Add early detection and graceful degradation when the model exhausts its
iteration budget without producing a text response:
- Inject 75% budget nudge telling model to start summarizing
- Strip all tool schemas on final iteration to force text-only response
- Adaptive web_fetch maxChars (60K→20K→10K) via iteration progress context
- Bump DefaultMaxIterations 20→30 for web-research workloads
- Fix TrimLeft → TrimPrefix to strip only one underscore separator
- Add registryName to indexedResult for parallel path cache
- Use registryName for bootstrapToolAllowlist and loopDetector checks
- Add server-side sanitization for ToolCallPrefix input
- Remove dead chatMessages alias and unrelated ParseStripAssistantPrefill
- Remove orphaned stripAssistantPrefill i18n keys
- Add 13 unit tests for StripToolPrefix
* feat(agent): support tool call prefix stripping for proxy providers
Proxy providers like LiteLLM and OpenRouter may prepend a prefix to
tool call names returned by the model (e.g. "proxy_exec" instead of
"exec"). This broke tool policy validation, registry lookup, and
hardcoded name checks for "team_tasks" and "spawn" in both serial
and parallel execution paths.
Add per-agent toolCallPrefix configuration that strips the configured
prefix from incoming tool call names before registry resolution. The
stripping is applied at resolveToolCallName() which is called before
permission checks, registry execution, and spawn/team_tasks detection.
- Add StripToolPrefix() supporting literal ("proxy_") and template
("{tool_name}") patterns
- Add toolCallPrefix to ToolPolicySpec with backward compat from old
"toolPrefix" JSON key
- Fix config save using spread operator to prevent dropping new fields
- Add UI input in Tool Policy section with i18n (en/vi/zh)
* chore: ignore AI tool config directories
Add .gemini/, .claude/, .opencode/ to .gitignore to prevent
committing user-specific AI tool configurations.
Address issues identified in PR #137 with a cleaner approach:
- Sandbox isolation: add SandboxCwd/ResolveSandboxPath helpers to map
filesystem tool paths to agent-scoped container subdirectories,
preventing cross-agent file access via read/write/edit/list tools
- DooD volume mounting: add resolveHostWorkspacePath with multi-strategy
container ID detection (/proc/self/mountinfo, HOSTNAME, os.Hostname)
and 5s timeout on docker inspect
- Sandbox hints: expand from 1 pattern (binary not found) to 6 patterns
(permission denied, network disabled, read-only FS, missing file,
resource limits) with MaybeFsBridgeHint for filesystem tools
- Nginx DNS: add Docker resolver (127.0.0.11) with dynamic upstream
variable to handle backend container IP changes
- MCP args: switch from comma-separated to space-separated parsing
with quote support for --flag="value with spaces" patterns
- Refactor: rename ExecTool.workingDir to workspace for consistency,
extract sandbox.DefaultContainerWorkdir constant
Detect exit code 127 and "command not found" patterns in sandbox
execution paths and append a [SANDBOX] hint to the error output,
so the model can inform the user about sandbox environment limitations.
Only applies to sandbox execution — host paths are untouched.
Team task announces rebuilt session keys via BuildScopedSessionKey(),
producing different format than WS sessions (ws:direct:uuid vs
ws-userId-ts). Now stores and forwards origin_session_key through
task metadata and dispatch, matching subagent announce pattern.
Also adds "ws" to InternalChannels to suppress spurious warnings.
* fix(memory): auto-backup + overwrite warning for memory files
When an agent writes to a memory file that already has content:
1. Old content is backed up to memory/.prev/{filename} (best-effort)
2. Write succeeds normally (no blocking)
3. Tool result includes warning with previous content (truncated to
4000 chars) so the agent can notice and merge if needed
Backup files are hidden from list_files and not indexed/embedded.
No database migration needed - uses existing PutDocument.
* fix(memory): support append mode + fix UTF-8 truncation in overwrite warning
- Add appendMode parameter to MemoryInterceptor.WriteFile() so
write_file(append=true) actually appends to memory files instead
of silently replacing content. Uses "\n\n---\n\n" separator.
- Fix UTF-8 truncation in overwrite warning: use []rune instead of
byte slicing to avoid cutting multi-byte characters.
- Fix char count in warning message: use len([]rune()) not len().
- Remove .prev/ backup mechanism (prevBackupPath, isPrevPath, backup
writes, ListFiles filter) in favor of simpler approach: append
mode prevents data loss, warning enables self-correction.
---------
Co-authored-by: Luvu182 <208665161+Luvu182@users.noreply.github.com>
Co-authored-by: viettranx <viettranx@gmail.com>
* feat(feishu): add list_group_members tool + native @mention support
Add agent tool to list all members in a Lark group chat via
GET /open-apis/im/v1/chats/{chat_id}/members with auto-pagination.
- New tool `list_group_members`: returns member IDs + display names
- Native @mention conversion: @ou_xxx in outbound messages becomes
real Lark mentions (post: "at" element, card: <at> markdown tag)
- Auto-sync discovered members into ContactStore
- GroupMemberProvider interface on Channel for extensibility
- Add im:chat.members:read scope to channel setup UI
Requires Lark app scope: im:chat.members:read
* fix(feishu): URL-encode API params + channel-aware tool filtering
- URL-encode chatID (PathEscape) and pageToken (QueryEscape) in
ListChatMembers to prevent path traversal and query corruption
- Add slog.Warn when ListGroupMembers fails for observability
- Add ChannelAware interface so tools can declare required channel types
- ListGroupMembersTool now declares RequiredChannelTypes=["feishu"]
- loop.go filters out channel-specific tools when channel type mismatches,
so list_group_members only appears for Feishu conversations
---------
Co-authored-by: Luvu182 <208665161+Luvu182@users.noreply.github.com>
Co-authored-by: viettranx <viettranx@gmail.com>
Subagent announce messages told the parent LLM "There are still X active
runs" — a bare number. The LLM had to reconstruct which subagents were
done vs running from conversation history, causing hallucination when
results arrived out of order.
Replace with a full roster listing each subagent by label and status,
including per-agent config limits. Remove dead CountRunningForParent()
and unused rosterFunc field on AnnounceQueue.
Track per-tool execution time statistics in session metadata. When a tool
call exceeds its adaptive threshold (2x historical max, min 120s default),
send a direct outbound notification to the user.
- ToolTimingMap: parse/serialize/record/threshold from session metadata
- StartSlowTimer: fires once per tool call, auto-cancels on completion
- Team config: slow_tool toggle (default on, always direct, never leader)
- UI: toggle in team settings with i18n (en/vi/zh)
- Store: add GetSessionMetadata to session store interface
* fix(tts): config save + Edge provider registration + dark mode chat bubbles
- Wrap TTS config payload in `raw` field for config.patch RPC (#229)
- Always register Edge TTS provider (free, no API key) instead of gating on `enabled` flag
- Fix low-contrast user message bubbles in dark mode chat
* fix(tts): skip duplicate media dispatch when temp file already delivered
When both the agent loop and the message tool dispatch the same TTS
temp file, the first dispatch succeeds and cleanup deletes it. Filter
out missing temp media files before sending to prevent "file not found"
errors and spurious error notifications on Telegram/Slack/Discord.
* feat(tts): include edge-tts in Docker image when Python enabled
Edge TTS is free (no API key) and serves as a universal TTS fallback.
Install it alongside Python in both ENABLE_PYTHON and ENABLE_FULL_SKILLS builds.
* chore(docker): expose build args from .env for compose builds
Pass ENABLE_OTEL, ENABLE_PYTHON, ENABLE_FULL_SKILLS as env-driven
build args so .env can control Docker build features without editing
docker-compose.yml directly.
* fix(tts): hot-reload TTS config on settings change via pub/sub
TTS providers were only registered at startup, so changing provider/API
key via the Web UI had no effect until container restart. Add a
tts-config-reload bus subscriber that rebuilds the TTS manager on
config changes, matching the pattern used by quota, cron, and web_fetch.
Always create a TtsTool at startup (even without providers) so the
reload subscriber can populate it when settings are first configured.
* fix(tts): protect TtsTool.UpdateManager with RWMutex to prevent data race
UpdateManager() can be called from the config reload goroutine while
Execute() reads t.manager concurrently from agent goroutines. Add
sync.RWMutex following the same pattern as WebFetchTool.UpdatePolicy().
Also update setupTTS doc comment which incorrectly stated it could
return nil — Edge TTS is now always registered.
---------
Co-authored-by: viettranx <viettranx@gmail.com>
* fix(subagent): inherit parent agent's provider instead of alphabetical fallback
Subagents previously used a fixed provider (alphabetically first from the
registry, often "anthropic") regardless of which provider the parent agent
used. This caused invalid combos like anthropic/glm-5 when a zai-coding
agent spawned subagents.
- Pass provider registry to SubagentManager for runtime resolution
- Inject parent provider name into context (WithParentProvider)
- Resolve activeProvider from parent context before LLM call
- Fix trace spans to show actual resolved provider, not default
* fix(providers): api_base fallback from config/env for DB providers
DB providers with empty api_base now inherit from config/env vars
(e.g., GOCLAW_ANTHROPIC_BASE_URL). Prevents proxy API keys from being
sent to the real provider API endpoint.
- Add APIBaseForType() method on ProvidersConfig
- registerProvidersFromDB falls back to config when api_base is empty
- ProvidersHandler uses resolveAPIBase() for model listing
- Add api_base, display_name, settings to provider validation whitelist
* fix(tracing): pass resolved provider name to subagent span emitters
- emitSubagentSpanStart now accepts providerName param instead of
reading sm.provider.Name() — ensures root subagent span reflects
the inherited parent provider, not the fallback default
- registerInMemory now uses resolveAPIBase() so DB providers with
empty api_base inherit the config/env fallback (same as startup path)
---------
Co-authored-by: viettranx <viettranx@gmail.com>
ShellDenyGroups was defined in SystemPromptConfig but lacked full propagation
through parser, Loop fields, context injection, and system prompt population.
Per-agent overrides from other_config JSONB had zero runtime effect.
Changes:
- agent_store.go: Add ParseShellDenyGroups() to extract overrides from JSONB
- loop_types.go: Add shellDenyGroups field to Loop and LoopConfig, wire in NewLoop
- resolver.go: Wire agent-parsed shell deny groups into LoopConfig
- loop.go: Inject shellDenyGroups into context via store.WithShellDenyGroups
- loop_history.go: Populate ShellDenyGroups in system prompt config
- message_test.go: Fix macOS symlink path normalization in test expectations
Fixes test failures on macOS where /var/folders symlinks to /private/var/folders.
plainto_tsquery ANDs all terms, so queries like "teenager coffee
instagram" return 0 results when any single term is missing from the
tsvector. Switch to to_tsquery with OR-joined terms so tasks matching
any keyword are returned, ranked by relevance. Sanitize input with
a whitelist (letters, digits, hyphens, underscores) to prevent
tsquery syntax injection.
- Add RunMediaPaths context key to track media files from current run
- Collect persisted media paths in agent loop after enrichment
- Auto-copy media files to {workspace}/attachments/ when leader creates task
- Append attached files hint in dispatch content so members know what to read
- Scope task_number per (team_id, chat_id) instead of global per team
- Fix NULL chat_id comparison with COALESCE
- Use hard link first, copy fallback to save disk space
- Validate filenames and use restrictive file permissions (0640)
- Add append=true parameter for chunked file writing
- Add ~12000 char warning in tool description and system prompt
- Helps models avoid API truncation on large file writes
- Add display_name to task create response (assignee name)
- Add owner_display_name and created_by_display_name to list/get items
- Pass to_agent_display via dispatch metadata (zero extra DB queries)
- Use display name in announce messages to leader for correct attribution
- Remove handleDelegateAnnounce() dead code (no sender emits delegate:* messages)
- Remove delegate tool reference from intent_classify.go
- Rename LaneDelegate → LaneTeam with backward-compat env var fallback
- Rename ChannelDelegate → ChannelTeammate across all team tool files
- Comment out lifecycle guards in team_tasks_lifecycle.go (TODO: reviewer workflow)
- Update string literals in cron.go, task_ticker.go
- Gate tool_status placeholder_update to non-streaming runs only
- Skip FinalizeStream on tool.call to prevent mid-run content loss
Cherry-picked valuable changes from PR #206:
- hasReadImageProvider supports chain format {"providers":[...]} config
- create_image/video/audio verify file persistence after write with diagnostic logging
- HistoryEntry gains Media field + CollectMedia() for group media context on @mention
- Zalo extractContentAndMedia refactored: all media types via DetectMIMEType/BuildMediaTags, 20MB limit
- Discord/Zalo pass media paths to Record() and collect historical media on @mention
- Zalo send_helpers logs directory contents when checkFileSize stat fails
Search now acquires the team create lock and marks listed, so leaders
can use search instead of list before creating tasks — reduces token
consumption by returning only matching tasks instead of the full board.
Flip write_file deliver param default from false to true so result files
(reports, articles, generated content) are automatically sent as document
attachments to chat channels without requiring explicit LLM opt-in.
Add .md MIME type to mimeFromExt for proper markdown file delivery.
Remove premature EventTeamTaskAssigned broadcast in executeCreate() that
caused duplicate "assigned to" Telegram notifications. Assignment
notification now only fires at actual dispatch (post-turn, fallback, or
unblocked).
Add TeamNotifyQueue (2s debounce, cap 20) to batch rapid-fire task
notifications per chat — reduces N messages to 1 when leader dispatches
multiple tasks at once. In leader mode this also reduces agent turns
from N to 1.
Also fix: ResetTaskStatus now clears progress_percent/progress_step on
retry, and retry broadcast includes TaskNumber/Subject for correct
notification formatting.
* feat: Implement MCP manager for server connections, tool registration, and deferred tool loading for agents.
* feat: Add tests for deferred tool activation logic within the tool registry and agent loop.
* fix(mcp): prevent deny list bypass via lazy activation + fix idempotency race
- Add PolicyEngine.IsDenied() to check deny patterns (incl. group: expansion)
before allowing lazily-activated deferred tools to execute
- Check IsDenied() in both single-tool and parallel execution paths in loop.go
- Make ActivateToolIfDeferred idempotent by checking activatedTools before
returning false, preventing concurrent goroutines from being blocked
- Add tests for deny-on-first-call, group deny patterns, and idempotent
concurrent activation
---------
Co-authored-by: viettranx <viettranx@gmail.com>
- Fix progress event payload missing TaskNumber, Subject, OwnerAgentKey,
ProgressPercent, ProgressStep — notifications were rendering empty
- Fix progress notification format to include task name (consistent with
dispatched/failed) and guard empty ProgressStep
- Change percent tool schema from number to integer for clarity
- Add pre-run member task reminder injecting task context before LLM loop
- Add mid-loop progress nudge every 10 iterations with suggested percent
based on iteration ratio (handles maxIter=0 unlimited case)
- Enhance leader cross-session reminder to show progress % when available
- Strengthen TEAM.md member guidance: focus, result quality, progress rules
- Add progress bar to task list table view (matches kanban card pattern)
Remove fields the model doesn't need from tool response DTOs:
- owner_agent_id → model uses owner_agent_key
- created_by_agent_id → model uses created_by_agent_key
- task_number → model uses identifier (e.g. "T-018-5e65")
Reduces context token consumption. UI (WS API) unaffected —
uses separate TeamTaskData struct with all fields.
- Add WithToolTeamWorkspace/ToolTeamWorkspaceFromCtx context key for
team workspace path (accessible but not necessarily default)
- Create WorkspaceInterceptor for team-specific write validation
(RBAC, quota, blocked extensions, event broadcasting)
- File tools (read_file, write_file, list_files, edit) allow access
to team workspace via allowedWithTeamWorkspace() helper
- read_file/list_files hint team workspace path when file not found
- Registry detects empty tool call args and returns actionable hint
(DashScope/Qwen large-output truncation workaround)
- Move task dispatch from mid-turn to post-turn to prevent dependent
tasks from completing before the current agent's run finishes
- Add team create lock to serialize list→create flows across concurrent
group chat sessions, preventing duplicate task creation
- Require list-before-create gate: agents must call team_tasks(list)
before creating tasks
- Make assignee required on task creation
- Add pagination (50 per page) to task list with offset support
- Slim task list/get/search responses with dedicated structs to reduce
context token usage
- Add task board snapshot in announce messages to leader
- Workspace: allow subdirectory paths in read/delete, show directories
in list output
- UI: reduce kanban card title font size for better visual balance
Team task announce was writing to wrong session (direct instead of group)
because origin_peer_kind was hardcoded as "direct" in dispatch metadata.
This caused leaders to miss completed task results in group conversations.
- Store peer_kind and local_key in task metadata at creation time
- Resolve peer_kind from context → metadata → "direct" fallback in all
dispatch paths (tool, gateway, unblocked)
- Use actual origPeerKind in announce handler session key + request
- Add origin_local_key to gateway dispatch for forum topic routing
- Clarify ask_user guidance: bot must present question directly
- Guide members to use team_tasks progress instead of team_message
- Improve error message when non-owner calls progress action
- Move cache_control from request root (ignored by API) to per-block
placement on last system block and last tool definition
- Change system prompt time format to date-only for better cache stability
- Add builtin datetime tool for precise timestamps (cron, memory, etc.)
- Add atMs past-time validation in cron handleUpdate (was only in handleAdd)
- Update cron description to guide model to use datetime tool first
viettranx