71 Commits

Author	SHA1	Message	Date
CoreRasurae	4fe14c5b77	feat(agents): add offset-based pagination to file_read ... `file_read` always read from position 0; `max_chars` (ADR-2030 D-3) only bounded where the returned content stopped, so once a large file was truncated there was no way to retrieve the next chunk short of falling back to the `shell` tool (`tail -c +N`, `sed -n`, `dd skip=...`) — the same class of awkward, error-prone detour ADR-2030 D-4 already eliminated for file paths mistaken for shell commands. Adds an optional `offset` character-position argument (default 0, byte-for-byte identical output when omitted) to both the tool schema (`llm_tools._BUILTIN_TOOL_SCHEMAS["file_read"]`) and the implementation (`ToolAgent._file_read_tool`). Combined with `max_chars` it returns the window `[offset, offset + max_chars)`, and the `[FILE_READ_SUCCESS]` header now signals continuation: `MORE_CONTENT_AT_OFFSET: N` when more content remains (N is the exact next offset, so callers never compute `offset + max_chars` themselves), or `NO_MORE_CONTENT_AT_OFFSET: N` when the requested offset is at or beyond end-of-file — a success response with empty content, not an error, since that is the expected terminal state of a correctly-paginating caller. Negative or non-integer offsets raise `ExecutionError`, mirroring the existing `max_chars` validation pattern. See docs/adr/ADR-2033-file-read-offset-pagination.md for the full set of design decisions, including why directory-listing pagination and byte-level seeking are deferred to a future issue. ISSUES CLOSED: #83	2026-07-31 08:53:31 +00:00
CoreRasurae	4b354f5b2d	feat(agents): surface timeout budget in tool schemas and errors ... The `shell` and `http_request` built-in tools are the only two whose execution actually honors the tool agent's `timeout` config field (ToolAgent._execute_shell_command, ToolAgent._http_request_tool), but their LLM-facing schemas (_BUILTIN_TOOL_SCHEMAS in llm_tools.py) never told the model a timeout applied, and neither tool's timeout error named the config field that controls it. - _BUILTIN_TOOL_SCHEMAS["shell"] and ["http_request"] descriptions now state that execution is bounded by the agent's configured `timeout` (seconds), matching the existing precedent of documenting `file_read`'s `max_chars` behavior in its schema. - _execute_shell_command's timeout ExecutionError now names `timeout` as the adjustable config field, in addition to the elapsed seconds it already reported. - _http_request_tool now catches asyncio.TimeoutError distinctly from the generic `except Exception`, so a request timeout produces a message naming the config field instead of the previous generic "HTTP request failed" text (which gave no indication a timeout was the cause). Non-timeout HTTP failures are unaffected — same exception handling, same message shape as before. No change to enforcement mechanics: `self.timeout` remains the sole config field (default 1s, Actor Configuration Standard §4.5). This is schema/message content only, within the extension precedent already established by ADR-2030 (D-1 schema registry, D-4/D-6 error-quality improvements) — no new ADR was needed for this change. Tests: extended features/tool_agent.feature (updated the two existing shell/http_request timeout scenarios to also assert the new config-field hint) and features/llm_tools_coverage.feature (new scenarios asserting both schema descriptions mention the timeout budget). Full suite: 141 features / 2782 scenarios / 13172 steps passed; coverage 96.8% (>=96.5% threshold). Robot integration_tests: 27/27 suites passed. ASV benchmark_regression: no significant change, as expected for a messaging-only change. ISSUES CLOSED: #82	2026-07-30 19:29:54 +01:00
CoreRasurae	a8f68b8262	fix(graph): enforce USD budget at each agent invocation with pre-flight gate ... Enforces the USD budget (max_cost_usd) as a hard pre-flight gate before every node execution and at each main-agent ainvoke round inside the multi-turn tool loop, with pruning passes silently skipped when the budget is exhausted. Pre-flight gate (pure_graph.py): _check_budget_pre_flight() raises ExecutionError(kind='cost', reason='budget_exhausted') immediately before node execution in both _execute_from_node and all three branches of _stream_from_node. In-node budget gating (llm.py): _check_budget_before_invoke() gates every _retry_ainvoke() call inside _execute_tool_loop, process_message, and stream_message. _should_skip_pruning() skips pruning passes when budget exhausted. _accumulate_cost_after_invoke() computes USD cost from token counts using the pricing table and advances current_accumulated_cost so subsequent budget checks see the updated cost. ContextVar plumbing (retry.py, pure_graph.py): current_accumulated_cost, current_max_cost_usd, and current_pricing ContextVars carry per-node budget state from PureLangGraph into LLMAgent. Set before node.execute() via _set_budget_context_for_node, reset in finally blocks via _reset_budget_context. Reviewer fixes (rui.hu #80): - Move cost accumulation inside try block in terminal streaming branch to prevent ContextVar leak (was set after finally reset). - Standardize log precision to $%.6f across all budget messages. - Remove orphaned 'already exhausted' step (criterion unreachable). - Add dedicated streaming budget enforcement scenario. - Remove redundant = None default on _reset_budget_context pricing_token. ISSUES CLOSED: #76	2026-07-12 19:51:39 +01:00
CoreRasurae	2072bc3ad1	feat(langgraph): implement real type:tool graph node execution via ToolAgent dispatch ... Wires NodeConfig.tool and NodeConfig.static_config through to ToolAgent.invoke() so type: tool graph nodes execute real tool calls instead of returning no-ops. - NodeConfig: adds tool (str | None) and static_config (dict) - Node._execute_tool: creates ToolAgent, invokes with static config merged with threaded previous output, returns ToolMessage - runtime_dispatch: auto-creates ToolAgent for type: tool nodes without pre-registered agents - pure_graph: parses tool:/config: YAML keys into NodeConfig - BDD: 11 scenarios, 33 steps, all passing Closes #75	2026-07-09 14:06:59 +00:00
hurui200320	9ec578f229	feat(agents): add tool_agent_class parameter to AgentFactory and create_executor ... Extends AgentFactory.__init__ and create_executor() with a keyword-only tool_agent_class: type[ToolAgent] = ToolAgent parameter so callers can inject a custom ToolAgent subclass at runtime without monkey-patching module globals (resolves the five-module setattr loop in cleveragents-webapp). **AgentFactory** (agents/factory.py): - Accepts tool_agent_class keyword argument; validates it is a ToolAgent subclass at construction time (fail-fast, CONTRIBUTING.md §Argument Validation). - Stores as self._tool_agent_class and uses it to populate self.agent_types['tool'], so every create_agent('tool') call instantiates the supplied subclass. - Forwards tool_agent_class to LLMAgent when creating 'llm'-typed agents so the LLM tool-calling loop honours the injected subclass. - Default ToolAgent is preserved, so all existing callers are unaffected. **LLMAgent** (agents/llm.py): - Accepts a keyword-only tool_agent_class: type[ToolAgent] = ToolAgent argument, validated as a ToolAgent subclass at construction time. - The multi-turn tool-call loop (_execute_tool_loop) constructs its ephemeral tool executors via self._tool_agent_class(...) across all three dispatch sites (regular loop, budget-exhaustion synthesis round, and stuck-model synthesis round), so the LLM tool-calling path no longer falls back to the module-level ToolAgent. This closes the gap that previously prevented removing the webapp monkey-patch for the LLM tool-calling path. **Executor / create_executor** (runtime.py): - Executor.__init__ accepts and validates tool_agent_class; stores as self.tool_agent_class. - create_executor() accepts tool_agent_class and forwards it to Executor. **Dispatch paths** (runtime_dispatch.py): - _execute_tool: uses executor.tool_agent_class(...) instead of the module-level ToolAgent(...) so the direct-construction path (bypassing AgentFactory) also honours the injected subclass. - _execute_graph and _execute_graph_stream: pass tool_agent_class=executor.tool_agent_class to AgentFactory so graph nodes using tool-typed agents pick up the subclass. - _execute_llm and _execute_llm_stream: pass tool_agent_class=executor.tool_agent_class to AgentFactory so the single-LLM actor's internally-created LLMAgent (and its tool-calling loop) honours the injected subclass. Without this, a single LLM actor that makes tool calls silently fell back to the base ToolAgent. - _execute_multi_actor: passes tool_agent_class=executor.tool_agent_class to the sub-Executor so every actor inside a multi-actor bundle (tool, graph, or LLM) honours the injected subclass. Without this, all dispatch paths reachable through a multi-actor bundle fell back to the base ToolAgent. - Removed the now-unused module-level ToolAgent import; the _execute_tool docstring cross-reference is fully qualified so it still resolves. All six dispatch paths that construct or dispatch tool agents now forward the injected subclass. **ReactiveCleverAgentsApp** (core/application.py): - Removed the redundant register_agent_type('tool', ToolAgent) call from load_configuration and the dead type_map re-registration block in _create_agents; AgentFactory.__init__ already owns the 'tool' registration. This drops the last hardcoded module-level ToolAgent references flagged in the issue #73 audit. **Tests** (features/tool_agent_class_injection.feature + steps): - 17 BDD scenarios covering: default class when parameter omitted; custom subclass stored on factory and executor; invalid class rejected with ConfigurationError (factory, executor, and LLMAgent); factory create_agent returns custom instance; _execute_tool dispatch instantiates custom subclass; the LLM tool loop instantiates the injected subclass; isinstance LSP compatibility through the factory; the _execute_llm dispatch path forwarding tool_agent_class to AgentFactory (single-LLM actor making a tool call); and the _execute_multi_actor dispatch path forwarding tool_agent_class to the sub-Executor. The two new dispatch-path scenarios were verified to fail without the source fixes (instantiation_count=0), confirming they guard the injection contract for the single-LLM and multi-actor paths. **Updated tests** (features/steps/runtime_coverage_steps.py, features/steps/runtime_extended_coverage_steps.py): - Removed the now-dead patch('cleveractors.runtime_dispatch.ToolAgent') mocks (and their mock instances) that no longer intercept _execute_tool after it switched to executor.tool_agent_class. The tool-actor scenario continues to pass via the real ToolAgent; the normal-coverage branch uses patch.object(ToolAgent, 'process_message', ...) to stub tool execution. **Updated test** (features/steps/credential_executor_paths_steps.py): - The _execute_multi_actor credentials-forwarding scenario intercepts Executor.__init__ with a capturing_init to assert the credentials dict reaches the sub-Executor. Now that _execute_multi_actor forwards tool_agent_class to the sub-Executor, capturing_init accepts and forwards that keyword so the assertion continues to pass. **Removed obsolete test** (features/application_coverage_gaps.feature): - Dropped the '_create_agents registers built-in agent types' scenario and its step definitions; the behaviour it asserted (redundant re-registration) was removed from _create_agents. ISSUES CLOSED: #73	2026-07-08 09:23:34 +00:00
CoreRasurae	ceb577e405	fix(nodes): propagate ExecutionError from retry mechanism instead of swallowing it ... When all LLM communication retries are exhausted, call_with_retry() raises ExecutionError(kind=\"timeout\"). Two broad except Exception handlers in Node._execute_agent() and Node.execute() systematically caught this, converting the timeout into an error string in graph state or — in the streaming path — an empty response. Added targeted except ExecutionError as e handlers that check e.kind and re-raise. Non-timeout ExecutionError instances (tool errors, etc.) continue to be caught and converted to error strings for graceful graph continuation. Extend _is_http_comms_error() in retry.py to recognize provider SDK connection errors (APIConnectionError, APITimeoutError) by class name, so they are retried with exponential backoff instead of propagating immediately. Add safety net in process_message() to wrap non-retried connection errors as ExecutionError(kind='timeout') for correct propagation through the node layer (issue #71). Two new BDD scenarios verify: - ExecutionError(kind=\"timeout\") propagates with correct kind/reason - Non-timeout ExecutionError is still caught gracefully ISSUES CLOSED: #71	2026-07-07 13:47:02 +01:00
CoreRasurae	52451b2550	test(agents): disable retry in error-path tests to avoid unnecessary delays ... Pre-existing tests that inject error-raising mock chat models (raising LangChainException or RuntimeError) were triggering the full exponential-backoff retry loop (~63.5s per test). With 6 such tests, this added minutes to the suite. Set max_retries=0 in each error-path step definition so the retry loop fails fast on the first attempt, preserving the test contract while avoiding unnecessary sleep delays.	2026-07-03 16:00:20 +01:00
CoreRasurae	55734cd512	feat(agents): add LLM agent retry mechanisms with exponential backoff ... - Documents the retry feature implementation (issue #69, ADR-2032) Introduce configurable retry mechanisms (ADR-2032) for all LLM agent communications (main agent and pruning agent) with exponential backoff. - New config fields: max_retries (default 7), max_retry_time (default 60s) -1 sentinel disables the corresponding guard (infinite retries/time) - Exponential backoff: initial 0.5s, doubled each retry - Termination: whichever comes first (max_retries or max_retry_time), with -1 disabling the respective check - Counter and accumulated wait time reset on success - Timeout ExecutionError includes URL, actor graph name, and retry stats - Uniform scope: applies to both main agent and pruning agent - Timeout raises ExecutionError(kind="timeout") with URL, actor graph name, retry count, and accumulated wait time in the message (D-7) - Same client/connection reused across retries, no re-establishment (D-6) - Uniform scope: applies to both main agent and pruning agent (D-5) - Graph name threaded through state metadata and ContextVar for error reporting - BDD tests (7 scenarios) for retry behaviour including -1 sentinel cases - Reject max_retries/max_retry_time values < -1 with ConfigurationError - Reset current_graph_name ContextVar after node execution - Wrap no-tools astream() with per-chunk granular retry via call_with_retry - Update ADR D-7 with symmetrical disabled-guard example - Update CHANGELOG to reflect full feature scope ISSUES CLOSED: #69	2026-07-03 16:00:11 +01:00
hurui200320	306f114b8a	feat(agents): add tool-call support to LLMAgent.stream_message ... Extract the entire multi-turn tool-call orchestration from the inline loop inside process_message() into a new shared private method _execute_tool_loop(). Both process_message() and stream_message() now delegate to this single implementation, eliminating the previous capability gap where stream_message() had no tool-calling support. Design: - _ToolLoopResult dataclass: carries final_response, accumulated_prompt, accumulated_completion, budget_exhausted, and synthesis_was_run. Both callers read token counts from this object instead of local sentinels. - _ToolLoopError exception: wraps any exception raised inside the helper and carries partial accumulated token counts + any_invocation_made flag. Callers catch this, set their _captured_prompt/_captured_completion sentinels (billing integrity), then re-raise the original cause. - _execute_tool_loop(): extracted verbatim from process_message(). Contains the full ainvoke() loop, per-tool ToolAgent dispatch, token accumulation across rounds, token-budget exhaustion + synthesis path (§4.4.7 D-4), stuck-model synthesis prompt, and tool-output pruning pass (§4.4.8). On exception, wraps in _ToolLoopError for the caller. process_message() refactor: - Replaces ~490 lines of inline tool loop with a 15-line delegation. For tool-configured actors: awaits _execute_tool_loop(), extracts response text and token counts from the result. For no-tool actors: single plain ainvoke() (unchanged behaviour). Billing-integrity sentinels (_captured_prompt/_captured_completion) are set from _ToolLoopError on exception and from the result on success. All existing Behave scenarios continue to pass unmodified. stream_message() integration: - When tools are configured: awaits _execute_tool_loop() and yields the final response as a single content chunk. Token counts come from the result's accumulated_prompt/accumulated_completion fields (spanning all tool rounds + pruning passes). Memory update is performed before the generator returns. Billing sentinels are updated from _ToolLoopError on exception. - When no tools are configured: the existing astream() token-by-token path runs completely unchanged. Single-chunk trade-off: when tools are involved, the user already waits for tool execution before the final answer begins, so yielding that answer as one chunk vs. token-by-token has minimal UX impact. The alternative (re-calling astream() after ainvoke() produced the final response) would double LLM cost for every tool-using request. Webapp workaround: the actor_config_has_tools fallback in cleveragents/cleveragents-webapp#328 (stream_fallback_to_non_stream in _execute_via_cleveractors_stream / _generate_actor_sse) can now be removed. Tests added: - features/llm_agent_tool_loop.feature + steps: 10 Behave scenarios covering _execute_tool_loop() internals independently (single round, multi-round, max_rounds exhaustion, token accumulation, budget exhaustion, stuck-model synthesis, pruning pass, tool dispatch error, exception path with partial counts, pre-invocation ConfigurationError). - features/llm_agent_stream_tool_calls.feature + steps: 7 Behave scenarios for stream_message() with tool-call support (single chunk output, two-round token sum, final text, no-tools astream unchanged, no-tools token counts, exception billing integrity, memory update). - robot/llm_tool_calling.robot: two new Robot integration tests: 'Streaming Path Exercises Tool Loop Via Execute Stream' and 'Streaming Path Accumulates Tokens Across Two Tool Rounds'. - robot/ToolCallingTestLib.py: new keywords and create_executor_with_multi_round_tool_calling_agent factory. Quality gates: format ✓ lint ✓ typecheck ✓ security_scan ✓ unit_tests ✓ integration_tests ✓ coverage_report ✓ (97.0%) ISSUES CLOSED: #67	2026-07-02 07:47:39 +00:00
CoreRasurae	8a7e353767	fix(actors): llm token consumption undercounted ... - enforces pruning model token-tracking metadata availability for cost tracking - also temporarily enforces pruning-model match ISSUES CLOSED: #65	2026-07-01 14:55:58 +01:00
CoreRasurae	bcce5dc42a	fix(agents): validate tool name against declared tools list before dispatch ... ISSUES CLOSED: #63	2026-06-30 12:06:05 +01:00
CoreRasurae	964e87cc0f	feat(tool-loop): add token-budget awareness and tool output pruning ... Implement token-budget awareness (§4.4.7) and tool output pruning (§4.4.8-9) for the multi-turn tool-call loop to prevent context-window exhaustion. Token budget: tracks estimated token count before each ainvoke(), emits warning at 75% budget, injects synthesis prompt at exhaustion with one final tool-call round. Estimation via chat_model.get_num_tokens() with len(content)//4 heuristic fallback. Config key: token_budget_percent. Tool output pruning: implicit LLM extraction pass scoped to file_read calls. Schema augmentation injects output_prune (bool) and output_prune_context (string) meta-arguments. Pruning model responds with [PRUNE_INFO_START/END] and [PRUNE_OUTPUT_START/END] markers. Config keys: allow_tool_output_pruning, pruning_model. - Capture output_prune value in budget synthesis flow with proper guard - Fix ADR D-4: synthesis prompt is HumanMessage not SystemMessage - Fix Robot tests: replace /dev/null with temp file path ADR-2031 documents all specification extensions. 21 new BDD scenarios, 5 Robot integration tests, 16 ASV benchmarks. ISSUES CLOSED: #61	2026-06-26 18:21:14 +01:00
CoreRasurae	227e2feece	feat(agents): implement multi-turn tool-call loop and sandbox improvements ... Implements the complete LLM agent tool-calling pipeline — the tool-calling path was broken in multiple ways: tools from agent config were not passed to the LLM, tool execution was single-pass (the model could not see tool results or make follow-up calls), tool errors were discarded, and shell/ python_exec tools were never registered. Key changes: - Multi-turn tool-call loop: passes tools to the LLM via ainvoke(tools=...) and re-invokes after each batch of ToolResults, with a configurable round limit (tool_max_rounds config / TOOL_MAX_ROUNDS env var, default 20). - Tool schema module (llm_tools.py): normalize_tool_entry() converts string tool names and config dicts to OpenAI function-calling format with full parameter schemas and LLM-facing guidance (max_chars for file_read, sandbox builtins for python_exec, etc.). - file_read enhancements: directory listing with file sizes and type indicators, max_chars truncation to prevent context overflow, shell command detection with redirect to shell tool, file-not-found recovery hints with parent directory listing. - python_exec sandbox: stdout capture so print() produces visible output, NameError guidance directing LLM to file_read/file_write instead of using open(). - shell/python_exec tool registration in builtin_tools when allow_shell / exec_python is enabled. - create_subprocess_shell for shell commands (was create_subprocess_exec, which blocks pipes/redirections/chains). - Tool error propagation: ExecutionError/ConfigurationError details included in ToolMessages for LLM self-correction. - Stuck-model recovery: injects synthesizing prompt when model exhausts tool rounds without producing content. - LLM provider error details preserved in ExecutionError messages instead of generic "LLM processing failed". - Empty message filtering in _prepare_conversation_history() prevents whitespace-only messages from polluting downstream conversation history in multi-agent pipelines. Closes #59	2026-06-23 12:39:20 +01:00
CoreRasurae	8f986c1e31	test(coverage): add coverage scenarios for pre-existing code paths ... Additional BDD scenarios covering registry resolver errors, cache TTL expiry, runtime dispatch normalization, template base edge cases, validation actor coverage gaps, and YAML Jinja loader deferred rendering.	2026-06-23 10:49:16 +01:00
hurui200320	ed9c4dc95d	fix(llmagent): do not close shared cached httpx clients in cleanup() ... LLMAgent.cleanup() previously iterated over hard-coded provider SDK client attributes (root_async_client, root_client, _async_client, _client) and called close() on each. Recent langchain-anthropic and langchain-openai versions cache their default httpx clients via module-level lru_cache functions. Closing those clients poisoned the cache: every subsequent ChatAnthropic/ChatOpenAI instance in the same process received the same closed httpx client and failed with a connection error. Fix: cleanup() now only releases the agent's own reference to the chat model (self._chat_model = None). The removed _KNOWN_CLIENT_ATTRS class variable has been deleted and ClassVar removed from the typing import. The concurrent-idempotency guarantee is preserved: the lock is still acquired before nulling _chat_model, so two concurrent cleanup() calls cannot both see a non-None model and attempt conflicting operations. The four provider SDK client-closing scenarios in credential_injection.feature and llm_missing_coverage.feature are removed as they tested the old (buggy) behaviour. Their step definitions are removed from credential_cleanup_steps.py (now only carries the resolve_class_ref patch step) and llm_missing_coverage_steps.py is updated with the corrected assertions. Six new regression BDD scenarios tagged @tdd_issue @tdd_issue_57 are added in features/llm_cleanup_shared_client.feature, covering all four provider SDK client attribute paths (Anthropic _async_client/_client, OpenAI root_async_client/root_client) and two end-to-end two-agent scenarios that prove a second agent can run successfully after the first is cleaned up. ISSUES CLOSED: #57	2026-06-17 17:19:44 +00:00
CoreRasurae	57b1a2e70e	test(unit): increase BDD coverage to 97% for registry, LLM, YAML template error branches ... Adds 4 Behave BDD feature files with 15 scenarios covering previously untested error-path branches: - registry_resolver_errors.feature (7 scenarios): Invalid semver, empty version lists, alias mismatches, ID/local/unknown reference error handling - llm_agent_config_validation.feature (4 scenarios): Empty name, non-dict config, non-TemplateRenderer, non-str credential key validation - llm_provider_url_canonicalization.feature (2 scenarios): URL canonicalization with explicit port and scheme lowercasing - yaml_template_non_dict_errors.feature (2 scenarios): Non-dict YAML parse results in template engine and inline Jinja handler Coverage: 96.8% -> 97.1%	2026-06-15 23:13:46 +01:00
CoreRasurae	517dfb4cce	feat(registry): implement local namespace reference resolution ... Implements the local:<path> reference scheme per Package Registry Standard v1.0.0 §5.3. Closes #46	2026-06-12 17:11:35 +01:00
hurui200320	d93c32ccee	feat(streaming): add Executor.execute_stream() returning AsyncIterator[str] for token-by-token delivery ... Implements the full streaming execution path for the CleverThis router: - **LLMAgent.stream_message(message, context)** (agents/llm.py): Async generator calling self.chat_model.astream(messages) and yielding str(chunk.content) per chunk. Captures token counts from the final chunk's usage_metadata using the same _safe_int() fallback chain as process_message(). Sets _last_token_usage and last_token_usage_var after exhaustion. - **Node.stream_agent(state)** (langgraph/nodes.py): Async generator that mirrors _execute_agent() but uses agent.stream_message() for LLMAgent instances and falls back to process_message() for non-LLM agents. Captures per-node token usage into self._last_stream_usage after stream exhaustion. - **PureLangGraph.execute_stream()** (langgraph/pure_graph.py): Async generator mirroring execute(). Calls _stream_from_node() which uses astream() for all AGENT nodes (buffering tokens for intermediate nodes, yielding only from the terminal node) and ainvoke() for non-AGENT nodes. Stores final state and node usages in _last_stream_state / _last_stream_node_usages for post-stream access. Same limit enforcement (timeout_ms, max_depth, max_model_calls, max_tool_calls) as _execute_from_node(). - **_execute_llm_stream() / _execute_graph_stream()** (runtime_dispatch.py): Async generator dispatch functions mirroring _execute_llm() / _execute_graph(). Set executor.last_result after stream exhaustion using collected token usage. - **Executor.execute_stream(message)** (runtime.py): Public async generator that dispatches to _execute_llm_stream() for llm actors and _execute_graph_stream() for graph actors. Raises ConfigurationError for tool/multi_actor types. Resets last_result to None at entry; callers must exhaust the iterator. - **Executor.last_result** (runtime.py): ActorResult | None attribute set by execute_stream() after the iterator is exhausted. - Intermediate AGENT nodes use astream() internally (tokens buffered, not yielded to caller); terminal AGENT nodes yield buffered tokens. This avoids double- running the terminal LLM while satisfying AC2. - Non-AGENT (FUNCTION/TOOL) nodes always use ainvoke() via node.execute(). - timeout_ms in execute_stream() buffers all tokens via asyncio.wait_for() around _collect_stream_tokens(), then yields them — this is simpler than wrapping an async generator directly. - Tool and multi_actor actor types raise ConfigurationError (no streaming path). - 60 new BDD scenarios in features/execute_stream.feature covering all acceptance criteria: token delivery, last_result population, token count extraction, limit enforcement (timeout, model_calls, tool_calls), error paths, graph streaming, non-AGENT node paths, parallel execution paths, and edge cases. - All 2395 existing unit tests continue to pass. - Coverage: 96.5% (meets threshold). ISSUES CLOSED: #16	2026-06-12 10:08:12 +00:00
CoreRasurae	e8bd348c77	feat(registry): implement RegistryError hierarchy with typed exceptions ... Add RegistryError base with message, details, and original_reference fields. Implement all 9 typed exception types per Package Registry Standard §13.2: PackageNotFoundError, InvalidPackageIdError, InvalidPackageReferenceError, VersionNotFoundError, ValidationError, AuthenticationRequiredError, AccessDeniedError, ConflictError, and RegistryNetworkError. RegistryNetworkError carries status_code and url for transport failures. __str__ includes original_reference when available. exception_for_status() maps HTTP codes to typed exceptions. _ERROR_TYPE_MAP enables error-type parsing from JSON error bodies. 23 Behave BDD scenarios in features/registry_exceptions.feature. 12 Robot Framework integration tests in robot/exceptions.robot. Existing registry_http_client tests updated for InternalServerError removal. ISSUES CLOSED: #29	2026-06-11 20:44:54 +01:00
CoreRasurae	054b432197	feat(registry): implement RegistryCache with LRU eviction and TTL ... - Add RegistryCache with LRU eviction, TTL expiry, SHA-1 content validation - Add CacheFactory for centralised cache configuration (Factory Method pattern) - Add CacheStats dataclass for hit/miss/eviction observability - Implement singleflight coalescing for concurrent-miss protection - Use tuple keys for resolve store to prevent separator collision - Fix cancellation-safe singleflight via plain Future + asyncio.shield - Independent max_size budgets per store (content + resolve) - Concurrent test assertions strengthened - All test imports moved to module level per CONTRIBUTING.md - Unused dead code removed from tests ISSUES CLOSED: #28	2026-06-11 19:26:04 +01:00
hurui200320	17d99abce7	feat(execution-limits): add structured ExecutionError kind/reason fields; enforce all 5 execution limits in PureLangGraph ... AC1: ExecutionError gains kind (categorical: depth, model_calls, tool_calls, timeout, cost) and reason (sub-code: budget_exhausted or missing_pricing_entry) fields, both defaulting to empty string. All existing raise ExecutionError(msg) call sites are backward-compatible. AC2: PureLangGraph.__init__ accepts limits: dict[str, Any] and pricing: dict[str, Any] (both default to {}). When limits['max_depth'] is supplied, a depth breach raises ExecutionError(kind='depth') instead of silently returning the current message (the old behaviour was a silent data-loss bug). When no max_depth limit is supplied, the legacy heuristic (max(2000, len(nodes)*50)) is used with a silent cap for backward compatibility with existing callers that do not pass limits. AC3: max_model_calls is checked before each NodeType.AGENT execution; counter increments after the check so the limit is enforced before the (limit+1)-th invocation. Breach raises ExecutionError(kind='model_calls'). AC4: max_tool_calls is checked before each NodeType.TOOL execution; same pattern as model_calls. Breach raises ExecutionError(kind='tool_calls'). AC5: execute() wraps _execute_from_node() in asyncio.wait_for() when limits['timeout_ms'] is set. asyncio.TimeoutError is caught and re-raised as ExecutionError(kind='timeout') so the router can map it to HTTP 429. AC6+AC7: After each LLM node, cost is computed from the pricing table (rates are USD per million tokens, matching ADR-2029 example values such as gpt-4.1-mini prompt=/bin/zsh.15/1M). Cost breach raises ExecutionError(kind='cost', reason='budget_exhausted'). A missing provider or model entry raises ExecutionError(kind='cost', reason='missing_pricing_entry') — proceeding with assumed zero cost is forbidden per ADR-2029. An empty pricing dict ({}) disables all cost checking (cost tracking was not requested). AC8: ExecutionError re-exported from cleveractors/__init__.py and added to __all__. Wire: runtime_dispatch._execute_graph() now passes executor.limits and executor.pricing to PureLangGraph so the limits reach the enforcement layer. ExecutionError is also re-raised (not swallowed) in _execute_from_node()'s broad except block so limit errors always propagate to the caller. New BDD tests (features/execution_limits.feature, 19 scenarios): - ExecutionError defaults kind and reason to empty strings - ExecutionError accepts kind and reason keyword arguments - ExecutionError exported from cleveractors top-level package - Graph exceeding max_depth raises ExecutionError(kind='depth') - Two AGENT nodes with max_model_calls=1 raises on 2nd call - Two TOOL nodes with max_tool_calls=1 raises on 2nd call - Slow node with timeout_ms=50 raises ExecutionError(kind='timeout') - LLM node exceeding max_cost_usd raises budget_exhausted - Unknown provider/model in pricing raises missing_pricing_entry - Empty pricing table skips cost calculation entirely - Executor limits/pricing propagation verified Quality gates: lint pass, typecheck 0 errors, unit_tests 2317/2317 pass, integration_tests pass, coverage 97.17% (threshold 96.5%). ISSUES CLOSED: #15	2026-06-11 09:37:37 +00:00
hurui200320	2664ebfd75	feat(ActorResult): implement ActorResult and NodeUsage types; capture per-node token counts from LangChain responses ... AC1: Move NodeUsage and ActorResult from runtime.py into the spec-mandated cleveractors/result.py module. runtime.py now imports and re-exports both types for backward compatibility. result.py defines __all__, uses dict[str, Any] | None (PEP 585) for the state field, and makes nodes a required positional field (no default) so the contract that nodes is always non-empty is statically enforced. AC2: LLMAgent.process_message() extracts real token usage from response.usage_metadata (primary) with isinstance(dict) guard, falling back to response.response_metadata['token_usage'] with its own isinstance(dict) guard. If neither is available, logs a WARNING with a cause parameter that distinguishes the three failure conditions (empty usage_metadata, missing response_metadata, or missing token_usage key) so operators can triage billing discrepancies. Non-numeric token values are coerced via _safe_int() with fallback to 0 and a warning log. AC3: Token counts are available alongside the response string via the side-channel attribute LLMAgent._last_token_usage, keeping process_message() return type as str (preserving the Agent base-class contract unchanged). AC4: Node._execute_agent() reads _last_token_usage, provider, and model from any LLMAgent after execution and includes a _node_token_usage dict in the state-updates return. PureLangGraph._node_usages accumulates one (node_id, provider, model, prompt_tokens, completion_tokens) 5-tuple per LLM node invocation during _execute_from_node(). PureLangGraph.execute() now returns a 3-tuple: (response, final_state, node_usages). process_message() unpacks and discards the extra elements. AC5: Executor._execute_graph() uses the node_usages list from execute() to build NodeUsage objects. Executor._execute_llm() reads _last_token_usage directly from the LLMAgent instance instead of calling _estimate_tokens(). State isolation verified by a new test that calls execute() twice on the same Executor with different mocked token counts and asserts independent results. AC6: Aggregation invariant enforced in both _execute_llm() and _execute_graph() via sum() over nodes list. AC7: ActorResult and NodeUsage re-exported from cleveractors/__init__.py via cleveractors.result (path updated from cleveractors.runtime). Cleanup: Delete runtime_tokens.py (estimate_tokens, estimate_graph_tokens) and the private _estimate_tokens() function in runtime.py — both are superseded by real LangChain usage_metadata extraction. Remove unused import logging and logger = logging.getLogger(__name__) from runtime.py (dead code after dispatch logic was moved to runtime_dispatch.py). Move NodeUsageTuple type alias in pure_graph.py to after the import blocks. Use tuple unpacking for last_usage in runtime_dispatch._execute_llm(). New/updated BDD tests (actor_result_token_counting.feature, 25 scenarios): - AC2: usage_metadata primary path (input_tokens/output_tokens) - AC2: truthy non-dict usage_metadata treated as absent (isinstance guard) - AC2: truthy non-dict response_metadata treated as absent (isinstance guard) - AC2: response_metadata={} (empty dict) logs warning, counts 0 - AC2: response_metadata fallback path (prompt_tokens/completion_tokens) - AC2: no usage data -> warning logged, counts 0 - AC2: response_metadata attribute present but None - AC2: stale token reset before each call - AC5: state isolation — execute() twice produces independent results - AC6: aggregation invariant for single-node and multi-node actors - AC6: placeholder node_id exact format assertion for _execute_graph - AC6: placeholder node_id exact format assertion for _execute_multi_actor - AC7: import correctness from cleveractors and cleveractors.result CHANGELOG.md updated with Changed (real token extraction, result module move) and Removed (runtime_tokens.py) entries under [Unreleased]. Quality gates: lint pass, typecheck 0 errors, unit_tests 2157/2157, integration_tests 86/86, coverage 97.2% (threshold 97%). ISSUES CLOSED: #14	2026-06-11 03:05:20 +00:00
CoreRasurae	284d2a9f00	feat(registry): extend TemplateType and integrate PackageReference into template system ... Extend TemplateType enum with TEMPLATE, SKILL, ACTOR, MCP, LSP. Add optional package_ref field to ComponentReference for registry references. Create ReferenceResolver with multi-server RegistryClient pool and caching. Update InstantiationContext for registry-aware resolution with _original_reference. Extend TemplateRegistry, EnhancedTemplateRegistry, and TemplateStore for 8 types. Add registry reference detection in instantiate_from_config. Add GenericTemplate for new package types without specialized template classes. ISSUES CLOSED: #27	2026-06-10 12:50:38 +01:00
CoreRasurae	4e12902492	feat(registry): implement ReferenceResolver with version alias resolution ... Introduce ReferenceResolver implementing the Package Registry Standard v1.0.0 §5.3 and §4.2: ReferenceResolver: - parse(ref_str) → PackageReference handles all three reference schemes (registry, ID, local), wrapping ValueError into InvalidPackageReferenceError - resolve(ref_str) → PackageId resolves references to concrete IDs: ID refs directly, registry refs via RegistryClient, local refs reserved - Integrates with RegistryClient for remote HTTP resolution Version resolution (§4.2): - resolve_version(version, available_versions) resolves concrete (vX.Y.Z) and mutable aliases (latest, vx, x, vX.x, vX.Y.x) - is_concrete_version() / is_version_alias() classification helpers Integration tests: 35 Robot Framework test cases against fake registry server with version alias resolution support Benchmarks: 7 ASV benchmark classes covering parse, resolve, and version resolution across small/medium/large version lists ISSUES CLOSED: #26	2026-06-10 12:04:00 +01:00
CoreRasurae	8fa9e7652d	feat(registry): implement Canonicalizer with NFC normalization and SHA-1 hashing ... Create Canonicalizer class implementing Package Registry Standard v1.0.0 §6: - NFC normalization of all string values via unicodedata.normalize - Lexicographic key sorting for deterministic output - Lifecycle field stripping (version, release_date) - RFC-8785 canonical JSON serialization (no whitespace, sorted keys) - SHA-1 content-addressed hashing via compute_package_id() - Internal reference resolution with configurable resolver callback 10 integration test cases covering: - Complex content canonicalization with lifecycle stripping - Deterministic output across multiple runs - Key sorting enforcement - Unicode NFC normalization - PackageId computation for all 8 package types - Different content produces different output ISSUES CLOSED: #25	2026-06-10 11:33:50 +01:00
hurui200320	76c4c74201	feat(create_executor): implement create_executor() factory and Executor.execute() returning ActorResult ... - **`src/cleveractors/runtime_types.py`** — `ActorResult` and `NodeUsage` dataclasses extracted to break circular imports (CONTRIBUTING.md §Import Guidelines). 100% coverage. - **`src/cleveractors/runtime.py`** — `create_executor()` factory, `Executor` class with `execute()` dispatching to module-level functions in `runtime_dispatch`. Imports `_execute_*` at module level (no function-local imports). 100% coverage. - **`src/cleveractors/runtime_dispatch.py`** — Four dispatch functions: `_execute_llm`, `_execute_graph`, `_execute_tool`, `_execute_multi_actor`. All imports at module level except `from cleveractors.runtime import Executor` inside `_execute_multi_actor` (the only remaining circular dep — Executor calls _execute_multi_actor which creates a new Executor; cannot be resolved without further restructuring). 99.69% coverage (only the `if TYPE_CHECKING:` guard line is uncovered — not a real gap). - **`src/cleveractors/runtime_tokens.py`** — Token estimation via tiktoken with heuristic fallback. 100% coverage. Key design decisions and fixes: - **Circular import resolution**: `ActorResult`/`NodeUsage` moved to `runtime_types.py`; both `runtime.py` and `runtime_dispatch.py` import from it at module level. - **`messages` forwarded to `_execute_llm`**: builds `context={"conversation_history":[...]}` passed to `LLMAgent.process_message` for multi-turn support. - **`parallel_execution` aligned with `PureGraphConfig` default (True)**: reads from legacy `route` dict or v2.0 `routes.main` dict; defaults to True. - **Factory normalization**: always merges `actors` into `agents` (actors take precedence) so configs using both keys work correctly. - **Generic exception in agent creation re-raised as `ConfigurationError`**: no double-logging; exception message preserved in `ConfigurationError`. - **`cleveragents_block` guarded against `None`**: uses `or {}` coercion. - **Type annotations added**: `top_provider`, `top_model`, `top_sp`, `temperature_raw`, `max_tokens_raw`, `config_block`, `tools` in dispatch functions. - **No `finally` in `_execute_tool`**: documented with comment (ToolAgent has no cleanup). - **Dead step removed**: `step_rxe_invalid_node_type` was unreferenced and incorrect. - **BDD coverage**: `parallel_execution=False` override verified via `PureGraphConfig` constructor args; conversation history forwarding verified; AC7 immutability for multi-actor path verified. - ✅ `nox -e lint` — passes - ✅ `nox -e format` — passes - ✅ `nox -e typecheck` — passes (0 errors, 1 expected warning) - ✅ `nox -e unit_tests` — 2113 scenarios pass, 0 failures, 0 skipped - ✅ `nox -e integration_tests` — 76 tests pass - ✅ `nox -e coverage_report` — 97.21% (9870 covered, 283 missing, 10153 total) PR-modified files: `runtime_types.py` 100%, `runtime.py` 100%, `runtime_dispatch.py` 99.69% (1 uncoverable TYPE_CHECKING guard), `runtime_tokens.py` 100% ISSUES CLOSED: #13	2026-06-10 09:41:39 +00:00
CoreRasurae	3fc0a3fa57	test(coverage): add BDD unit tests to close coverage gap toward 97% ... Add comprehensive BDD Behave unit tests across six modules to improve line coverage. New feature files and step definitions for: - validation/_actor: graph/llm/tool/multi_actor config validation - runtime_tokens: token estimation with tiktoken and fallback heuristics - dynamic_router: ContentBasedCondition, RouterNode, graph extension - runtime: v2 route format, config blocks, conversation history, multi-actor - progress: ProgressBarManager remaining count resolution - route: MESSAGE_ROUTER rules to metadata in to_graph_config Also fix credential executor and graph cleanup step mocks to return proper tuple values matching updated runtime signatures. Refs: #39	2026-06-09 16:06:32 +01:00
hurui200320	f281fa3b24	feat(credentials): refactor LLMAgent/AgentFactory for per-request credential injection and extended provider routing ... Implements Wave 3 of the cleveractors-core integration (ADR-2026, ADR-2028). ## AgentFactory - Added optional credentials: dict[str, dict[str, str]] | None parameter to __init__. Stored as self.credentials. - When creating LLMAgent instances, the full credentials dict is forwarded via a new credentials constructor kwarg, threading per-request API keys without touching the stored actor config_dict. - Added explicit type annotation ReactiveAgentFactory: type[AgentFactory]. - _create_agent_instance raises ConfigurationError when credentials is not None but does not contain an entry for the requested provider (AC4 compliance). ## LLMAgent - Added optional credentials: dict[str, str] | None parameter to __init__. Stored as self._credentials; config_dict is never modified. - LangChain client construction deferred to first access (lazy init) via a chat_model property backed by _ensure_chat_model(). The property setter allows test-injection of mock models without going through the lazy-init path. - _create_chat_model() dispatches to credential-injection / standalone paths. - Extracted to llm_client.py and llm_providers.py for modularity. - Defined shared module-level constants DEFAULT_MODEL, DEFAULT_TEMPERATURE, DEFAULT_MAX_TOKENS to eliminate duplicated magic numbers across llm.py, runtime.py, and factory.py. - Added public credentials property exposing self._credentials. - Added type narrowing guard in chat_model property to satisfy static analysis. - Added temperature_override type validation in process_message. - Guarded cleanup() null-assignment with _chat_model_lock (TOCTOU fix). ## SSRF Prevention (ADR-2028) - _validate_base_url(): https-only, userinfo rejection, localhost rejection, raw IP literal rejection, numeric-hostname (dotted-hex/octal/compact) rejection, percent-decode loop with iteration cap (10), trailing-dot FQDN strip. - DNS resolution intentionally omitted (blocking call in async path). - Canonical URL reconstruction uses lowercased scheme. ## Credential Leak Prevention - All logger.exception(...) replaced with logger.debug(..., exc_info=True). - Error messages use type(e).__name__ instead of str(e). - process_message exception chains broken with from None to prevent LangChain authentication errors from leaking via __cause__. ## Runtime fixes - Executor._execute_llm: shallow copy for setdefault mutations. - Executor._execute_graph: restructured try/finally for resource cleanup. - Uses DEFAULT_MODEL, DEFAULT_TEMPERATURE, DEFAULT_MAX_TOKENS from llm.py. - Defensive copy of credentials dict in Executor.__init__ to prevent caller mutation after construction. ## Test improvements - Added BDD scenarios for llm_imports.py ImportError fallback branches using sys.modules manipulation with a custom import hook, exercised via subprocess. - Added BDD scenario for AgentFactory.create_agents_from_config() credential threading to all created agents. - Removed tautological dataclass test scenarios (NodeUsage, ActorResult) from runtime_coverage.feature. - Fixed misleading scenario title in runtime_coverage.feature (exception chain suppressed, not chained) and updated step to assert __cause__ is None. - Added CHANGELOG entry for ReactiveAgentFactory backward-compatibility alias. - Removed duplicate log emission in _check_known_provider_domain. ISSUES CLOSED: #12	2026-06-08 11:10:46 +00:00
CoreRasurae	85ebe29f8e	feat(registry): add PackageType, PackageId, PackageReference, PackageContent core types ... Package Registry Standard v1.0.0 (§3, §5) data types: - PackageType enum matching §3.2: ACTOR, GRAPH, STREAM, AGENT, TEMPLATE, SKILL, MCP, LSP - PackageId: frozen/hashable value object parsing pkg_<type>_<40-hex-sha1> format - PackageReference: frozen/hashable parser for REGISTRY, ID, and LOCAL reference formats - PackageContent: frozen wrapper for fetched package content with metadata - 34 Behave BDD scenarios covering valid parsing, error paths, hashability, str/repr ISSUES CLOSED: #23	2026-06-06 19:21:22 +01:00
CoreRasurae	ab17407576	test(runtime): add BDD coverage tests for runtime Executor API ... Add 16 Behave BDD scenarios covering: - create_executor() factory function and Executor initialization - Executor.execute() dispatch to LLM, graph, tool, multi-actor agents - _execute_llm credential injection and error handling - _execute_graph PureGraphConfig construction - _execute_tool tool configuration - _execute_multi_actor routing and error cases - _build_factory_config credential/limit injection - _estimate_tokens with tiktoken and fallback heuristic - NodeUsage and ActorResult dataclasses ISSUES CLOSED: #33 Refs: #33	2026-06-05 23:55:31 +01:00
CoreRasurae	92d5305a20	feat(registry): implement async RegistryClient using httpx ... Add async RegistryClient implementing the Package Registry Standard v1.0.0 HTTP API endpoints (§8) using httpx.AsyncClient. - GET /packages/{id} — retrieve package content by PackageId - GET /{type}/{ns}/{name}?version= — resolve references with version alias resolution (latest, vx, vX.Y.x, vX.x per §4.2) - GET /browse — discover published packages with type/namespace filters - GET /.well-known/cleverthis-packages — registry metadata discovery - Maps all 8 standard error types (§13.2) to typed exceptions - Supports anonymous reads (§9.1) and optional API key auth (§9.2) - Async context manager support (__aenter__/__aexit__) - 32 Behave BDD scenarios, 7 Robot Framework integration tests, ASV benchmarks, 100% registry module coverage ISSUES CLOSED: #24	2026-06-05 22:01:21 +00:00
hurui200320	149feae15f	feat(validate_dict): expose public validate_dict(config_dict, platform_limits) API ... Implement validate_dict() in cleveractors/validation.py and export it from cleveractors/__init__.py as the first router-facing API (ADR-2024, Wave 1). validate_dict(config_dict, platform_limits) -> dict[str, Any]: - Validates that both 'agents' and 'routes' top-level keys are present. - Validates each agent's type against the spec-defined vocabulary: {llm, tool, composite, chain, template_instance}. - Validates LLM agent 'provider' values against platform_limits['allowed_providers'] when present, or the default allowlist {openai, anthropic, google} when absent. - Validates route/edge field names: 'from'/'to' are rejected in favour of 'source'/'target' as required by the spec-conformant format (ADR-2025). - Validates graph node types against the spec vocabulary: {start, end, agent, function, tool, conditional, subgraph, message_router}. - Validates stream operator types against the spec vocabulary (all defined types). - Enforces structural platform limits from platform_limits when present: max_graph_depth (DFS longest-path check), max_subgraph_depth (recursive subgraph-reference depth), max_total_nodes (total nodes across all graph routes). - Returns config_dict unchanged when all checks pass (identity contract). - Pure static validator: no file I/O, no env-var reads, no app construction. Review fixes (Cycle 1): - M3: Added isinstance(node_name, str) guard for graph node keys in _validate_graph_route, with BDD scenario for non-string node key. - m1: Added # pragma: no branch to _count_total_nodes defensive guard in _limits.py (matching the identical guards in _subgraph_children). - m2: Added # pragma: no branch to dead defensive continue branches in _validate_graph_route edge validation loop. - m3: Added dedicated BDD scenario and step for dangling source node (src not in nodes_raw branch), plus renamed existing scenario from "source node" to "target node" for accuracy. - m4: Deleted 4 lines of commented-out InlineYAMLJinja setup code in features/environment.py. - n1: Added depth <= 0 ValueError guard in build_subgraph_chain helper. - n2: Renamed scenario "Edge referencing a non-existent source node" to "Edge referencing a non-existent target node" (the step tested a target). - n3: Changed > to >= in both _MAX_DFS_STEPS and _MAX_SUBGRAPH_STEPS guards to prevent off-by-one allowing one extra iteration beyond documented ceiling. - n4: Added adjacency target deduplication in _compute_graph_depth to avoid wasting DFS step budget on parallel edges. - Spec review: M1 (entry_point optional) and M2 (start/end implicit injection) were verified against the authoritative spec at docs/specification.md in the cleveragents-webapp repo (develop branch). The spec does NOT support either claim; entry_point is required and start/end must be explicitly declared. No changes needed for M1/M2. Review fixes (Cycle 2): - M1: Deduplicated _subgraph_children yields with a seen set to prevent duplicate subgraph references from consuming DFS steps redundantly and causing false "too complex" errors. - M2: Changed adjacency value type from list[str] to set[str] in _compute_graph_depth for O(1) deduplication, reducing worst case from O(N^3) to O(N^2) for dense graphs. - m1: Added BDD scenario for LLM agent without provider field failing when default "openai" is not in custom allowlist. - m2: Added positive BDD scenarios for valid node types start, tool, conditional, message_router, function. - m3: Added positive BDD scenarios for additional stream operator types filter, transform, reduce. - m5: Added depth_cache memoization to _compute_subgraph_depth to avoid recomputing depths across graph routes sharing subgraph trees. - m6: Moved cleveractors.validation._limits import from module level into after_scenario, guarded by hasattr. - n1: Split validate_dict_routes_steps.py (was 473 lines) into validate_dict_routes_steps.py (generic route steps) and validate_dict_graph_route_steps.py (graph-specific steps). - n2: Updated _MAX_DFS_STEPS module comment to accurately state O(N+E) complexity and that dense graphs can exhaust the ceiling quickly. BDD Behave scenarios cover all requirements. Robot Framework integration tests (robot/validate_dict.robot) cover API usage. ISSUES CLOSED: #10	2026-06-05 09:23:44 +00:00
hurui200320	6b80be2117	feat(merge_configs): expose public merge_configs(*dicts) API per §3.1 deep-merge algorithm ... Implement a new module-level merge_configs() function in src/cleveractors/config_utils.py that exposes the §3.1 deep-merge algorithm as a public, importable API. Motivation: The internal ReactiveConfigParser._merge_configs() method mutates its base dict in place and only accepts two dicts. The CleverThis router (wave 2 of cleveragents/cleveragents-webapp#275) needs a public merge_configs() to combine platform base configs with actor configs from the database without side-effects. Implementation: - merge_configs(*dicts) accumulates into a fresh dict, applying each overlay using the §3.1 rules: * Key absent → add via deep copy. * Both mappings → deep-merge recursively (_apply_merge helper). * Both sequences → extend (existing + deep copy of new). * Otherwise → replace with deep copy of new value. - Runtime type guard: non-dict arguments raise TypeError immediately (fail-fast per CONTRIBUTING.md argument validation guidelines). - Zero-argument call returns {}. - Inputs are never mutated (copy.deepcopy used for all stored values). - Internal _merge_configs on ReactiveConfigParser is unchanged. Exports: merge_configs added to cleveractors/__init__.py import and __all__ so consumers can use 'from cleveractors import merge_configs'. Tests: - 15 Behave BDD scenarios in features/merge_configs.feature covering zero-args, single-dict deep copy (including nested identity), empty-dict arguments, TypeError on None, absent-key insertion, scalar override, recursive mapping merge, sequence append, three-dict chained merge, deep nesting, input mutation guard, result-mutation independence from inputs, and type-mismatch replacement. - 4 Robot Framework integration tests in robot/config.robot with strengthened assertions (len check, overlay immutability). - 5 ASV benchmarks with realistically sized three-way merge fixtures (50 keys each with overlap). Project files: Added CONTRIBUTORS.md per CONTRIBUTING.md §PR Process rule 8. Coverage: 97% (config_utils.py: 100%). Quality gates: ruff lint ✓, pyright strict ✓ (0 errors), all BDD/Robot tests ✓. ISSUES CLOSED: #11	2026-06-03 14:44:18 +00:00
CoreRasurae	9753b31c7e	test: add coverage gap tests improving coverage from 81.4% to 96.90% ... Add 13 BDD scenarios covering previously uncovered code paths: - SAFE_BUILTINS validation (sandbox.py: 0% → 100%) - ConfigurationError re-raise path (config.py: 99.3% → 100%) - CLI hello/main functions (cli.py: 72.7% → 90.9%) - GraphState message truncation (state.py: 98.7% → 100%) - ProgressBarManager update/context rendering (progress.py: 0% → 87.7%) - MessageRouter regex/exact/contains routing (message_router.py: 0% → 59.4%) - RoutingAdapter parse_routing_command (routing_adapter.py) - DynamicRouterNode pattern-based routing (dynamic_router.py) - EnhancedTemplateRegistry unknown template type (enhanced_registry.py: 99.2%) - CompositeAgent null-graph error path (composite.py: 97.8% → 98.6%) Cover routing_adapter.py (17% → 100%): all GOTO/ROUTE patterns, create_routing_node, create_conditional_router, dynamic config conversion. Cover dynamic_router.py (21% → 87%): execute with empty/dict/string messages, extract_message with colon parsing, config creation, graph extension with edge generation. Cover message_router.py (59% → 78%): regex, exact, contains, prefix, suffix match types, invalid regex handling, non-string message, set_state. Exercise Node._prepare_conversation_history with invalid configs, _runtime error paths, _execute_message_router with rules, _execute_agent with current_message and metadata propagation, _execute_function with dynamic_router, and _execute_conditional with content_contains/content_not_contains/content_starts_with and custom condition types. Improves nodes.py from 71.3% to 72.5%. Exercise PureGraphConfig, PureLangGraph init with dict/config, RxPyLangGraphBridge registration/connection/lookup, ReactiveStreamRouter operator creation and condition functions, ReactiveConfigParser config/route/graph parsing, ToolAgent tool execution with JSON, space-separated, single, file_read, progress_bar invocations, and ReactiveCleverAgentsApp init/dispose/visualization.	2026-06-02 20:02:01 +01:00
CleverThis Engineering	df468f821f	style: apply end-of-file and whitespace fixes across feature files	2026-05-27 21:41:03 +00:00
CleverThis Engineering	e7816ad9c4	fix: resolve all bandit, semgrep, and vulture security findings ... - Set jinja2 autoescape to select_autoescape(default_for_string=False) in all template engines to prevent HTML escaping of YAML config values while still configuring explicit autoescape behavior - Replace silent try/except/pass with logging.debug in langgraph bridge - Add logging before try/except/continue in langgraph nodes - Replace assert with proper if/raise ValueError in reactive config parser - Add # nosec annotations on intentional eval()/exec()/subprocess uses with justifications (sandboxed code execution for tool agent) - Add # nosemgrep annotations for intentional eval()/exec() in sandboxed tools - Create vulture whitelist for TYPE_CHECKING imports used in string annotations - Fix unused lambda variables with _ prefix	2026-05-26 23:12:05 +00:00
CleverThis Engineering	5bbda89386	style: fix all lint errors across source and test files ... - Fix bare excepts (replace with except Exception:) - Remove wildcard star imports from test step files - Add explicit imports for ConfigurationError, AgentCreationError - Add 'from err' to raise statements in except blocks - Fix loop variable binding with default arguments (B023) - Organize imports, fix trailing whitespace and blank line whitespace - Bump Python to 3.13 and update tool configurations	2026-05-26 22:21:50 +00:00
CleverThis Engineering	2223a16d48	feat: Minor code changes and linting and formatting issues fixes	2026-05-26 21:53:20 +00:00
CleverThis Engineering	d827c93d83	Chore: RxPy routing is not garunteed to work in run mode so added some guardrails to prevent that.	2026-05-26 21:53:20 +00:00
CleverThis Engineering	1001020fa1	Chore: Moved any pytest tests over to behave and removed pytest entierly	2026-05-26 21:53:19 +00:00
CleverThis Engineering	656dbd5626	feat: added --load-context argument to run command	2026-05-26 21:53:19 +00:00
CleverThis Engineering	05aef2f41a	Fix: JINJA2 templates are now being honored.	2026-05-26 21:53:19 +00:00
CleverThis Engineering	aa75d0978f	Fix: Verbosity wasnt being respected and errors were still leaking to the top, this is fixed	2026-05-26 21:53:19 +00:00
CleverThis Engineering	c26197e90a	Feat: added verbosity flag to control logging output	2026-05-26 21:53:19 +00:00
CleverThis Engineering	9a9aa04745	Fixed linting and typechecking, tox fully works now	2026-05-26 21:53:19 +00:00
CleverThis Engineering	3da2958a16	tests: some improved code coverage	2026-05-26 21:53:19 +00:00
CleverThis Engineering	44a53a16aa	feat: Added temperature override for run and interactive commands for better testing	2026-05-26 21:53:19 +00:00
CleverThis Engineering	57254f57e5	fix: Fixed many of the remaining issues and got a mostly working integration test for scientific paper writer	2026-05-26 21:53:19 +00:00
CleverThis Engineering	d6477288fe	test: Added vetter coverage to network and decorators	2026-05-26 21:53:19 +00:00
CleverThis Engineering	82ea476c5f	feat: Added context tracking when using via run, and added context command for managing it	2026-05-26 21:53:18 +00:00