Three call sites hardcoded Path("/app/data/cookbook_state.json"), which only
exists in Docker; on a native run the real path is <repo>/data, so the state
file looked missing and cookbook serve-state was silently ignored. Two others
used os.environ.get("DATA_DIR", "data") (a relative fallback, since DATA_DIR is
never set as an env var). Route all five through core.constants.DATA_DIR so the
path is consistent and absolute on both Docker and native.
Part of #3331.
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
* fix(platform): add WSL compatibility functions and path translation
fix(cookbook): enhance model scan script to support additional HuggingFace cache paths
fix(hardware): improve cache key generation for remote SSH context
test(tests): add tests for WSL detection and path translation functionality
* fix(cookbook): prefer prebuilt wheels for llama-cpp-python and normalize package aliases
* fix: enable StrictHostKeyChecking in nvidia probe
refactor: consolidate ssh & powershell command execution to utility functions in core module
refactor: consolidate nvidia path candidates in to single variables in core module
tests: add tests for new utility functions
* fix: correct wrong variable name
* fix(skill-extractor): walk all brace candidates so stray braces in prose do not swallow valid JSON
The extractor sliced from the FIRST brace to the LAST brace to recover
JSON embedded in surrounding commentary. When the model emits stray
braces before the JSON object, the slice produces invalid JSON,
json.loads raises, and the exception is swallowed -- the skill is
silently lost.
Fix: walk each brace candidate left-to-right and attempt json.loads on
each slice. The first candidate that parses successfully wins. If none
parse, json.loads on the original text raises and the existing
JSONDecodeError handler logs and returns None as before.
Tested locally -- 8/8 tests passed:
tests/test_extract_skill_json_nonstring.py (2 passed)
tests/test_skill_extractor_rows.py (1 passed)
tests/test_search_content_extraction_parity.py (2 passed)
tests/test_deep_research_search_error.py (3 passed)
Closes#2199
* test(skill-extractor): add focused repro for stray-brace JSON recovery
* test(skill-extractor): add regression test for leading invalid-brace fragment
Addresses the remaining edge case from review: a response that *starts*
with a brace but the leading fragment isn't valid JSON (e.g.
'{not json}\n{"title": "Valid later", ...}') still needs to recover
the valid skill object that follows.
_extract_json_object (already on dev) handles this correctly — it tries
the whole de-fenced string first, then walks each '{' candidate left-to-
right regardless of whether the response begins with '{', so the leading
invalid fragment no longer short-circuits recovery of the real object.
Updates the comment at the call site to call this out explicitly and adds
a regression test covering exactly the scenario described in review.
PDF uploads are stored as markdown wrappers with pdf_source or pdf_form_source markers so the editor can preserve extracted text, form fields, and annotations. The library exposed that internal wrapper: auto-created PDF documents used the hashed storage filename as the title, and row/facet language reported markdown instead of pdf.
Derive chat-upload PDF titles from the original upload name, derive document-library display language from the PDF source marker for rows, filters, and facets, and keep markdown wrappers excluded from the markdown facet when they represent PDFs.
The expanded library card already renders PDF-backed documents through /api/document/{id}/render-pdf. Allow only that inline PDF preview endpoint to be framed by same-origin app pages while leaving normal routes on X-Frame-Options: DENY and frame-ancestors none.
Also tighten the existing PDF marker regression assertion so it matches the actual historical corruption signature instead of contradicting the preserved [Page 1 text]: marker.
Fixes#2468
#2753 made the agent loopback base port-configurable but only for
_COOKBOOK_BASE in tool_implementations. Several other in-process loopback
calls still hardcoded http://localhost:7000 and broke off port 7000:
cookbook_serve_lifecycle (model-endpoints x2, shell/exec), builtin_actions
(model/serve), task_routes (calendar x3), and the gallery/email calls in
tool_implementations.
Extract the resolution (ODYSSEUS_INTERNAL_BASE / APP_PORT / 7000 fallback,
127.0.0.1 to avoid IPv6 ambiguity) into core.constants.internal_api_base()
and route every call site through it. Rename the now-misnamed _COOKBOOK_BASE
to _INTERNAL_BASE since it serves gallery/email/calendar/serve too. Adds a
test for the resolver plus a regression guard against reintroducing the
literal.
Part of #2752.
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
* fix(security): harden untrusted_context_message against delimiter spoofing
Root cause: untrusted_context_message() did not sanitise content before
interpolating it into the <<<UNTRUSTED_SOURCE_DATA>>> / <<<END_UNTRUSTED_SOURCE_DATA>>>
delimited sandbox block. Malicious content embedding the literal delimiter
strings could prematurely close the sandbox and inject instructions that
the LLM treats as trusted.
Fix: add _escape_guard_markers() helper that replaces the guard marker
strings with structurally inert tokens (<<<_UNTRUSTED_DATA>>> and
<<<_END_UNTRUSTED_DATA>>>) before the content is wrapped. The function is
applied in untrusted_context_message() after casting content to str.
The existing ~13 call sites (chat_processor.py, agent_loop.py,
deep_research.py, chat_helpers.py, chat_routes.py) are unaffected because
they pass content through without inspecting the output delimiters.
Regression tests added in tests/test_prompt_security.py covering:
- _escape_guard_markers unit tests (open, close, both, benign passthrough)
- untrusted_context_message integration tests (delimiter spoofing
neutralisation, type coercion, None handling, metadata preservation)
Resolves#3056
* fix(security): sanitize label for newlines and guard markers
Addresses reviewer feedback on PR #3086:
- Normalize label: strip CR/LF to prevent pre-guard line injection
- Escape guard marker literals in label via _escape_guard_markers()
- Add regression tests for label-based newline injection, GUARD_OPEN
and GUARD_CLOSE in label, and exactly-one-structural-guard assertion
* fix(security): move Source label inside GUARD_OPEN block
The reviewer correctly identified that even after sanitizing the label,
any user-derived label text (e.g. `f"web page: {url}"`) still appeared
before GUARD_OPEN in the trusted framing zone, where the LLM treats it
as trusted instructions.
Fix: move the 'Source: {label}' line to inside the guarded block so
only the hardcoded UNTRUSTED_CONTEXT_HEADER sits before GUARD_OPEN.
The raw label is still kept in metadata["source"] for traceability.
_sanitize_label() and _escape_guard_markers() are kept for defence-in-
depth on the label stored inside the block.
Update test_label_newline_injection_is_blocked to assert no label-
derived instruction text appears before GUARD_OPEN (pre-guard zone is
now empty of any user-derived content).
* refactor(tests): replace local function copies in test_endpoint_resolver with real imports
The test file carried 9 verbatim copies of src/endpoint_resolver.py functions
to avoid import-pollution concerns, but these copies are a drift hazard — PR #3343
had to update both in parallel. Replace them with direct imports so future changes
to endpoint_resolver are automatically exercised by the test suite.
Also fixes _ollama_api_root in endpoint_resolver.py: the bare-URL Ollama case
(e.g. http://nas:11434 with empty path) was already handled correctly in the test
copy but was missing from the real function, which would return /chat instead of
/api/chat for native Ollama endpoints without an explicit /api prefix.
Closes#3351
* refactor: import _ollama_api_root from llm_core instead of duplicating it
endpoint_resolver already imports _detect_provider and _host_match from
llm_core. Add _ollama_api_root to that import and remove the local copy,
collapsing two implementations to one source of truth.
llm_core's version is a superset (also strips /api/chat|tags|generate paths),
and since normalize_base already removes those suffixes upstream the result
is identical for every input used here.
The document editor stores the AI-edit diff state (_diffModeActive,
_diffOldContent, _diffNewContent, _diffChunks) as a module-global
singleton bound to whatever document was active when the diff opened,
and every document shares one #doc-editor-textarea. When the active
document is switched while an unapproved diff is open, the stale diff
must be discarded first or a later exitDiffMode (tab switch /
Accept-Reject-All) flushes the old document's content into the new
active document and overwrites it (issue #2467).
Guard both paths that switch the active document for an AI update,
while activeDocId still points at the previously-active doc:
- handleDocUpdate(): a doc_update targeting a different document.
- streamDocOpen(): the AI streaming a NEW document — this runs first on
that path, so a guard only in handleDocUpdate would fire too late and
still overwrite the streamed document.
Both reuse the exact `if (_diffModeActive) exitDiffMode(true);` guard
switchToDoc() and enterDiffMode() already use.
Fixes#2467
GET /api/sessions fired full-table scans against sessions, documents, and
gallery_images on every call. Added DbSession.owner == user (line 265),
Document.owner == user (line 283), GalleryImage.owner == user (line 289),
and .limit(2000) to auto_sort_sessions (line 1013). All follow the existing
owner-scoping pattern at lines 700 and 1230. No behaviour change — the
response was already correct; this eliminates the over-fetch.
Long model names are truncated with ellipsis in two places with no way to see
the full name: the model-picker dropdown items and the chat-header model
indicator. Add a native title tooltip carrying the full name to both — the
dropdown item's name span (nameSpan.title = m.display) and the header label
(label.title = the full model id; empty for the 'Select model' placeholder).
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
parse_due_for_user only matched day-first format ('today at 3pm').
Time-first strings like '3pm today' or '11pm today' — which the tool
schema and tool_index both advertise as valid examples — fell through
all branches, hit dateutil or the legacy _parse_dt fallback, and in
many cases raised ValueError. do_manage_notes then stored the raw
string verbatim, and the ISO-only reminder scanner (action_ping_notes)
never fired the note.
Add a time-first regex branch immediately after the day-first branch
to handle '<time> today|tonight|tomorrow|tmrw|yesterday'. Existing
day-first parsing is unchanged.
Fixes#3302
* feat(platform): add support for Apple Silicon detection in platform compatibility
test(tests): enhance shell_routes tests for Apple Silicon compatibility
* fix issues with missing import
* fix: correct package name in package-lock.json and enhance package installation commands in shell_routes.py and cookbook.js
* feat: add Apfel startup and health checks on macOS
- bootstrap Apfel via Homebrew on arm64 macOS
- start `apfel --serve --port 11435` detached for Odysseus
- verify readiness via `/health`
- clean up the Apfel process on exit or Ctrl+C
* fix: duplicate variable declaration post-merge conflict
- Should fix `node` CI issues.
* fix: issues with the update status of the APFEL dependency.
- fixed by changing the main conditional that determines the update.
* Fix: Remove unnecessary whitespaces and formatting for the model_routes.py file.
* Fix: whitespace issues with the model_routes file
* Fix: Remove unnecessary whitespaces and formatting for the model_routes.py file. Final
* Fix: Fixed updates using PIP for APFEL instead of custom cmd
#2996 made context_compactor call resolve_endpoint('utility', owner=owner),
but the mock added by #2174 stubbed it as lambda which: ..., which rejects the
owner kwarg. Each PR passed alone; merged on dev the two compaction tests fail
with TypeError and the pytest job goes red. Widen the mock to lambda *a, **k.
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
routes.email_helpers._decode_header joined the runs from
email.header.decode_header() with " ". Those runs carry their own
surrounding whitespace (e.g. (b"Re: ", None)), and RFC 2047 §6.2 requires
the whitespace between two adjacent encoded-words to be dropped, so the
join produced a double space after an ASCII prefix ("Re: Jóse"), a
spurious space in "Name <addr>" senders, and a stray space between two
adjacent encoded-words ("Café 日本"). _decode_header backs the inbox list,
message read, search, and the background pollers, so the corruption hit
essentially every non-ASCII subject/sender.
Use email.header.make_header(...) for RFC-correct concatenation, keeping
the existing lossy per-part fallback for malformed/unknown MIME charsets
(make_header raises LookupError there) so the unknown-charset contract in
tests/test_email_decode_header.py still holds.
The sibling mcp_servers.email_server._decode_header was already fixed the
same way (commit 46999de); this brings the routes.email_helpers copy in
line, with regression coverage.
Supported by Claude Opus 4.8
Co-authored-by: SurprisedDuck <288741682+SurprisedDuck@users.noreply.github.com>
maybe_extract_skill() sliced the LLM response from the first '{' to the
last '}'. When a model emits a stray brace in prose before the real
object (e.g. "uses {placeholder} then {...}"), the slice starts at the
prose brace, json.loads fails, and a valid skill is silently dropped.
Factor parsing into _extract_json_object(), which tries the whole
(de-fenced) string first and then each '{' start position, returning the
first candidate that parses to a JSON object.
Adds tests/test_skill_extractor_json.py.
On summary LLM call failure, maybe_compact was returning system_msgs+recent
(dropping the older half) with was_compacted=False, misleading the caller into
thinking the list was unchanged. Return the original messages list unchanged so
no history is lost; the next trim_for_context call handles length if needed.
Fixes#2160
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
_fallback_memory_candidates matched both positive (prefer/like/love) and
negative (hate / do not like / don't like) sentiment verbs in one regex
alternation, then formatted every hit as "User prefers {X}.". So
"I hate cilantro" was stored as "User prefers cilantro." -- the inverse of
what the user said. These fallback facts are persisted to memory and later
re-injected into the model's context, so the inverted preference actively
misleads the assistant.
Capture the matched verb and branch on it: negatives become
"User dislikes {X}.", positives stay "User prefers {X}." (still filed under
the existing "preference" category).
Supported by Claude Opus 4.8
Co-authored-by: SurprisedDuck <288741682+SurprisedDuck@users.noreply.github.com>
The loop-breaker's runaway backstop counted per-tool-type call totals and
tripped whenever any tool was used >=15 times — treating 15+ DISTINCT calls
to one tool as a stuck loop. A real batch (e.g. "add these 18 birthdays to my
calendar" emits 18 distinct manage_calendar create_event calls in one round)
got flagged "calling manage_calendar over and over", the calls were discarded
(next round tools_sent=0), and 0 events were created.
Count IDENTICAL repeated call signatures instead (same tool AND args), via a
small, unit-testable _detect_runaway_call() helper. Genuine batches pass; a
model truly stuck repeating one call still trips the backstop. Adds a
regression test.
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
_ping_endpoint() probes the bare base URL for non-Ollama endpoints.
OpenAI-compatible servers like llama-swap return 404 on the /v1 prefix
but 200 on /v1/models, causing endpoints to appear offline despite being
fully functional.
Add a /models fallback when the base URL returns a non-auth 4xx.
Auth failures (401/403) are treated as definitive — probing /models
would just repeat the same rejection.
Fixes#3181
Co-authored-by: michaelxer <michaelxer@users.noreply.github.com>
* feat: assign folder='Tasks' to task sessions at creation
Task sessions (LLM, action, research) now set folder='Tasks' on their
DbSession row, matching the pattern used by the Assistant folder. This
enables sidebar lens filtering without changing existing session
behaviour.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* feat: add backfill script for task session folders
One-shot script to set folder='Tasks' on existing [Task]/[Research]
sessions that predate the folder assignment in task_scheduler.py.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* refactor: replace standalone backfill script with automatic migration
Convert scripts/backfill_task_folders.py into _migrate_backfill_task_folders()
in core/database.py, called from init_db(). The migration is idempotent (only
touches rows where folder IS NULL/empty) and runs automatically on upgrade,
so operators no longer need a manual step to tag pre-existing task sessions.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
AuthMiddleware is the outermost middleware, so a credential-less CORS preflight
(OPTIONS + Access-Control-Request-Method) was rejected with 401 before
CORSMiddleware could answer it. That blocks every cross-origin browser/WebView
client: the preflight fails, so the real request is never sent.
Let a genuine preflight through at the top of AuthMiddleware.dispatch via a pure,
unit-tested predicate (core.middleware.is_cors_preflight). Precise -- only
OPTIONS carrying Access-Control-Request-Method; a credentialed request is never
matched -- and no data access.
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Three IMAP connection leaks were recently fixed via try/finally
(#1325, #1330, #1423). This commit applies the same pattern to the
remaining callsites that still used inline logout-only cleanup.
routes/email_helpers.py:
- _fetch_sender_thread_context: conn was uninitialized when the outer
try/except returned early on connect failure, causing the finally
block to crash on conn.close()/conn.logout(). Merged the two
separate try blocks into one and added conn=None guard.
- _pre_retrieve_context: ctx_conn.logout() was inside the loop body
with no finally, so any exception in the folder/search loop leaked
the socket. Moved cleanup into a finally block with ctx_conn=None
guard.
mcp_servers/email_server.py:
- _list_emails: multiple inline conn.logout() calls on early-return
paths; exception between them leaked the socket. Wrapped in
try/finally.
- _read_email: same pattern — four separate logout() calls replaced
by a single finally block.
- _reply_to_email: logout() called before the error check, so an
exception in conn.select() leaked the socket. Wrapped in
try/finally.
- _download_attachment: same pattern as _reply_to_email.
Also adds tests/test_imap_leak_fixes.py with 9 regression tests (one
per function/failure-mode) that monkeypatch _imap_connect and assert
conn.logout() is called exactly once even when IMAP operations raise.
validate_caldav_url resolves and vets the initial host, but caldav's
niquests session follows 3xx redirects by default, so a validated public
URL can be redirected at request time to loopback/link-local/private
space, re-opening the SSRF the host check closes. The existing redirect
guard only covered the settings test-connection path.
Add a shared _build_dav_client helper that pins the session to zero
redirects (any 3xx then raises instead of silently following an
attacker-chosen Location), and route both the pull (_sync_blocking) and
write-back (_writeback_blocking) paths through it. Mirrors the
follow_redirects=False already used on the test-connection path.
Tests exercise the real DAVClient request path (a 302 toward an internal
host is refused, the sink is never contacted; the PROPFIND is asserted to
reach the public server first so the check can't pass vacuously), confirm
the helper disables redirects on the installed client, guard against a
raw DAVClient creeping back in, cover mixed public/internal DNS results
in both orderings, and add the resolves-to-no-usable-records fail-closed
branch.
* fix(security): add HSTS and Permissions-Policy headers to SecurityHeadersMiddleware
Strict-Transport-Security is sent only when the connection is HTTPS
(detected via request.url.scheme or X-Forwarded-Proto: https), so
plain-HTTP dev deployments behind a reverse proxy are unaffected.
Permissions-Policy disables camera, microphone, and geolocation APIs
unconditionally — Odysseus does not use them, and this prevents a
successful XSS from requesting browser-native sensor access.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix(security): scope Permissions-Policy microphone directive to same-origin
Reviewers on PR #3081 (alteixeira20, NubsCarson) flagged that
microphone=() blocks mic access for same-origin (self) too, breaking
Odysseus's own voice/STT flow (getUserMedia({audio: true}) in
static/js/voiceRecorder.js). Scope it to microphone=(self) so
third-party origins stay locked out while the app's own UI keeps mic
access; camera and geolocation remain fully disabled as unused.
Adds focused middleware tests covering HSTS scoping (HTTPS direct,
X-Forwarded-Proto, absent on plain HTTP) and the Permissions-Policy
same-origin microphone contract.
---------
Co-authored-by: Claude Sonnet 4.6 <noreply@anthropic.com>
* fix(webhooks): redact IPv6 addresses in sanitized error messages
sanitize_error() only stripped IPv4 literals, so a failed webhook
delivery to an internal IPv6 host (::1, fe80::/fc00:: ...) leaked the
address into Webhook.last_error, which is surfaced in the UI. The module
already treats internal IPv6 as sensitive (see _PRIVATE_NETWORKS and
src/url_safety.py); the scrubber just didn't keep up.
Add an IPv6 redaction pass covering bracketed, full 8-group, and
::-compressed forms. The pattern is scoped to leave clock times
("12:34:56"), MAC addresses, and C++ "::" tokens untouched, and the
::-branch uses a lookahead over a flat character class so there is no
nested quantifier to backtrack on (no ReDoS on long colon/hex runs).
Adds tests/test_webhook_sanitize_error_ipv6.py.
* webhook: validate IPv6 candidates with ipaddress, not a regex grammar
Per review on #3038: instead of hand-rolling the IPv6 grammar in a regex
(brittle, and easy to over-match colon-heavy text), use a loose regex to
find candidate tokens and let ipaddress.ip_address() decide. Only tokens
it parses as IPv6 are redacted, so the false-positive guards (clock times,
MACs, "std::vector") now come from the stdlib instead of a custom pattern.
This also covers cases the old pattern missed -- zone ids (fe80::1%eth0)
and IPv4-mapped addresses -- and no longer partially mangles invalid
colon strings (a 9-group token is preserved whole rather than losing its
first 8 groups). The bracketed branch is a single greedy class with no
X*:X* backtracking; verified ~1ms on 40k-char adversarial input.
Extends the test file with zone-id, IPv4-mapped, and invalid-token cases.
* webhook: redact bracketed/scoped/IPv4-mapped IPv6 as one unit
Review on #3038 found a few IP forms left partially redacted or malformed
by sanitize_error():
[fe80::1%eth0]:8080 -> [[redacted]]:8080
[::ffff:192.168.0.1]:8080 -> [[redacted][redacted]]:8080
::ffff:192.168.0.1 -> [redacted][redacted]
Two causes: the bracketed branch's character class dropped zone ids, so
scoped addresses fell through to the bare branch and left the brackets and
port behind; and the IPv4 pass ran first, stripping the embedded v4 of an
IPv4-mapped address so the v6 pass then redacted the "::ffff:" remnant
separately.
Fix:
- run the IP-candidate pass before the IPv4 pass, so IPv4-mapped forms are
matched and redacted whole
- match the full bracketed authority ([...] + optional %zone + :port) as a
single token, and redact a v4-or-v6 literal inside [ ] as one [redacted]
- extend the bare branch with a bounded (exactly-3) dotted-quad tail for
IPv4-mapped forms; exactly-3 so it can't swallow a partial suffix and
accidentally preserve an otherwise-valid address
Each form now collapses to a single [redacted]; the candidate finder stays
linear (~1.3ms on 40k-char adversarial input). Adds regression tests for
the three reported forms and keeps the timestamp/MAC/std::vector coverage.
* fix(security): close DNS-rebinding hole on diffusion_server
scripts/diffusion_server.py used to ship `allow_origins=["*"]` with the
default `--host=127.0.0.1` bind. Combined, that left the OpenAI-compatible
image API reachable from any browser tab via DNS-rebinding: an attacker page
resolves its own domain to 127.0.0.1 mid-fetch, the browser forwards the
request to the loopback server, the server processes it (no Host check), and
the wildcard CORS reply lets the attacker page read the result + drive the
GPU. CWE-346 + CWE-942 + CWE-352 (DNS-rebinding bridge).
Fix:
- Drop the wildcard CORS at module load (default-deny).
- Install `TrustedHostMiddleware` with a loopback allowlist so DNS-rebound
requests are rejected by the middleware before any route runs.
- Add additive `--allowed-host` / `--allowed-origin` CLI flags so operators
who need browser access on a specific origin can opt in explicitly without
re-introducing the wildcard.
Tests: tests/test_diffusion_server_security.py (9 cases) pin the allowlist
helpers, the default-deny CORS behavior, and the live middleware paths via
Starlette's TestClient.
Detected by Aeon + semgrep + manual review.
Severity: medium.
CWE-346 / CWE-942 / CWE-352.
* test(diffusion-server): drive ASGI app via httpx, not TestClient portal
The TrustedHost/CORS integration tests used `with TestClient(app) as
client:`, whose context-manager form spins up an anyio blocking portal to
run the app lifespan. Under the repo's pytest setup (anyio plugin active, a
stray asyncio_mode option, no pytest-asyncio) that portal deadlocks —
`test_trusted_host_middleware_rejects_attacker_host` hung indefinitely in
review before emitting any assertion output.
Replace the TestClient usage with a tiny _asgi_get() helper that drives the
ASGI app over httpx.ASGITransport on a fresh event loop (asyncio.run). No
portal, no lifespan, no dependency on the host project's async test plugins.
Host is taken from the request URL so TrustedHostMiddleware sees the exact
hostname under test; Origin goes through headers. Assertions are unchanged.
Focused test now passes in 0.12s; full file 9 passed.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: aeonframework <aeonframework@users.noreply.github.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Alexandre Teixeira