Feat/audit#21
Merged
Merged
Conversation
- PlanRunner: orchestrates LLM planning with skill selection, context metadata collection, action normalization, plan hashing, and DB upsert - ExecuteRunner: executes planned actions via ToolRouter with policy guard, step reference resolution, action logging, and cancellation support
- PolicyGuard: evaluate tool calls with risk classification, block unconfirmed high-risk actions - ToolRouter: dispatch drive.* tools (listFolder, createFolder, move/rename/delete file/folder) against real services - Add classify_tool_side_effect, classify_tool_risk, normalize_action_risk helpers
- AnthropicPlannerClient: calls Anthropic API (compatible) for plan generation with system/user prompts - worker.py: background consumer that polls agent.plan/agent.execute jobs and delegates to runners
- PlanService: enqueue plan jobs with rate-limit enforcement, validate token/steps budgets - ExecuteService: enqueue execute jobs with plan-hash verification and high-risk confirmation check - BackgroundJobService: add agent_phase column support to job enqueue
- Add agent_llm_base_url, remove agent_llm_provider from settings and env - Wire settings into PlanService and ExecuteService via deps - Add anthropic SDK dependency (>=0.104.1)
- Replace unsafe :user_id IS NOT NULL AND owner_user_id = :user_id with owner_user_id = CAST(:user_id AS BIGINT) across all skill/mcp repository queries - Fixes NullPointer when user_id is passed as None and ensures consistent type handling
- Add --no-agent-worker and --agent-worker-count CLI flags - Spawn agent worker subprocesses when AGENT_ENABLED=true
- V13 migration seeds builtin:organizeByType skill with tool whitelist and plan template - test_agent_repositories: add acceptance tests for null user_id with BIGINT casts - test_agent_plan_execute_runtime: add runtime integration tests - test_agent_routes: add route-level plan/execute/cancel tests
…s frontend - Add AgentReasoningEffort, AgentActionRiskLevel types - Add reasoning effort selector to TaskInputDock with i18n labels - Wire reasoningEffort state through useAgentSession into plan/execute payloads - Add high-risk confirmation dialog in runExecute flow - Update mock handlers with riskLevel/requiresConfirmation fields and 409 on unconfirmed high-risk - Update Library dev page with new props
- Add downloadFile, getPreviewUrl, previewFile mock stubs to LeftSidebar.spec.ts, MyFiles.spec.ts, upload.spec.ts - Fixes test failures from unresolved imports
…runners - Add reference_rules module with is_symbolic_id_placeholder and parse_step_reference utilities - Add _validate_action_inputs to plan_runner to reject symbolic placeholders and out-of-range step references at plan time - Update _resolve_references in execute_runner to detect unresolved symbolic placeholders at execution time - Update system prompt to instruct LLM to use only .field syntax
…ter and services - Add mode='json' to all model_dump calls in tool router to ensure datetime fields are serialized as ISO strings - Use jsonable_encoder for action_log inputs_json and outputs_json - Use jsonable_encoder for background_jobs payload before DB storage - Use mode='json' in execute_service payload model_dump
…ization - Add tests for symbolic placeholder rejection at plan and execute time - Add tests for reference acceptance and future-step rejection - Add tests for jsonable_encoder normalization in action_log and background_jobs - Add tests for plan_runner commit/rollback behavior
…el failures - Add extractErrorMessage helper to pull response.data.message, Error.message, or fallback in that order - Apply extractErrorMessage to plan, execute, and cancel catch blocks instead of hardcoded fallback strings
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Basic agentic functions