2183 Commits

This Branch

This Branch

All Branches

Author	SHA1	Message	Date
brent.edwards	d4cbbaa41f	fix(e2e): use Extract JSON From Stdout for WF05 decision tree parsing ... WF05 was manually parsing the plan tree JSON output without unwrapping the CLI envelope, causing the decision_id count to return 0. The envelope wraps the decision array in a 'data' key. Updated WF05 to use Extract JSON From Stdout which automatically unwraps the envelope, ensuring the decision walker operates on the actual decision array instead of the envelope structure. Also added a documentation note to Extract JSON From Stdout explaining which tests need refactoring to use this keyword instead of custom JSON extraction logic.	2026-04-26 17:41:39 +00:00
brent.edwards	8bce7f45d0	fix(e2e): restore M5/WF14 tests broken by JSON envelope and stale config ... Two independent root causes were causing 10 non-quota E2E failures. **M5 acceptance (9 tests + 1 cascade):** `format_output(..., "json")` wraps all CLI JSON output in the spec-required envelope `{"command":…,"status":"ok","exit_code":0, "data":{…},…}`. The M5 tests were looking for payload fields (`total_tokens`, `resolved_view`, `acms_config`, `tier_metrics`, `plan_id`) at the top level of the extracted JSON object, but they live inside `data`. Fix: update `Extract JSON From Stdout` in `common_e2e.resource` to auto-unwrap the envelope when both `exit_code` and `data` keys are present, so callers receive the payload dict transparently. Output from helper scripts that do not use the CLI envelope (e.g. the WF04 snapshot helper) is returned unchanged. Also includes the already-staged m5_acceptance.robot rework that fixed the argument-passing bug (args were previously joined into single space-containing strings instead of passed individually to Run CLI), removed stale `tdd_expected_fail` tags, and standardised indentation. **WF14 server mode (1 test):** `~/.cleveragents/config.toml` is shared across all parallel pabot workers and across nox sessions (the `CLEVERAGENTS_HOME` env var isolates the SQLite DB but not the config file path, which is hardcoded to `Path.home() / ".cleveragents"`). A previous `server_stubs.robot` run wrote `server.url = "https://stub.example.com"` to that file. The `config set` CLI command could not overwrite the entry because tomllib parses TOML dotted keys as nested dicts (`{"server": {"url": "…"}}`), causing `config set` to write a new flat key alongside the original nested entry rather than replacing it. Fix: add `WF14 Clean Global Config` — a keyword that uses Python/tomlkit directly to remove `server.url`, `server.token`, and `server.namespace` from the global config TOML, handling both nested (`[server]` table) and flat quoted-key representations. The keyword is called from both `WF14 Suite Setup` (so any stale value is removed before the test's `config set` runs) and `WF14 Suite Teardown` (so subsequent runs start clean). Failures are silently ignored — a missing config file is normal on a fresh environment. ISSUES CLOSED: #8459	2026-04-26 17:41:39 +00:00
brent.edwards	84f5d6eb6b	fix(plan): wire sandbox_root into plan execute pipeline ... Fixes three sub-bugs that caused LLM-generated files to be silently lost during plan execute: Bug 1 — Duplicate sandbox branch on second execute call: _create_sandbox_for_plan() was called before phase detection, causing a fatal duplicate-branch error when plan execute was invoked a second time on a plan already in Execute/complete. Moved sandbox creation inside the Execute/QUEUED block so repeated invocations skip it safely. Bug 2 — Changeset entries discarded after execute: The lightweight ChangeSet from LLMExecuteActor was never persisted. Now serialised to plan.error_details["changeset_entries_json"] by PlanExecutor._run_execute_with_stub(); PlanApplyService._resolve_changeset() reconstructs a SpecChangeSet from this metadata when no changeset_store is wired, so plan diff shows the generated file changes. Bug 3 — Wrong output format for plan execute --format plain: execute_plan() used the now-removed _execute_output_dict (envelope format) instead of _plan_spec_dict, inconsistent with plan use / plan apply. Switched to _plan_spec_dict; removed the dead _execute_output_dict function. Also removes tdd_expected_fail from the M1 acceptance E2E test and adds 6 new BDD regression scenarios (@tdd_issue_1313) covering all three fixes. ISSUES CLOSED: #1313	2026-04-26 17:41:13 +00:00
brent.edwards	d81b5853a8	fix(e2e): restore M5/WF14 tests broken by JSON envelope and stale config ... Two independent root causes were causing 10 non-quota E2E failures. **M5 acceptance (9 tests + 1 cascade):** `format_output(..., "json")` wraps all CLI JSON output in the spec-required envelope `{"command":…,"status":"ok","exit_code":0, "data":{…},…}`. The M5 tests were looking for payload fields (`total_tokens`, `resolved_view`, `acms_config`, `tier_metrics`, `plan_id`) at the top level of the extracted JSON object, but they live inside `data`. Fix: update `Extract JSON From Stdout` in `common_e2e.resource` to auto-unwrap the envelope when both `exit_code` and `data` keys are present, so callers receive the payload dict transparently. Output from helper scripts that do not use the CLI envelope (e.g. the WF04 snapshot helper) is returned unchanged. Also includes the already-staged m5_acceptance.robot rework that fixed the argument-passing bug (args were previously joined into single space-containing strings instead of passed individually to Run CLI), removed stale `tdd_expected_fail` tags, and standardised indentation. **WF14 server mode (1 test):** `~/.cleveragents/config.toml` is shared across all parallel pabot workers and across nox sessions (the `CLEVERAGENTS_HOME` env var isolates the SQLite DB but not the config file path, which is hardcoded to `Path.home() / ".cleveragents"`). A previous `server_stubs.robot` run wrote `server.url = "https://stub.example.com"` to that file. The `config set` CLI command could not overwrite the entry because tomllib parses TOML dotted keys as nested dicts (`{"server": {"url": "…"}}`), causing `config set` to write a new flat key alongside the original nested entry rather than replacing it. Fix: add `WF14 Clean Global Config` — a keyword that uses Python/tomlkit directly to remove `server.url`, `server.token`, and `server.namespace` from the global config TOML, handling both nested (`[server]` table) and flat quoted-key representations. The keyword is called from both `WF14 Suite Setup` (so any stale value is removed before the test's `config set` runs) and `WF14 Suite Teardown` (so subsequent runs start clean). Failures are silently ignored — a missing config file is normal on a fresh environment. ISSUES CLOSED: #8459	2026-04-26 17:40:43 +00:00
HAL9000	3975ad244c	docs(spec): fix invariant reconciliation, add ACMS thread safety, v3.6.0 section, ADR-049 ... - Fix invariant glossary entry: reconciliation now documented as occurring at each phase boundary (before Strategize, Execute, Apply, and after Apply) rather than only at start of Strategize (closes #9899) - Add Multi-Phase Invariant Enforcement note to Plan Lifecycle section - Add ACMS Thread Safety subsection documenting threading.RLock concurrency contract for ContextAssemblyPipeline and all context stores (closes #9859) - Append M7: Advanced Concepts and Deferred Features (v3.6.0) milestone spec section covering advanced invariant lifecycle, ACMS observability, plan hierarchy enhancements, and CLI communication pattern migration - Add ADR-049: CLI Communication Pattern documenting local CLI exemption from A2A boundary requirement and M9 migration path	2026-04-26 17:09:22 +00:00
HAL9000	0f0efe02e3	docs: update examples.json index with actor management workflow entry ... ISSUES CLOSED: #7539	2026-04-26 16:52:36 +00:00
HAL9000	3cfa344110	docs(spec): align checkpoint trigger names and config key path with implementation	2026-04-26 13:50:38 +00:00
HAL9000	9f583fffeb	ci: retrigger CI pipeline	2026-04-26 13:01:45 +00:00
HAL9000	4c88458519	ci: retrigger CI pipeline	2026-04-26 13:01:45 +00:00
HAL9000	da214b101d	ci: retrigger CI pipeline ... Retrigger CI to resolve transient docker build failure. All code quality gates (lint, typecheck, unit_tests, integration_tests, coverage) pass. The docker failure is a CI infrastructure issue unrelated to this documentation-only PR.	2026-04-26 13:01:45 +00:00
HAL9000	d6f4395869	docs(agents): remove extraneous blank line in CHANGELOG entry ... Clean up double blank line in the CHANGELOG.md entry for the PR merge supervisor criteria documentation (#8107). ISSUES CLOSED: #8107	2026-04-26 13:01:45 +00:00
HAL9000	7c21ca0d6f	docs(agents): sync pr-merge-pool-supervisor merge criteria with runtime behaviour ... Add Ten Merge Criteria section to pr-merge-pool-supervisor.md documenting all 10 criteria the supervisor checks at runtime. Previously only 6 criteria were documented, creating inconsistency with actual runtime behaviour. Also update CHANGELOG.md and CONTRIBUTORS.md. Closes #8107 # Conflicts: # CONTRIBUTORS.md	2026-04-26 13:01:44 +00:00
HAL9000	087314843a	ci: remove benchmark-regression job from PR workflow (already in scheduled) ... The benchmark-regression job was supposed to be moved to the dedicated benchmark-scheduled.yml workflow (commit 44fc7411, issue #9040), but was accidentally left in master.yml. This caused every PR to run a 1h+ ASV benchmark suite on the docker-benchmark runner, blocking PR feedback. The benchmark-regression job now lives exclusively in .forgejo/workflows/benchmark-scheduled.yml and runs on a nightly schedule (2 AM UTC) and on workflow_dispatch. The benchmark-publish job (push to master) remains in master.yml as intended.	2026-04-26 12:35:39 +00:00
HAL9000	05b58630bf	style(a2a): apply ruff format to integration test helper ... Apply ruff format to robot/helper_a2a_session_plan_lifecycle_integration.py to fix CI lint failure. Changes are purely formatting (whitespace, line wrapping) with no logic modifications. ISSUES CLOSED: #10032	2026-04-26 12:35:39 +00:00
HAL9000	d306e6e33c	test(a2a): add integration tests for full A2A session and plan lifecycle ... - Added robot/a2a_session_plan_lifecycle_integration.robot: Robot Framework integration test suite with 7 test cases covering the full A2A session and plan lifecycle, event queue publish/subscribe, guard enforcement (budget cap, tool call limit, denylist), and plan rollback with session state consistency. - Added robot/helper_a2a_session_plan_lifecycle_integration.py: Helper script with real service instances (no mocks) using in-memory SQLite, wiring up A2A facade, session service, plan lifecycle service, event queue, and autonomy guardrail service. ISSUES CLOSED: #10032	2026-04-26 12:35:39 +00:00
HAL9000	446f447c03	style(test): apply ruff format to tdd_lsp_path_containment_steps.py	2026-04-26 12:19:34 +00:00
HAL9000	e619c1be58	test(lsp/runtime): fix TDD tag to reference bug issue #10490 and add safe-file scenario ... Corrected @tdd_issue_10489 → @tdd_issue_10490 so the tag references the actual bug issue (path traversal vulnerability) rather than the TDD issue itself. Added a positive scenario verifying _read_file can still read files inside the workspace, as suggested in review. ISSUES CLOSED: #10489	2026-04-26 12:19:34 +00:00
HAL9000	b5110761c2	lsp/runtime: add failing test proving LspRuntime._read_file has no workspace path containment check ... Added features/tdd_lsp_path_containment.feature: Introduces a TDD Behave feature with a @tdd_expected_fail scenario to demonstrate that LspRuntime._read_file() does not enforce workspace path containment, guiding a fix. Added features/steps/tdd_lsp_path_containment_steps.py: Provides step definitions for executing the new test. ISSUES CLOSED: #10489	2026-04-26 12:19:34 +00:00
HAL9000	1b515d529c	fix(tests): align LLM trace test session factory with flush-based save ... Use a shared session in test setup so that data flushed (but not committed) by LLMTraceRepository.save() remains visible to subsequent read operations within the same test scenario. This matches the production UnitOfWork pattern where a single session is shared across repository calls. Also fixes ruff format violations (single quotes, string concatenation) that caused the CI lint gate to fail. ISSUES CLOSED: #10034	2026-04-26 11:19:10 +00:00
HAL9000	2a5a37d774	fix: remove type: ignore comments and use proper method assignment ... Removed type: ignore[method-assign] comments from spy function assignments in test code. Replaced direct method assignment with object.__setattr__ to properly handle method replacement without type suppression, maintaining code quality standards.	2026-04-26 11:19:10 +00:00
HAL9000	b4b6fe6484	fix: replace session.commit() with session.flush() in LLMTraceRepository.save() ... Refactor: replace session.commit() with session.flush() in LLMTraceRepository.save() to ensure changes are persisted within the UnitOfWork without prematurely committing the database transaction. - Updated LLMTraceRepository.save() to call session.flush() instead of session.commit() in src/cleveragents/infrastructure/database/llm_trace_repository.py. - Added two new BDD scenarios to features/llm_trace.feature: - 'Repository save() calls flush not commit' to verify save() uses flush not commit. - 'LLM trace rolled back when UnitOfWork transaction rolls back' to verify rollback. - Added corresponding step definitions to features/steps/llm_trace_steps.py. ISSUES CLOSED: #10034	2026-04-26 11:19:10 +00:00
HAL9000	927c5a4681	chore(ci): optimize Dockerfile layer order to cache Python dependency installation ... Reorder COPY and RUN instructions in both Dockerfile and Dockerfile.server to leverage Docker's layer caching for the dependency installation step. Previously, pyproject.toml, README.md, and src/ were all copied together before running uv pip install, meaning any source code change would invalidate the dependency layer cache and force a full reinstall. The new pattern copies only the dependency manifests (pyproject.toml and uv.lock) first, installs the build tool in a cached layer, then copies README.md and src/ for the actual wheel build. This ensures the uv pip install step is only re-executed when pyproject.toml or uv.lock change, not on every source code change. ISSUES CLOSED: #1667	2026-04-26 10:02:20 +00:00
HAL9000	4f2122cd9e	docs: add action module API reference, update CHANGELOG and nav ... - Add docs/api/action.md: full API reference for ActionConfigSchema and ActionArgumentSchema — field tables, factory methods, camelCase compat mapping, env-var interpolation, database integrity note, and complete YAML example - Update docs/api/index.md: add cleveragents.action entry to module index - Update mkdocs.yml: add 'Action Schema: api/action.md' to the API Reference nav - Update CHANGELOG.md [Unreleased] Fixed: document plan use UNIQUE constraint violation fix (#4174, #4197) - Clarify from_yaml_file ValueError semantics, document inputs_schema type as dict[str, Any] | None, and update A2A ASV benchmarks to pass method= requests after JSON-RPC migration ISSUES CLOSED: #7472	2026-04-26 09:39:17 +00:00
HAL9000	8320f9f4cb	fix: correct trusted automation profile description to match spec ... ISSUES CLOSED: #9156	2026-04-26 09:12:01 +00:00
HAL9000	48dd67eb84	test: add TDD bug-capture test for #991 — AuditService TOCTOU race ... Rebase onto latest master to resolve CHANGELOG.md merge conflict. All test files unchanged from the approved PR commit (6c411be1). CHANGELOG entry added to the ### Added section under ## [Unreleased]. ISSUES CLOSED: #1095	2026-04-26 08:50:22 +00:00
HAL9000	263cbfa45f	ci: re-trigger CI after transient typecheck runner failure ... The previous CI run failed with a 5-second typecheck job failure, indicating a transient runner issue rather than a code problem. Pyright reports 0 errors locally (only checks src/, not features/). All quality gates pass: lint, typecheck, unit_tests.	2026-04-25 04:21:23 +00:00
HAL9000	ef5b8bf986	test(agents/graphs/auto_debug): add expected-fail test for _analyze_error in-place state mutation ... Added a TDD-style test for issue #10494 by introducing a new features/tdd_auto_debug_analyze_error_mutation.feature with a scenario tagged @tdd_issue @tdd_issue_10494 @tdd_expected_fail, and implemented Behave steps in features/steps/tdd_auto_debug_analyze_error_mutation_steps.py. The test captures the bug that _analyze_error mutates the state in-place and returns the full state object instead of a dict of updates, violating the LangGraph node contract. The @tdd_expected_fail tag inverts the test outcome so CI remains green while the bug exists. ISSUES CLOSED: #10494	2026-04-25 04:21:23 +00:00
HAL9000	a47e88016f	fix(agents): add PR diff and file list permissions to implementation-worker ... The implementation-worker has a PR fix mode that needs to read PR changes to understand what is failing. However, it lacked forgejo_list_pull_request_files and forgejo_get_pull_request_diff permissions, forcing it to clone the entire repository just to inspect what files changed. The pr-reviewer agent already has both permissions. This change aligns the implementation-worker permissions with its actual usage patterns and reduces unnecessary full-repo clones in PR fix mode. Also adds curl, printenv, and echo bash permissions that were inadvertently omitted from the permission block. These are needed for the agent to make Forgejo API calls, check environment variables, and perform basic shell operations. ISSUES CLOSED: #8175	2026-04-25 04:00:45 +00:00
HAL9000	49ecf7ff8f	fix(agents): add PR diff and file list permissions to implementation-worker ... The implementation-worker has a PR fix mode that needs to read PR changes to understand what is failing. However, it lacked forgejo_list_pull_request_files and forgejo_get_pull_request_diff permissions, forcing it to clone the entire repository just to inspect what files changed. The pr-reviewer agent already has both permissions. This change aligns the implementation-worker permissions with its actual usage patterns and reduces unnecessary full-repo clones in PR fix mode. Also adds curl, printenv, and echo bash permissions that were inadvertently omitted from the permission block. These are needed for the agent to make Forgejo API calls, check environment variables, and perform basic shell operations. ISSUES CLOSED: #8175	2026-04-25 04:00:45 +00:00
HAL9000	1732436edb	fix(agents): add PR diff and file list permissions to implementation-worker ... The implementation-worker has a PR fix mode that needs to read PR changes to understand what is failing. However, it lacked forgejo_list_pull_request_files and forgejo_get_pull_request_diff permissions, forcing it to clone the entire repository just to inspect what files changed. The pr-reviewer agent already has both permissions. This change aligns the implementation-worker permissions with its actual usage patterns and reduces unnecessary full-repo clones in PR fix mode. Also adds curl, printenv, and echo bash permissions that were inadvertently omitted from the permission block restructuring. These are needed for the agent to make Forgejo API calls, check environment variables, and perform basic shell operations. ISSUES CLOSED: #8175	2026-04-25 04:00:45 +00:00
HAL9000	f7deab5147	chore(agents): add concrete milestone API example and ISSUES CLOSED footer to agent-evolution-worker ... Added a concrete curl example showing how to query the earliest open milestone via the Forgejo API at PR creation time, and added the ISSUES CLOSED footer requirement to the PR compliance checklist in agent-evolution-worker.md. Also updated CHANGELOG.md to reflect these additions. ISSUES CLOSED: #8370	2026-04-25 03:14:00 +00:00
HAL9000	d17d4eebb5	chore(agents): fix CHANGELOG.md stale milestone reference in agent-evolution-worker entry ... Updated the CHANGELOG.md [Unreleased] > Changed entry for #8370 to accurately reflect the actual implementation: dynamic earliest-open-milestone assignment via Forgejo API (not hardcoded v3.2.0 ID: 105), and Type/Task label (not Type/Chore which does not exist in the org label system). ISSUES CLOSED: #8370	2026-04-25 03:14:00 +00:00
HAL9000	c4122f877a	chore(agents): add CONTRIBUTING.md compliance steps to agent-evolution-worker ... ISSUES CLOSED: #8370	2026-04-25 03:14:00 +00:00
brent.edwards	92c3bd8732	fix(tests): update resource_dag.robot SQLite pool and cycle detection types ... Replaced direct create_engine usage with StaticPool-based connection in resource_dag.robot to prevent SQLite connection sharing issues in tests. Updated cycle detection test to use distinct resource types (git-checkout and fs-directory) instead of the same type for both resources, improving test coverage of cross-type cycle detection. Split from PR #1204 per reviewer request for atomic commits. ISSUES CLOSED: #1226	2026-04-25 01:55:37 +00:00
hamza.khyari	60c50d3c9f	Merge pull request 'fix(plan): implement error recovery for execute phase per spec §35958 and §18323' (#10844) from bugfix/reexecute-errored-plan into master ... Reviewed-on: #10844 Reviewed-by: Luis Mendes <luis.mendes@cleverthis.com>	2026-04-24 23:03:44 +00:00
hamza.khyari	12fc9e089f	fix(plan): implement error recovery for execute phase per spec §35958 and §18323 ... Two recovery paths for execute/errored plans using the domain model's classify_error() for canonical error classification: 1. Transient failures (§28530/§35958): rate limit, timeout, connection errors — reset to execute/queued and re-execute with the same strategy. Preserves strategy_decisions_json for decision hierarchy reconstruction. 2. Non-transient failures (§18323-18329): strategy constraint, authentication, unknown — delegate to service.try_auto_revert_from_execute() which enforces MAX_REVERSIONS loop guard, increments reversion_count, records a reversion decision, and respects the delete_content automation threshold. Error findings are redacted via redact_error_details() and stored as JSON for the strategy actor (§18329). Error classification uses the domain model's classify_error() and ErrorCategory enum — single source of truth, no parallel frozenset. AuthenticationError correctly classified as non-transient per domain model. ISSUES CLOSED: #10843	2026-04-24 22:46:41 +00:00
freemo	43fdc31624	chore(agents): improve ca-continuous-pr-reviewer — skip PRs with unchanged HEAD since last review	2026-04-24 22:13:30 +00:00
CoreRasurae	9c5f19854d	ci(docker): Ensure CI uses CT docker proxy to workaround Docker ratelimit issues ... ISSUES CLOSED: #10856	2026-04-24 21:58:51 +01:00
HAL9000	e3212b5f8a	fix: reject migrations when prompt fails unexpectedly ... Handle unexpected exceptions from the interactive migration prompt by logging the failure and rejecting instead of proceeding. ISSUES CLOSED: #7503	2026-04-24 19:01:56 +00:00
HAL9000	4beeb747d0	test: add regression for unexpected migration prompt failure ... ISSUES CLOSED: #7503	2026-04-24 19:01:56 +00:00
HAL9000	122a9e70a1	chore(ci): remove build artifacts from tracking and add test_reports to .gitignore ... Remove test_reports/summary.txt and test_reports/test_results.json from git tracking as they are build artifacts that should not be committed to the repository. Add test_reports/ to .gitignore to prevent future accidental commits of these files. ISSUES CLOSED: #7503	2026-04-24 19:01:56 +00:00
HAL9000	36dc29b3ec	docs(contributors): add HAL 9000 contribution details for migration safety fix	2026-04-24 19:01:56 +00:00
HAL9000	49ed394d11	fix(migration): reject migrations on prompt failure instead of auto-approving ... Fixed MigrationRunner._default_prompt_for_migration silently auto-approving destructive database migrations when the interactive prompt raised any exception. The bare 'except Exception' handler was swallowing all errors and returning True (auto-approve), which could apply destructive schema migrations to production databases without user consent when stdin is broken, typer is unavailable, or any other prompt failure occurs. Changes: - Narrow exception handler from 'except Exception' to 'except (OSError, EOFError)' to only catch genuine non-interactive environment signals - Re-raise KeyboardInterrupt so Ctrl-C always works - Return False (reject) instead of True (auto-approve) on prompt failure - Log at WARNING level instead of DEBUG so the rejection is visible - Non-interactive environments (stdin not a TTY) now also return False by default - Updated docstring to document the new safe-default behavior - Added BDD regression tests for all new code paths - Added TDD feature file tdd_migration_prompt_auto_approve_7503.feature ISSUES CLOSED: #7503	2026-04-24 19:01:56 +00:00
HAL9000	8071539d1b	chore(ci): trigger CI re-run for transient infrastructure failure	2026-04-24 13:33:16 +00:00
HAL9000	b65f33af60	chore(ci): trigger CI re-run for transient infrastructure failure ... Previous CI run 14880 failed all jobs immediately (0-1 seconds) due to a CI infrastructure issue, not a code problem. The same failure pattern is observed on the master branch (run 14932), confirming this is a transient infrastructure outage. All quality gates pass locally: - lint ✓ (ruff check: all checks passed) - format ✓ (ruff format --check: 1974 files already formatted) - typecheck ✓ (pyright: 0 errors, 3 warnings — pre-existing optional deps) Refs: #10455	2026-04-24 11:05:02 +00:00
HAL9000	a567a78948	style(memory): apply ruff formatting to entity persistence files ... Apply ruff auto-formatting to fix CI lint gate failures. Two files had formatting issues detected by ruff format --check: - features/steps/tdd_memory_service_entity_persistence_steps.py - src/cleveragents/application/services/memory_service.py Changes are purely cosmetic: line wrapping adjustments, parenthesization style, and string formatting alignment per ruff rules. Refs: #10455	2026-04-24 11:05:02 +00:00
HAL9000	719315a30b	fix(memory): implement entity persistence in MemoryService or remove stub ... Added Behave scenario to ensure EntityStore metadata and mention counts persist across restarts. Implemented new step definitions to track metadata updates and assert mention counts. Kept existing persistence behavior unchanged while expanding TDD coverage to validate regression safety. ISSUES CLOSED: #10455	2026-04-24 11:05:02 +00:00
HAL9000	0f4824c26c	fix(memory): implement entity persistence in MemoryService or remove stub ... Implemented EntityStore._load_from_persistence() to load entities from a SQLite database on initialization using SQLAlchemy. Implemented EntityStore._persist_if_needed() to write dirty entities to the database using SQLAlchemy. Both methods raise RuntimeError on persistence failures instead of silently failing, eliminating the silent data-loss bug. Added import json to support entity serialization/deserialization. Added TDD Behave feature file with 4 scenarios verifying entity round-trip through persistence. All 4 TDD scenarios pass, all existing tests continue to pass. Coverage at 97.1%. ISSUES CLOSED: #10455	2026-04-24 11:05:02 +00:00
hamza.khyari	f0923e08ba	feat(plan): create per-project sandboxes for multi-project plans ... Per spec §19310-19312, each resource gets its own sandbox and Apply commits each sandbox separately. When a plan is linked to multiple projects with git-checkout resources: - _create_sandbox_for_plan creates a worktree for EACH resource (not just the first), returning a list of _SandboxInfo objects - LLM output is written to the first resource's worktree (primary) - _route_sandbox_files_to_worktrees moves files that belong to other resources into their worktrees by matching against each resource's git ls-files output - Each worktree is committed independently - _apply_sandbox_changes merges each resource's worktree separately, showing per-resource Apply Summary and Sandbox Cleanup panels - Partial apply: if one resource's merge fails, others still proceed Single-resource plans are fully backward compatible — same behavior as before. ISSUES CLOSED: #7270	2026-04-24 10:32:15 +00:00
HAL9000	663a6d2397	docs(decomposition): enhance docstrings and add spec section for selective subtree recomputation	2026-04-24 08:26:59 +00:00