backend writes desktop.log on the client machine,
so the bundle can never contain it — and the report rendered that as a
bare "(file not found)", which reads as "the app logged nothing" and
sends triage after a client-side bug it cannot see.
Name the writer and the path to collect by hand. Backend-written logs
are unchanged, a present desktop.log is still captured, and an empty one
still reports "(file empty)" — the app ran and logged nothing is a
different fact from the file being on another host.
* fix(agent): adopt .env credential/base-url edits at the turn boundary
A Settings save (desktop PUT /api/env, hermes setup) updates .env and
the saving process's os.environ, but a live session worker keeps the
base_url/api_key captured at agent init until restart — an open chat
silently kept calling the old endpoint (e.g. a local-server key sent to
api.openai.com, failing with an opaque 401).
Add AIAgent._try_refresh_env_client_credentials(), called at the start
of each conversation turn: re-resolve the provider's env-sourced
credentials (load_env() is mtime-memoized, so an unchanged file costs
one stat()) and rebuild the client via the existing
_replace_primary_openai_client machinery when the user edited them.
The refresh reacts only to env edits — resolved values changed since
the last look — never to mere divergence from the agent's current
values: credential-pool rotation and failover legitimately move the
session off the env credential, and stomping those back would flap.
Config model.base_url / pool custom endpoints keep precedence: edits
are only adopted while the session still runs on the registry default
or the previously-seen env value.
Lift _get_env_prefer_dotenv out of _seed_from_env to module level
(get_env_prefer_dotenv) so both the pool seeder and the per-turn
refresh share the same .env-over-os.environ resolution, including the
op:// indirection handling.
Fixes#67821
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* fix(agent): address sweeper review on env credential refresh
- Cover named custom providers (#67935): provider="custom" has no
PROVIDER_REGISTRY entry, so resolve the config block's key_env through
the same lookup the runtime resolver uses.
- Make the edit baseline transactional: a failed client rebuild rolls the
agent back and leaves _env_creds_seen un-advanced so the unchanged edit
is retried next turn.
- Recompute route-derived TLS material and default headers on a base-url
change, via a _reapply_route_client_config helper shared with
credential-pool rotation so the two paths cannot drift.
- Rebase onto main: get_env_prefer_dotenv keeps the scoped _get_secret
semantics from the profile-isolation fix (no raw os.environ reads).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
* chore: map jskang@lablup.com to rapsealk
---------
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
Co-authored-by: Brooklyn Nicholson <brooklyn.bb.nicholson@gmail.com>
* feat(desktop): expose native OS notifications to plugins via ctx.notifyNative
Desktop plugins can toast in-app (host.notify) but have no sanctioned way to
reach the OS notification pipeline the app's own approval/turn alerts use, so
a plugin surfacing a genuinely notable background event (e.g. a discovery
plugin finding a match) stays invisible once the user steps away from Hermes.
Add a curated per-plugin door instead of exporting the raw dispatcher:
- ctx.notifyNative({ title, body?, silent? }) on PluginContext — attributed
to the plugin id, routed through dispatchNativeNotification so every
existing gate applies (master + per-kind prefs, post-connect baseline,
away-from-app gating, throttle).
- New 'plugin' native-notification kind with its own Settings ▸ Notifications
toggle (default on), so users silence plugins without losing app alerts.
- New optional `tag` discriminator on the notify payload keys the renderer
throttle and main-process cross-window dedupe per plugin, so two plugins
can't collapse each other's session-less notifications.
Consumer: the Index Network desktop plugin wants background opportunity
alerts; anything in ~/.hermes/desktop-plugins gets the same door.
* feat(desktop): ctx.os — the curated OS door for plugins
Fold ctx.notifyNative into a ctx.os namespace so every way a plugin
reaches outside the app window lives behind one attributed door instead
of accreting one top-level ctx method per capability:
- ctx.os.notify — the native-notification door from the previous commit,
unchanged semantics (plugin kind pref, away-gating, per-plugin throttle).
- ctx.os.openExternal / ctx.os.revealPath / ctx.os.writeClipboard — the
existing window.hermesDesktop bridge capabilities, now sanctioned and
result-shaped: each resolves false (never throws) when the bridge or
member is missing, so a plugin branches on the result instead of
sniffing the preload surface or crashing on an older shell.
No new Electron surface: everything routes through bridge members the
app already ships; the notification path keeps every existing gate.
---------
Co-authored-by: seref <1573640+serefyarar@users.noreply.github.com>
* fix(credential-pool): clear exhaustion state on key rotation
When a user rotates an API key (e.g. via `hermes setup` after hitting a
rate limit), _upsert_entry updates the access_token on the existing pool
entry but preserves the stale last_status=exhausted from the old key.
On the next session the pool finds the entry, sees it exhausted, and
returns no usable credentials — even though the new key is valid.
Fix: when access_token changes on an existing entry, reset last_status,
last_error_code, last_error_reason, last_error_message, and
last_error_reset_at. The exhaustion state belongs to the old key, not
the new one.
* chore: add pasevin@gmail.com to AUTHOR_MAP
* fix: clear last_status_at on key rotation, remove unused pytest import
Address review feedback from teknium1 on PR #22622:
- Add last_status_at=None to the reset block (matches all other
token-sync reset paths in credential_pool.py)
- Assert last_status_at is None in the regression test
- Remove unused pytest import flagged by ruff + ty
Fold ctx.notifyNative into a ctx.os namespace so every way a plugin
reaches outside the app window lives behind one attributed door instead
of accreting one top-level ctx method per capability:
- ctx.os.notify — the native-notification door from the previous commit,
unchanged semantics (plugin kind pref, away-gating, per-plugin throttle).
- ctx.os.openExternal / ctx.os.revealPath / ctx.os.writeClipboard — the
existing window.hermesDesktop bridge capabilities, now sanctioned and
result-shaped: each resolves false (never throws) when the bridge or
member is missing, so a plugin branches on the result instead of
sniffing the preload surface or crashing on an older shell.
No new Electron surface: everything routes through bridge members the
app already ships; the notification path keeps every existing gate.
Show earlier spends the already-materialized DOM budget first and only asks the
session store for another page once that is exhausted, so the click stays cheap
and the store window stays as small as it can be.
Paging has no ceiling: each expand grows the window by one budget page until the
whole transcript is loaded. Branch persistence stays wired throughout —
setMessages is never dropped, so switchToBranch and applyBranchVisibility keep
working on a windowed session.
Co-authored-by: HexLab <8422520+HexLab98@users.noreply.github.com>
An oversized session rebuilt an unbounded runtime repository on every store
update and exhausted the renderer's V8 heap, crash-looping the window. The DOM
budget in thread/list.tsx bounds what PAINTS, but every message was still
normalized into the repository first, so a session only had to be heavy — not
visible — to kill the renderer.
selectTranscriptWindow keeps the tail that fits one render-weight page. Weight,
not message count: measured against a real 1,175-session store, a 400-message
cap disengages on 37 sessions that are heavy but short (one is 133 messages /
1.05MB) while firing on 92 long-but-light sessions that were never at risk.
The cut aligns off branch-group boundaries. useRuntimeMessageRepository records
a group's fork point the first time it sees the group, so a window starting
mid-group would re-parent the surviving branches to whatever happened to
precede them.
Co-authored-by: HexLab <8422520+HexLab98@users.noreply.github.com>
messageRenderWeight moves out of thread/list.tsx into lib/render-weight.ts.
The DOM page budget already spends render cost rather than message count —
the store window added next needs the same currency, and one weight function
keeps the two layers from drifting apart.
No behavior change.
Desktop plugins can toast in-app (host.notify) but have no sanctioned way to
reach the OS notification pipeline the app's own approval/turn alerts use, so
a plugin surfacing a genuinely notable background event (e.g. a discovery
plugin finding a match) stays invisible once the user steps away from Hermes.
Add a curated per-plugin door instead of exporting the raw dispatcher:
- ctx.notifyNative({ title, body?, silent? }) on PluginContext — attributed
to the plugin id, routed through dispatchNativeNotification so every
existing gate applies (master + per-kind prefs, post-connect baseline,
away-from-app gating, throttle).
- New 'plugin' native-notification kind with its own Settings ▸ Notifications
toggle (default on), so users silence plugins without losing app alerts.
- New optional `tag` discriminator on the notify payload keys the renderer
throttle and main-process cross-window dedupe per plugin, so two plugins
can't collapse each other's session-less notifications.
Consumer: the Index Network desktop plugin wants background opportunity
alerts; anything in ~/.hermes/desktop-plugins gets the same door.
Hitting ⌘1 (or cycling ⌃Tab onto the main tab) while Capabilities /
Messaging / Artifacts covered the workspace looked dead: the workspace
pane was already the zone's active tab behind the page, so fronting it
changed nothing on screen.
activateTreeTabSlot / cycleTreeTabInFocusedZone now return the activated
pane id, and the keybind handlers route back to the loaded session (or
the new-chat draft) when the landing pane is the workspace under a full
page — the same rule openSession already applies.
hermes model saves custom_providers models: {default: {context_length}} for
local Ollama. That dict shape was treated as an explicit catalog, so no-key
endpoints skipped live /v1/models probing and Desktop/Telegram only showed
the saved default — Refresh could not help. Keep list/string shapes as
allowlists; pin dict catalogs with discover_models: false.
Drop the manual web.search_backend / web.backend config-reading block
that duplicated _read_config_key in web_search_registry.py. The function
now delegates directly to get_active_search_provider() (which reads the
same config keys via the registry's canonical resolver) and falls back
to _get_search_backend() only when the registry has no providers loaded.
Also updates the TestXaiWebSearchBackendPreference tests to monkeypatch
the registry instead of load_config_readonly, and adds two new tests for
the legacy fallback path (no provider registered -> _get_search_backend).
Lock in backend preference, wire-name aliasing, and normalize mapping
so configured non-xai search providers stay on the Hermes client path.
Also init conflict-recovery generation on the telegram bare-adapter
helper so CI polling progress tests do not AttributeError.
ethernet
Teknium
brooklyn!
Jeffrey Quesnelle
kshitij
A.Alzaro
Jeongseok Kang
Aleksandr Pasevin
seref
HexLab98
Bryan Bednarski
xxxigm