Commit Graph

12 Commits

Author SHA1 Message Date
nesquena-hermes
711bb5a6c9 feat: real-time gateway session sync (Phase 1) (#274)
* feat: add real-time gateway session sync (Phase 1)

- Add gateway_watcher.py: background daemon polling state.db every 5s
  for gateway session changes (telegram, discord, slack, etc.)
- Extend get_cli_sessions() to include all non-webui sources
- Add SSE endpoint /api/sessions/gateway/stream for real-time push
- Add dynamic source badges (telegram=blue, discord=purple, slack=dark purple)
- Rename 'Show CLI sessions' to 'Show agent sessions'
- Wire watcher lifecycle into server start/stop
- 10 tests covering metadata, filtering, SSE, and watcher lifecycle
- Activated via the same checkbox as CLI session import

Addresses GitHub issue #272

* fix: SSE event name mismatch, TLS attribute, remove PLAN.md

- Fix critical SSE bug: frontend listened for 'gateway_session_update'
  but backend sends 'sessions_changed' -- events were silently dropped
- Fix frontend field check: data.changed -> data.sessions (matches
  the actual payload structure from gateway_watcher)
- Fix TLS: ssl.TLSv1_2 -> ssl.TLSVersion.TLSv1_2 (the bare attribute
  does not exist, would crash TLS setup and silently fall back to HTTP)
- Remove PLAN.md: implementation plan should not be committed to repo

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* fix: test isolation and slow-consumer sentinel in gateway sync

tests/test_gateway_sync.py:
- Fix _get_test_state_dir() path mismatch: the function was computing
  HERMES_HOME/webui-mvp-test but conftest.py sets HERMES_HOME=TEST_STATE_DIR,
  so state.db was written to a double-nested path the server never read.
  Now uses HERMES_WEBUI_STATE_DIR first (which conftest sets directly to
  TEST_STATE_DIR), fixing the 7/10 test failures in full-suite ordering.
- Fix conn cleanup: removed conn.close() from inside try blocks so the
  connection stays valid for _remove_test_sessions() in the finally block.
  Previously the closed conn caused ProgrammingError in finally (swallowed
  by bare except), leaving ghost sessions in state.db on test failure.

api/gateway_watcher.py:
- Fix slow-consumer queue eviction: when a subscriber queue fills (>10 events)
  and is removed from _subscribers, now puts a None sentinel into it so the
  SSE handler unblocks and closes the connection, letting EventSource
  auto-reconnect. Without this the connection stayed open but received no
  further events.

* fix: test isolation — set HERMES_WEBUI_TEST_STATE_DIR in conftest

The gateway sync tests write directly to state.db and must use the same
path the test server reads from.  Previously they computed the path
independently, which broke when test_auth_sessions.py set a different
HERMES_WEBUI_STATE_DIR in the test-process environment at import time.

tests/conftest.py:
- Set HERMES_WEBUI_TEST_STATE_DIR=TEST_STATE_DIR in the test process's
  os.environ (via setdefault) so gateway tests can read it reliably.
  Using setdefault preserves any explicit override the caller may pass.

tests/test_gateway_sync.py:
- Simplify _get_test_state_dir(): check HERMES_WEBUI_TEST_STATE_DIR first
  (now reliably set by conftest), fall back to HERMES_HOME/webui-mvp-test.
  Remove the workaround that tried to snapshot HERMES_HOME at import time.

Result: 658/658 tests pass in full-suite ordering (was 651 pass / 7 fail).

---------

Co-authored-by: bergeouss <bergeouss@users.noreply.github.com>
Co-authored-by: Nathan Esquenazi <nesquena@gmail.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-11 20:53:12 -07:00
nesquena-hermes
27c2fd6c08 v0.46.0: security, Docker UID/GID, model discovery, i18n, cancel fix
* fix: decode HTML entities before markdown processing + zh/zh-Hant translations (#239)

Adds decode() helper in renderMd() to fix double-escaping of HTML entities
from LLM output (e.g. &lt;code&gt; becoming &amp;lt;code&amp;gt; instead
of rendering). XSS-safe: decode runs before esc(), only 5 entity patterns.

Also adds 40+ missing zh (Simplified Chinese) translation keys and a new
zh-Hant (Traditional Chinese) locale with 163 keys.

Fix applied: removed duplicate settings_label_notifications key in both
zh and zh-Hant locales.

Fixes #240

* fix: restore custom model list discovery with config api key (#238)

get_available_models() now reads api_key from config.yaml before env vars:
  1. model.api_key
  2. providers.<active>.api_key / providers.custom.api_key
  3. env var fallbacks (HERMES_API_KEY, OPENAI_API_KEY, etc.)

Also adds OpenAI/Python User-Agent header and a regression test covering
authenticated /v1/models discovery.

Fixes users with LM Studio / Ollama custom endpoints configured in
config.yaml whose model picker silently collapsed to the default model.

* feat: Docker UID/GID matching to avoid root-owned .hermes files (#237)

Adds docker_init.bash with hermeswebuitoo/hermeswebui user pattern so
container files match the host user UID/GID. Prevents .hermes volume
mounts from being owned by root when using a non-root host user.

Configure via WANTED_UID and WANTED_GID env vars (default 1000/1000).
Readme updated with setup instructions.

Fix applied: removed duplicate WANTED_GID=1000 line in docker-compose.yml
that was overriding the ${GID:-1000} variable expansion.

* security: redact credentials from API responses and fix credential file permissions (#243)

Adds response-layer credential redaction to three endpoints:
  - GET /api/session — messages[], tool_calls[], and title
  - GET /api/session/export — download also redacted
  - SSE done event — session payload in stream
  - GET /api/memory — MEMORY.md and USER.md content

Adds api/startup.py with fix_credential_permissions() at server startup.
Adds 13 tests in tests/test_security_redaction.py.

Merged with #237 container detection changes in server.py.

* fix: cancel button now interrupts agent and cleans up UI state (#244)

Wires agent.interrupt() into cancel_stream() so the backend actually
stops tool execution when the user clicks Cancel, rather than only
stopping the SSE stream while the agent keeps running.

Changes:
  - api/config.py: adds AGENT_INSTANCES dict (stream_id -> AIAgent)
  - api/streaming.py: stores agent in AGENT_INSTANCES after creation,
    checks CANCEL_FLAGS immediately after store (race condition fix),
    calls agent.interrupt() in cancel_stream(), cleans up in finally block
  - static/boot.js: removes stale setStatus(cancelling) call
  - static/messages.js: setBusy(false)/setStatus('') unconditionally on cancel

Race condition fix: after storing agent in AGENT_INSTANCES, immediately
checks if CANCEL_FLAGS[stream_id] is already set (cancel arrived during
agent init) and interrupts before starting. Check is inside the same
STREAMS_LOCK acquisition, making it atomic.

New test file: tests/test_cancel_interrupt.py with 6 unit tests.

* docs: v0.46.0 release notes, bump version, update test counts

---------

Co-authored-by: Nathan Esquenazi <nesquena@gmail.com>
2026-04-11 10:17:52 -07:00
nesquena-hermes
ed9023a431 fix: wire auto_install_agent_deps into server.py startup (#216)
* fix: wire auto_install_agent_deps into server.py startup; add api/startup.py to ARCHITECTURE.md

* fix(tests): kill stale process on test port before server start in conftest

Stale servers left by QA harness runs (ports 8792/8793 etc.) or prior
test sessions could interfere with conftest starting its own server on
TEST_PORT (8788). If the port was already occupied, _wait_for_server
hit the wrong server and tests got unexpected 404s/500s, failing
non-deterministically — the 'conftest isolation issue' seen this session.

Fix: run fuser -k on TEST_PORT before launching the new server process,
with a 0.5s sleep for port release. The full suite now runs 571/571
reliably regardless of what other servers were previously active.

---------

Co-authored-by: Nathan Esquenazi <nesquena@gmail.com>
2026-04-10 00:56:07 -07:00
Cyprian Kowalczyk
011034dc71 feat: optional HTTPS/TLS support via cert and key env vars (#199)
Add optional HTTPS support controlled by two env vars:
  HERMES_WEBUI_TLS_CERT=/path/to/cert.pem
  HERMES_WEBUI_TLS_KEY=/path/to/key.pem

- Wraps server socket with ssl.SSLContext (min TLSv1.2)
- Dynamic scheme detection for startup messages (http:// vs https://)
- Graceful fallback to HTTP if cert loading fails — server never crashes
  due to bad TLS config, just prints a warning and continues
- Auth cookie Secure flag already set when HTTPS is detected via getpeercert
- 6 end-to-end tests: config flags, HTTPS handshake, HTTP still works,
  fallback on bad paths

Addresses #191 (HTTPS support issue).
2026-04-09 18:08:29 -07:00
Cyprian Kowalczyk
04678b7b6e feat(server): add 30s connection timeout to prevent slow-client thread exhaustion (#198)
Set Handler.timeout = 30. Python's BaseHTTPRequestHandler.setup()
calls self.request.settimeout(timeout), which raises socket.timeout
on idle or slow connections after the configured duration.

This defends against Slowloris-style attacks where a client holds
connections open indefinitely, exhausting threads in ThreadingHTTPServer.
Also recovers threads from crashed clients with hung TCP connections.

Addresses #194.
2026-04-09 18:05:18 -07:00
nesquena-hermes
a064542df9 release: v0.39.0 — security hardening, 12 fixes (#171)
* Security: harden auth, CSRF, SSRF, XSS, and env race conditions

Twelve fixes from a full security audit:

CRITICAL
- Add CSRF Origin/Referer validation on all POST endpoints
  (prevents cross-origin abuse of self-update, settings, file ops)

HIGH
- Unify password hashing: config.py now uses PBKDF2 (600k iters)
  instead of single-iteration SHA-256
- Add per-IP rate limiting on login (5 attempts/60s, 429 on excess)

MEDIUM
- Validate session IDs as hex-only before filesystem operations
  (prevents path traversal via crafted session ID)
- SSRF: resolve DNS before private-IP check in model fetching
  (prevents DNS rebinding to internal services)
- Warn loudly when binding non-loopback without password set
- SSE env var mutations: wrap sync chat + streaming restore in _ENV_LOCK
- Force Content-Disposition:attachment for HTML/XHTML/SVG uploads
  (prevents stored XSS via uploaded files)

LOW
- Extend HMAC session signature from 64 to 128 bits
- Add resolve()+relative_to() check on skills path construction
- Set Secure flag on session cookie when connection is HTTPS
- Sanitize exception messages to strip filesystem paths

No breaking changes. All fixes are backward-compatible.

* fix: use getattr for Secure cookie SSL detection

handler.request.getpeercert raises AttributeError on plain sockets
(non-SSL). Use getattr(..., None) to safely check for SSL.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

* tests: add sprint 29 security hardening coverage (PR #171)

33 tests covering all 12 security fixes:
- CSRF origin/referer validation
- Login rate limiting (5 attempts/60s)
- Session ID hex validation (path traversal prevention)
- Error path sanitization (_sanitize_error)
- Secure cookie getattr safety
- HMAC signature length (64->128 bit)
- Skills path traversal prevention
- Content-Disposition for HTML/SVG/XHTML
- PBKDF2 password hashing verification
- Non-loopback startup warning
- SSRF DNS guard code presence
- _ENV_LOCK export from streaming module

* release: v0.39.0 — security hardening, 12 fixes (#171)

---------

Co-authored-by: betamod <matthew.sloly@gmail.com>
Co-authored-by: Nathan Esquenazi <nesquena@gmail.com>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-07 22:26:03 -07:00
Nguyễn Công Thuận Huy
4d333acbbc chore: add missing type hints across 10 files 2026-04-05 13:30:20 +07:00
Varun Chopra
d05e15e612 fix: resolve pip packages from site-packages instead of agent dir
When `pip install --target .` is run inside the hermes-agent checkout,
third-party package directories (openai/, pydantic/, requests/, etc.)
end up alongside real Hermes source files. With the agent dir at the
front of sys.path (insert(0)), Python resolves imports from those local
directories, breaking whenever the host platform differs from the
container (e.g. macOS .so files inside a Linux image).

Fix: append agent dir to sys.path instead of prepending. This lets
site-packages resolve pip packages correctly while still allowing
Hermes-specific modules (run_agent, hermes/, etc.) to resolve since
they do not exist in site-packages.

Also improves verify_hermes_imports() to surface the actual exception
message in startup logs, making it much easier to diagnose why a
module failed to import.
2026-04-04 23:29:33 +05:30
Nathan Esquenazi
1b1cd124f6 fix: stop leaking stack traces to clients in HTTP 500 responses
Tracebacks exposed file paths, module names, and potentially secret
values from local variables. Now logged server-side only; clients
receive a generic error message.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-03 06:41:32 -07:00
Nathan Esquenazi
b8b62722ec feat: Sprint 19 — password auth, security headers, login page
Auth system (off by default, zero friction for localhost):
- New api/auth.py module: password hashing (SHA-256 + STATE_DIR salt),
  signed HMAC session cookies (24h TTL), auth middleware
- Enable via HERMES_WEBUI_PASSWORD env var or Settings panel
- Minimal dark-themed login page at /login (self-contained HTML)
- POST /api/auth/login, /api/auth/logout, GET /api/auth/status
- Settings panel: "Access Password" field + "Sign Out" button
- password_hash added to settings.json (null = auth disabled)

Security hardening:
- Security headers on all responses: X-Content-Type-Options: nosniff,
  X-Frame-Options: DENY, Referrer-Policy: same-origin
- POST body size limit: 20MB cap in read_body() to prevent DoS

Closes #23. 9 new tests. Total: 304 passed, 0 regressions.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
2026-04-03 05:53:26 -07:00
Hermes
7019c25021 Hermes Web UI — Sprints 11-14: multi-provider models, settings, session QoL, alerts, polish
Sprint 11 (v0.13): multi-provider model support, streaming smoothness
- Dynamic model dropdown populated from configured API keys (OpenAI, Anthropic,
  Google, DeepSeek, GLM, Kimi, MiniMax, OpenRouter, Nous Portal)
- Scroll pinning during streaming (no forced scroll when user has scrolled up)
- All route handlers extracted to api/routes.py (server.py now ~76 lines)

Sprint 12 (v0.14): settings panel, SSE reconnect, session QoL
- Settings panel (gear icon) -- persist default model and workspace server-side
- SSE auto-reconnect on network blips
- Pin/star sessions to top of sidebar
- Import session from JSON export

Sprint 13 (v0.15): cron alerts, background errors, session duplicate, tab title
- Cron completion alerts: toast per completion + unread badge on Tasks tab
- Background agent error banner when a non-active session errors mid-stream
- Session duplicate button
- Browser tab title reflects active session name

Sprint 14 (v0.16): Mermaid diagrams, file ops, session archive/tags, timestamps
- Mermaid diagram rendering inline (dark theme, lazy CDN load)
- File rename (double-click in file tree) and create folder
- Session archive (hide without deleting, toggle to show)
- Session tags -- #hashtag in title becomes colored chip + click-to-filter
- Message timestamps (HH:MM on hover, full date as tooltip)

Test suite: 224 tests across 14 sprint files + regression gate, 0 failures.
2026-03-31 07:02:47 +00:00
Nathan Esquenazi
a4e2174c29 Hermes WebUI v0.1.0 — initial public release 2026-03-30 20:40:19 -07:00