When a comment-triggered task resumes an existing session, the agent
may mistake the new comment for a previous one and skip it. Add [NEW
COMMENT] tag to the prompt and reinforce in AGENTS.md workflow that
the agent must respond to THIS specific comment, not prior ones.
When an agent CLI process hangs (e.g. a tool call blocks on unreachable
I/O), the daemon's scanner blocks indefinitely on stdout, preventing the
Result from ever being sent. This causes tasks to stay in "running"
state permanently with no further events.
Three-layer fix:
1. Agent backends (claude, opencode, openclaw, gemini): add a watchdog
goroutine that closes the stdout/stderr pipe when the context is
cancelled, forcing the scanner to unblock. Also set cmd.WaitDelay
so Go force-closes pipes after 10s if the process doesn't exit.
2. daemon executeAndDrain: add an independent drain timeout (backend
timeout + 30s buffer) with context-aware select on both the message
channel and the result channel, so the daemon never blocks forever.
3. daemon ping path: add context-aware select so pings don't deadlock
if the agent backend stalls.
Closes#925
Co-authored-by: Devv <devv@Devvs-Mac-mini.local>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(daemon): prevent duplicate runtime registration on profile switch
The daemon_id included a profile name suffix (e.g. "hostname-staging"),
so switching profiles created a new daemon_id that bypassed the UPSERT
dedup constraint, leaving orphaned runtime records in the database.
Three changes:
- Remove profile suffix from daemon_id — use stable hostname only.
The unique constraint (workspace_id, daemon_id, provider) already
prevents collisions within the same workspace.
- Auto-migrate agents from old offline runtimes to the newly registered
runtime during DaemonRegister (same workspace/provider/owner).
- Add TTL-based GC in the runtime sweeper to delete offline runtimes
with no active agents after 7 days.
Closes MUL-695
* fix(daemon): address code review issues on PR #906
1. Move gcRuntimes() to the main sweep loop — previously it was inside
sweepStaleRuntimes() after an early return, so it only ran when new
runtimes were marked stale. Now it runs every sweep cycle independently.
2. Fix DeleteStaleOfflineRuntimes to exclude runtimes with ANY agent
reference (not just active ones). The FK agent.runtime_id is ON DELETE
RESTRICT, so archived agents also block deletion.
3. Scope MigrateAgentsToRuntime to the same machine by matching
daemon_id LIKE '<current_daemon_id>-%'. This prevents cross-machine
agent migration when the same user has multiple devices.
When an agent is triggered via @mention (not as the issue assignee),
the generated CLAUDE.md had no explicit agent identity. The agent would
infer its identity from the issue's assignee field, causing it to skip
work intended for it.
Now CLAUDE.md always includes "You are: <agent-name> (ID: <agent-id>)"
so the agent knows exactly who it is regardless of the issue assignee.
Closes MUL-709
- cmd_daemon.go: use filepath.Join for PID/log file paths instead of string concat with "/"
- codex_home.go: use os.TempDir() instead of hardcoded "/tmp" for cross-platform fallback
Co-authored-by: Devv <devv@Devvs-Mac-mini.local>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
On Windows, os.Symlink requires Developer Mode or admin privileges.
Extract symlink creation into platform-specific files: on non-Windows,
behavior is unchanged (os.Symlink). On Windows, try os.Symlink first,
then fall back to directory junctions (mklink /J) for dirs and file
copy for files.
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
1. Security: add isBlockedEnvKey() blocklist that rejects MULTICA_*
prefix and critical system vars (HOME, PATH, USER, SHELL, TERM,
CODEX_HOME) from custom_env injection
2. Observability: log warnings when json.Unmarshal fails on custom_env
(agentToResponse + claim endpoint)
3. UX: use stable auto-increment IDs for env entry React keys instead
of array index to prevent input focus/state issues on add/remove
Add per-agent custom_env configuration that gets injected into the agent
subprocess at launch time. This enables users to configure custom API
endpoints (ANTHROPIC_BASE_URL), API keys (ANTHROPIC_API_KEY), and cloud
provider modes (CLAUDE_CODE_USE_BEDROCK, CLAUDE_CODE_USE_VERTEX) without
requiring code changes.
Changes:
- Migration 040: add custom_env JSONB column to agent table
- Backend: custom_env in agent CRUD API + claim endpoint
- Daemon: merge custom_env into subprocess environment variables
- Frontend: env var editor in agent settings (key-value pairs with
visibility toggle for sensitive values)
Closes#816
Related: #807, #809
Main introduced executeAndDrain/mergeUsage refactor. Resolve by keeping
main's refactored structure and re-applying EnvRoot to the switch/case
in runTask. Rename newTestDaemon → newGCTestDaemon to avoid collision
with the helper added in daemon_test.go on main.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The repoCache.Sync() call in loadWatchedWorkspaces runs synchronous git
clone/fetch operations that can take minutes for large repos. Because
heartbeatLoop and pollLoop only start after loadWatchedWorkspaces returns,
the runtime's last_seen_at is never updated during the sync, causing the
server's sweeper to mark it offline after 45 seconds.
Move repo cache sync to a background goroutine so heartbeat and poll
loops start immediately after runtime registration.
Closes#825
The context cancel in pruneRepoWorktrees was called explicitly after
CombinedOutput inside a loop. Extract to a helper method so defer
cancel() works correctly (scoped to the function, not the loop).
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
- Move WriteGCMeta from runTask() to handleTask() so it runs after
task completion, not at start. Mid-task crashes leave orphan dirs
that get cleaned by GCOrphanTTL.
- Strengthen isBareRepo to check both HEAD and objects/ directory.
- Remove empty workspace directories after all task dirs are cleaned.
- Add 30s context timeout to git worktree prune to prevent hangs.
- Add comprehensive unit tests for shouldCleanTaskDir (8 scenarios),
cleanTaskDir, gcWorkspace empty-dir cleanup, isBareRepo, and
WriteGCMeta/ReadGCMeta roundtrip.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(daemon): add fallback for failed session resume
When the daemon tries to resume a prior session (--resume flag for
Claude, --session for OpenCode, session/resume RPC for Hermes) and the
session no longer exists, the agent fails immediately. This adds a
fallback that retries the execution with a fresh session instead of
marking the task as blocked.
Extracts the execute+drain logic into a reusable executeAndDrain method
to avoid code duplication between the initial attempt and the retry.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(daemon): narrow session resume retry and merge usage
Address review feedback:
1. Narrow retry trigger: only retry when result.SessionID == "" (no
session was established), not on any failure with PriorSessionID set
2. Merge token usage from both attempts so billing is accurate
3. Log errors when the retry itself fails to start
4. Add unit tests for mergeUsage, fallback behavior, and no-retry
when session was already established
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Isolation directories accumulate indefinitely because they're preserved
for session reuse but never cleaned up after the issue is closed.
This adds a background GC loop that periodically scans local workspace
directories and removes those whose issue is done/canceled and hasn't
been updated for 5 days (configurable via MULTICA_GC_TTL). Orphan
directories with no metadata are cleaned after 30 days.
Changes:
- Write .gc_meta.json (issue_id, workspace_id) at task completion
- Add GET /api/daemon/issues/{issueId}/gc-check endpoint for status queries
- Add gcLoop goroutine to daemon with configurable interval/TTL
- Prune stale git worktree references from bare repo caches each cycle
- New env vars: MULTICA_GC_ENABLED, MULTICA_GC_INTERVAL, MULTICA_GC_TTL,
MULTICA_GC_ORPHAN_TTL
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
When a task is triggered by a comment, the agent prompt now includes
the comment content directly. This prevents the agent from ignoring
the comment when stale output files exist in a reused workdir.
Closes#805
Previously only Claude and Codex had log-scanning-level token usage
reporting (Flow B). This adds scanners for the remaining three runtimes:
- OpenCode: reads JSON message files from ~/.local/share/opencode/storage/message/
- OpenClaw: reads JSONL session files from ~/.openclaw/agents/*/sessions/
- Hermes: reads JSONL session files from ~/.hermes/sessions/
All three are registered in Scanner.Scan() and follow the same
(date, provider, model) aggregation pattern as existing scanners.
Allow agents to pipe comment content through stdin instead of the
--content flag, avoiding shell escaping issues with backticks, quotes,
and other special characters in markdown content.
Usage: cat <<'COMMENT' | multica issue comment add <id> --content-stdin
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Codex tasks running in workspace-write sandbox mode could not resolve
api.multica.ai because the hardcoded sandbox parameter in thread/start
overrode any config.toml settings, and the default sandbox policy blocks
network access.
Changes:
- Remove hardcoded `sandbox: "workspace-write"` from thread/start RPC —
let Codex read sandbox config from its own config.toml instead
- Auto-generate config.toml in per-task CODEX_HOME with
`sandbox_mode = "workspace-write"` and `network_access = true`,
preserving any existing user settings
- Fix Reuse() to restore CodexHome for Codex provider on workdir reuse
Closes#368
Registers `gemini` as a sixth supported agent provider alongside claude,
codex, opencode, openclaw, and hermes.
- Daemon config probes for `gemini` on PATH (MULTICA_GEMINI_PATH /
MULTICA_GEMINI_MODEL env overrides mirror the other providers).
- New agent.geminiBackend in pkg/agent/gemini.go: spawns
`gemini -p <prompt> --yolo -o text [-m <model>] [-r <session>]`,
reads stdout to completion, and returns a single MessageText plus
the standard Result struct (Status / Output / DurationMs).
- Execution environment writes a GEMINI.md file into the task workdir
(mirroring the existing CLAUDE.md / AGENTS.md injection for other
providers) so Gemini discovers the Multica runtime meta-skill
through its native mechanism.
Tests:
- pkg/agent/gemini_test.go — unit coverage for buildGeminiArgs
(baseline, model override, resume session, omit-when-empty).
- internal/daemon/execenv/TestInjectRuntimeConfigGemini — verifies
GEMINI.md is written and that CLAUDE.md/AGENTS.md are NOT.
Scope (intentional for v1):
- Text output only (`-o text`). Streaming tool events via
`--output-format stream-json` is a follow-up once we have a
reliable reproduction of Gemini's event schema.
- No MCP config plumbing. Gemini's `--allowed-mcp-server-names`
filter pairs well with the per-agent MCP work on feat/per-agent-mcp;
stacking the two can land as a follow-up.
- No token usage scraping (Gemini's accounting lives on the Google
Cloud side, not a local JSONL log like claude/codex).
- No session resume wiring beyond accepting the ExecOptions field —
the daemon does not yet persist Gemini session IDs because the text
output mode does not expose them.
Migration / env changes:
- New optional environment variables MULTICA_GEMINI_PATH and
MULTICA_GEMINI_MODEL. Default path is the string "gemini" (resolved
via PATH at daemon startup). If no Gemini install is detected, the
provider is simply absent from the runtime — no behavior change for
existing deployments.
Replaces the hardcoded assignment-triggered workflow in buildMetaSkillContent()
with a minimal version that defers to agent Skills and Identity. Keeps platform
capability docs and status management steps intact.
Fixes#669
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
* feat(daemon): add minimum Claude Code version check during runtime registration
The daemon now validates the detected agent CLI version against a
minimum requirement before registering a runtime. Claude Code requires
>= 2.0.0 (when --output-format stream-json and --permission-mode
bypassPermissions were introduced). Older versions are skipped with a
warning log, preventing silent failures.
Closes#569
* feat(daemon): add minimum Codex CLI version check (>= 0.100.0)
The `codex app-server --listen stdio://` flag was introduced in v0.100.0.
Older versions lack this flag and fail silently. Add codex to the
MinVersions map so the daemon skips outdated codex CLIs with a clear
warning, matching the existing Claude version check.
Refs #490
* feat(agent): add Hermes Agent Provider via ACP protocol
Integrate Hermes as a new agent backend using the ACP (Agent
Communication Protocol) JSON-RPC 2.0 over stdio — the same pattern
as the Codex provider but with ACP-specific methods.
- New hermesBackend spawns `hermes acp` and drives initialize →
session/new → session/prompt lifecycle
- Handles session/update notifications: agent_message_chunk,
agent_thought_chunk, tool_call, tool_call_update, usage_update
- Auto-approves tool executions via HERMES_YOLO_MODE env var
- Supports session resume, model override, system prompt injection
- Token usage extracted from PromptResponse and usage_update events
- Auto-detected at daemon startup via MULTICA_HERMES_PATH env var
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat(ui): optimize runtime icons and fix create-agent dialog overflow
- Replace OpenClaw pixel-art icon (32 rects) with clean vector paths
- Add Hermes provider icon (NousResearch mascot, 48x48 webp data URI)
- Use provider-specific icons in runtime selector instead of generic Monitor
- Fix dialog overflow: add min-w-0 to grid item so truncate works
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(agent): add required mcpServers param to Hermes ACP session/new
ACP SDK v0.11.2 requires mcpServers as a mandatory field in
NewSessionRequest. Without it, Pydantic validation fails with
"Invalid params" and the agent immediately errors out.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Per-task CODEX_HOME isolated session logs in per-task directories, making
them invisible from the global ~/.codex/sessions/ where users expect to
find them. Symlink the sessions directory back to the shared home so
Codex writes session logs to the global location while keeping skills
isolated per task.
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The bare cache used a mirror-style fetch refspec
(+refs/heads/*:refs/heads/*) which collided with worktree-locked
refs/heads/agent/<task> branches once those branches were pushed
back to origin as PRs. git fetch aborted with "refusing to fetch
into branch ... checked out at ...", the error was swallowed as a
warning, and every subsequent checkout reused the snapshot from
the original clone.
Fix:
- Clone / migrate bare caches to a remote-tracking layout
(+refs/heads/*:refs/remotes/origin/*) so fetched heads never
land in refs/heads/*.
- Resolve the base ref from refs/remotes/origin/HEAD with a
5-level fallback (verified origin/HEAD symref to origin/main
or origin/master to the bare HEAD bridged into origin/<same>
to single-entry origin/* scan to bare HEAD for legacy caches).
- Refuse to guess when refs/remotes/origin/* has multiple
candidates and none match a known fallback, so CreateWorktree
fails loudly instead of basing work on an arbitrary branch.
- Refresh refs/remotes/origin/HEAD after every successful fetch,
not just on the legacy migration path, so a cache that was
already modern picks up an upstream default-branch change.
- Verify the primary symref target actually exists so a phantom
refs/remotes/origin/HEAD from a broken set-head does not
surface a deleted branch.
- Detect legacy caches on the fly and rewrite refspec +
refs/remotes/origin/* + refs/remotes/origin/HEAD in place so
existing clones self-heal on next use.
- Serialize per-bare-repo mutation (both Sync and CreateWorktree)
with sync.Map-backed mutexes so concurrent fetch and worktree
add on the same repo cannot race on git's own lockfiles.
- Narrow the already-exists retry to actual branch-collision
errors so a path-collision no longer silently leaks a branch
into the bare repo.
Implement the Master Agent chat feature allowing users to chat with agents
directly from a floating window, separate from the issue-based workflow.
Backend:
- New chat_session and chat_message tables (migration 033)
- Make issue_id nullable on agent_task_queue for chat tasks
- REST API: create/list/get/archive sessions, send/list messages
- EnqueueChatTask in TaskService with session_id persistence
- WS events: chat:message, chat:done
- Daemon: chat task type with separate prompt builder
- ClaimTaskByRuntime populates chat context (session, message, repos)
Frontend:
- ChatSession/ChatMessage types + API client methods
- core/chat: TanStack Query options, mutations with optimistic updates, WS updaters
- features/chat: Zustand store, ChatFab (floating button), ChatWindow with
real-time streaming via task:message events
- Mounted in dashboard layout (bottom-right corner)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(daemon): update existing worktree to latest remote on reuse
When an agent receives a new task on the same issue, the execution
environment is reused and the repo worktree already exists on disk.
Previously, `multica repo checkout` would fail because `git worktree add`
cannot create a path that already exists — so the agent worked on stale
code from the prior task.
Now `CreateWorktree` detects existing worktrees and updates them:
fetch origin, reset working tree, then checkout a new branch from the
latest remote default branch. The previous task's branch is preserved.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(daemon): propagate actual branch name and use correct ref in worktree reuse
- Return (string, error) from updateExistingWorktree so collision-retried
branch name propagates to WorktreeResult
- Use baseRef directly instead of origin/baseRef — bare clone refspec maps
remote branches to local refs, so remote-tracking refs may not exist
- Remove redundant fetch (worktree shares object store with bare clone)
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
1. Separate ReportTaskUsage endpoint (POST /api/daemon/tasks/{id}/usage)
so usage is captured independently of complete/fail — fixes usage loss
for failed/blocked tasks.
2. Add usage tracking for all four providers:
- Claude: already done (stream-json message.usage)
- OpenCode: extract from step_finish.part.tokens
- OpenClaw: extract from step_end.data token fields
- Codex: extract from turn/completed and task_complete usage fields
3. Remove usage from CompleteTask payload — all usage goes through the
dedicated endpoint now.
Extract token usage from Claude Code's stream-json output in real-time
during task execution, replacing the inaccurate global JSONL log scanner.
- New `task_usage` table: tracks (task_id, provider, model) level usage
- Agent SDK: parse `message.usage` from assistant messages, accumulate
per-model and return in Result
- Daemon: convert agent usage to entries, send with CompleteTask
- Server: store usage on task completion, expose workspace-level
aggregation APIs (GET /api/usage/daily, GET /api/usage/summary)
When repos are present, sub-steps c/d/e/f are now distinct instead of
having two 'c' steps. Each branch (with/without repos) now has its own
complete set of correctly lettered sub-steps.
Prepend the directory of the running multica binary to PATH in the
agent's environment variables. This fixes the issue where isolated
runtimes (e.g. Codex sandbox) cannot find the multica CLI, causing
agent tasks to fail immediately with "command not found: multica".
Closes#451
Add OpenClaw as a fourth supported agent runtime alongside Claude Code,
Codex, and OpenCode. OpenClaw CLI (`openclaw agent -p ... --output-format
stream-json`) is integrated via the same Backend interface pattern.
Changes:
- Add openclawBackend in server/pkg/agent/openclaw.go with NDJSON
event stream parsing (text, thinking, tool_call, error, step, result)
- Register "openclaw" in the agent factory (agent.go)
- Add MULTICA_OPENCLAW_PATH / MULTICA_OPENCLAW_MODEL env var detection
in daemon config
- Include "openclaw" in AGENTS.md config injection alongside codex/opencode
- Add comprehensive unit tests for all event handlers and processEvents
* refactor(cli): overhaul help output to match gh CLI style
- Add gh-style grouped help with CORE/RUNTIME/ADDITIONAL COMMANDS sections
- Use UPPERCASE section headers (USAGE, FLAGS, EXAMPLES, LEARN MORE)
- Format commands as "name: description" with automatic alignment
- Add ENVIRONMENT VARIABLES and EXAMPLES sections to root help
- Apply consistent templates to root, subcommand, and leaf commands
- Update descriptions from "Manage X" to "Work with X" for gh parity
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(execenv): add explicit instruction for agents to always use multica CLI
Agents were using curl/wget to access Multica attachment URLs directly,
which fails due to authentication. Add a prominent "Important" section
to the generated CLAUDE.md template that explicitly prohibits direct
HTTP access and instructs agents to escalate missing CLI functionality
to their workspace owner.
---------
Co-authored-by: Devv <devv@Devvs-Mac-mini.local>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
1. Update CLAUDE.md template to document --limit, --offset, --since
params and guide agents to use pagination when comments are large
2. Add GetJSONWithHeaders to API client; CLI now prints "Showing X of Y
comments" to stderr when paginating
3. Cap --since without --limit at 500 server-side to prevent unbounded
result sets
* fix(agent): instruct agents to use download_url for attachments
Agents were not aware of the signed vs unsigned URL distinction in
attachments, causing failures when trying to read images. Added an
Attachments section to the generated CLAUDE.md/AGENTS.md template that
tells agents to always use `download_url`. Also increased signed URL
expiry from 5 to 30 minutes to better accommodate agent processing time.
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat(cli): add `multica attachment download` command
Adds a dedicated CLI command for downloading attachments by ID. The
command fetches attachment metadata from the API (which returns a fresh
signed URL), downloads the file, and saves it locally. This eliminates
the need for agents to understand signed vs unsigned URLs.
Changes:
- New `multica attachment download <id>` CLI command
- New `GET /api/attachments/{id}` backend endpoint
- `DownloadFile` helper on APIClient
- Updated CLAUDE.md template to document the command
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(cli): sanitize filename and add download size limit
- Use filepath.Base on attachment filename to prevent path traversal
- Add 100MB size limit to DownloadFile (matches upload limit)
- Include response body in download error messages for debugging
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Devv <devv@Devvs-Mac-mini.local>
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* feat(daemon): add opencode as supported agent provider
Add opencode backend alongside claude and codex. The backend spawns
`opencode run --format json`, parses streaming JSON events (text,
tool_use, error, step_start/finish), and supports --prompt for system
prompts. Includes CLI detection, AGENTS.md runtime config, native skill
discovery via .config/opencode/skills/, and 21 tests covering handlers,
JSON parsing, and integration-level processEvents scenarios.
* chore: add .tool-versions to gitignore
* feat(daemon): support direct download update for non-Homebrew installs
Previously, CLI auto-update only worked for Homebrew installations. Non-brew
binaries would fail with "not installed via Homebrew". Now the daemon and
`multica update` fall back to downloading the release binary directly from
GitHub Releases when Homebrew is not detected.
Also fixes:
- Daemon restart now uses the current executable's absolute path instead of
searching PATH, ensuring the updated binary is used
- Brew installs preserve the symlink path so the new Cellar version is picked up
- Daemon startup logs now include the CLI version
- Update UI auto-clears "restarting" status after 5s to show the new version
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
* fix(cli): remove dead DetectNewBinaryPath and guard against nil latest version
Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
The runtime detail page was showing the agent CLI version (claude/codex)
as "CLI Version" because metadata.version stored the agent version from
agent.DetectVersion(). The multica CLI version was never sent.
Fix: daemon now sends cli_version in the registration request, server
stores it as metadata.cli_version alongside the existing agent version,
and frontend reads metadata.cli_version.
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
