diff --git a/docs/timeline.md b/docs/timeline.md
index 03a9cadd..434a57dc 100644
--- a/docs/timeline.md
+++ b/docs/timeline.md
@@ -10,14 +10,14 @@ The following chart shows all 29 epics across 9 legendary workstreams, 7 milesto
 @startgantt
 
 title CleverAgents Core — Epic-Level Project Schedule
-footer Generated 2026-02-23 | 29 Epics | 9 Legendaries | 7 Milestones | 6 Developers | ~1350 SP
+footer Generated 2026-02-26 | 29 Epics | 9 Legendaries | 7 Milestones | 6 Developers | ~1350 SP
 
 Project starts 2026-02-03
 saturday are closed
 sunday are closed
 printscale weekly zoom 2
 
-today is 2026-02-24
+today is 2026-02-26
 today is colored in #FF6666
 
 <style>
@@ -155,8 +155,8 @@ ganttDiagram {
 
 [Concurrency + Cleanup (#365)] as [CONCUR] on {Luis} requires 2 days
 [CONCUR] starts 2026-02-23
-[CONCUR] is 0% completed
-[CONCUR] is colored in #E8E8E8/Silver
+[CONCUR] is 100% completed
+[CONCUR] is colored in PaleGreen/SeaGreen
 
 ' ================================================================
 '  PROJECTS & RESOURCES — WORKSTREAM B (Legendary #373 - DONE) — v3.1.0
@@ -189,17 +189,17 @@ ganttDiagram {
 
 [Actor + Skill Schema C0 (#354)] as [C0] on {Jeff} {Aditya} {Luis} {Brent} requires 7 days
 [C0] starts 2026-02-11
-[C0] is 93% completed
-[C0] is colored in LightSkyBlue/SteelBlue
+[C0] is 100% completed
+[C0] is colored in PaleGreen/SeaGreen
 
 [Actor Runtime + Tool Exec (#355)] as [M1] on {Jeff} {Hamza} {Luis} {Brent} requires 3 days
 [M1] starts at [PROJ]'s end
-[M1] is 34% completed
+[M1] is 60% completed
 [M1] is colored in LightSkyBlue/SteelBlue
 
 [Actor Graphs + Tool Sources (#356)] as [M2] on {Aditya} {Jeff} {Luis} {Brent} {Hamza} requires 5 days
 [M2] starts at [M1]'s end
-[M2] is 5% completed
+[M2] is 15% completed
 [M2] is colored in LightSkyBlue/SteelBlue
 
 ' ── v3.2.0 MILESTONE ─────────────────────────────────────────
@@ -219,7 +219,7 @@ ganttDiagram {
 
 [Decisions + Validations M3 (#357)] as [M3] on {Luis} {Hamza} {Brent} {Jeff} requires 4 days
 [M3] starts at [M2]'s end
-[M3] is 10% completed
+[M3] is 50% completed
 [M3] is colored in LightSkyBlue/SteelBlue
 
 ' ── v3.3.0 MILESTONE ─────────────────────────────────────────
@@ -240,12 +240,12 @@ ganttDiagram {
 
 [Corrections + Checkpoints M4 (#358)] as [M4] on {Luis} {Jeff} {Brent} {Hamza} requires 3 days
 [M4] starts at [M3]'s end
-[M4] is 27% completed
+[M4] is 40% completed
 [M4] is colored in LightSkyBlue/SteelBlue
 
 [Security + Safety Hardening (#362)] as [SEC] on {Luis} requires 3 days
 [SEC] starts 2026-03-02
-[SEC] is 20% completed
+[SEC] is 60% completed
 [SEC] is colored in LightSkyBlue/SteelBlue
 
 [Provider Fixes + Runtime (#363)] as [PROV] on {Luis} requires 2 days
@@ -285,7 +285,7 @@ ganttDiagram {
 
 [Large Project Autonomy (#369)] as [LARGE] on {Luis} {Brent} {Jeff} requires 4 days
 [LARGE] starts 2026-03-17
-[LARGE] is 52% completed
+[LARGE] is 55% completed
 [LARGE] is colored in LightSkyBlue/SteelBlue
 
 ' ── v3.5.0 MILESTONE ─────────────────────────────────────────
@@ -390,8 +390,8 @@ legend right
   **Risk Register**
   |= Epic |= Completion |= Risk |= Blocker |
   | PRE_TEST (#24) | 0% | Medium | Test framework gap |
-  | M1 (#355) | 34% | Medium | Hamza/Luis blocked |
-  | SEC (#362) | 20% | High | Security-critical |
+  | M1 (#355) | 60% | Medium | MCP adapter + agent skills loader |
+  | SEC (#362) | 60% | Medium | Remaining security hardening |
   | PROV (#363) | 0% | Low | Provider fixes deferred |
   | A9 (#352) | 0% | Low | Config deferred to v3.4 |
   ----
@@ -401,7 +401,7 @@ legend right
   | <back:LightSkyBlue>    </back> | In Progress |
   | <back:#E8E8E8>    </back> | Not Started |
   | <back:Gold>  <>  </back> | Milestone Target |
-  | <back:#FF6666>  |  </back> | Today (2026-02-24) |
+  | <back:#FF6666>  |  </back> | Today (2026-02-26) |
   ----
   29 Epics | 9 Legendaries | 7 Milestones | 6 Developers
   ~1350 total story points | ~2700 estimated hours
@@ -428,7 +428,7 @@ saturday are closed
 sunday are closed
 printscale weekly zoom 2
 
-today is 2026-02-24
+today is 2026-02-26
 today is colored in #FF6666
 
 <style>
@@ -1933,7 +1933,7 @@ legend right
   | <back:LightSkyBlue>      </back> | Issue: In Progress |
   | <back:#E8E8E8>      </back> | Issue: Not Started |
   | <back:Gold>  <>  </back> | Milestone Target |
-  | <back:#FF6666>  |  </back> | Today (2026-02-24) |
+  | <back:#FF6666>  |  </back> | Today (2026-02-26) |
   ----
   213 Issues | 29 Epics | 9 Legendaries | 7 Milestones
   ~1350 total story points | ~2700 estimated hours
@@ -1948,13 +1948,15 @@ This section provides a high-level overview of the CleverAgents implementation r
 
 ### Current Status Summary
 
-As of Day 16 (2026-02-24), the project has **~144 closed issues out of ~259 assigned issues** across all milestones (72% overall completion rate). The foundational layers — domain models, persistence, YAML schemas, CLI commands, quality automation, and CI/CD — are substantially built out. Today's progress includes the **actor compiler** (Jeff, #158 merged — YAML to LangGraph StateGraph compilation), **agent skills discovery** (Jeff, #161 merged), **actor CLI alignment** (Jeff, #288 merged), **CLI plan/action extensions** (Jeff, #325 merged), and a **database parallelism fix** (Brent, #418 merged).
+As of Day 18 (2026-02-26), the project has **~168 closed issues out of ~225 assigned issues** across all milestones (~75% overall completion rate). A major productivity surge over Days 16-18 closed **65 issues in 3 days**, substantially advancing M1, M2, M3, and M4. The foundational layers — domain models, persistence, YAML schemas, CLI commands, quality automation, CI/CD, validation pipeline, and security hardening — are now mature.
 
-The primary blocker for M1 completion remains the **sandbox merge/apply pipeline** (#155, Luis — active branch) and the **E2E source-code smoke test suite** (#156, Brent). The M2 blockers are the **MCP adapter** (#159, Aditya — in review) and remaining **actor YAML extensions** (#157, Aditya — active branch). The actor compiler (#158) — previously the M2 critical-path blocker — was merged today.
+**Days 17-18 highlights**: Apply pipeline + validation-gated apply + diff review merged (Luis, PR #417 — #155, #175, #176, #178, #303, #186). Changeset persistence (Jeff, #163). Plan resume + concurrency locks (Jeff, #327, #328). Provider cost controls (Jeff, #324). Decision persistence (Hamza, PR #131 — #171). Async resource cleanup (Hamza, #321). Actor YAML hierarchy (Aditya, PR #432 — #157). Security hardening — template rendering + exception handling (Hamza, #319, #320). Brent QA batches: M1/M2 E2E suites (#156, #169, #326, #402, #403) and M3/M4 E2E suites (#179, #180, #404, #405, #187). CI concurrency groups (#452).
 
-!!! warning "Schedule Risk: 7-9 Days Behind"
+**M1 is essentially complete** — only #323 (remove FakeListLLM defaults) remains open. The M2 blockers are the **MCP adapter** (#159, Aditya), **agent skills loader** (#160, Aditya), and **skill persistence/CLI** (#166, #167, #168). The M3 target date is today; 3 issues remain (#172 decision recording, #173 invariant enforcement, #174 plan explain CLI).
 
-    The project is running **7-9 days behind** the original aggressive schedule. M1 (target Feb 15) is overdue by 9 days; M2 (target Feb 22) is overdue by 2 days. A major workload rebalancing from Luis (33 remaining issues, lowest velocity) to Jeff (highest velocity, capacity available) is being executed to close the gap. Any `MoSCoW/Could Have` items in M1-M2 should be deferred to avoid further slippage.
+!!! warning "Schedule Risk: 5-7 Days Behind (Narrowing)"
+
+    The project is running **5-7 days behind** the original aggressive schedule, improved from 7-9 days at Day 16. The 65-issue closure burst over Days 16-18 substantially narrowed the gap. M1 (target Feb 15) is overdue by 11 days but essentially complete. M2 (target Feb 22) has 6 remaining issues. M3 (target Feb 26) has 3 remaining issues. The workload rebalancing executed on Day 16 is paying off — Jeff's throughput has been exceptional, and the validation/security batch from Luis/Hamza landed cleanly.
 
 ### Parallel Workstreams
 
@@ -1962,14 +1964,14 @@ Development is organized into six concurrent workstreams plus continuous quality
 
 | Track | Focus Area | Status |
 | :---- | :--------- | :----- |
-| **A** — Plan Lifecycle | Action/plan domain models, persistence, CLI alignment, execute/apply wiring | Core models and persistence done; CLI extensions merged today (#325). Execute/apply pipeline in progress. Config service (#352) deferred to M5. |
-| **B** — Projects & Resources | Resource registry, project model, resource types, sandbox strategies, CLI | **Substantially complete.** Registry, models, CLI, handler runtime (git-checkout + fs-directory), DAG metadata all done. Auto-discovery remaining. |
-| **C** — Actors, Tools, Skills | Actor YAML schema/loader/compiler, tool registry/runtime, skill framework, MCP adapter, validation runner | **Actor compiler DONE** (#158 merged today). Agent skills discovery DONE (#161 merged today). Actor CLI aligned (#288 merged today). MCP adapter in review (Aditya PR #382). Agent skills loader in review (Aditya PR #416). Validation runner remaining. |
-| **D** — Decisions & Apply | Decision tree model, change tracking, diff review, apply pipeline, correction engine | Decision domain model and context snapshots done. **Decision persistence in review** (Hamza PR #131). Correction engine models done. Apply pipeline partially done; sandbox merge (#155) active. |
-| **E** — Subplans | Subplan domain model, orchestration, parallel execution, three-way merge | Domain model done; orchestrator and merge logic remaining |
+| **A** — Plan Lifecycle | Action/plan domain models, persistence, CLI alignment, execute/apply wiring | **Substantially complete.** Apply pipeline with validation gating merged (#155). Changeset persistence merged (#163). Concurrency locks + plan resume merged (#327, #328). Config service (#352) deferred to M5. |
+| **B** — Projects & Resources | Resource registry, project model, resource types, sandbox strategies, CLI | **Complete.** Registry, models, CLI, handler runtime (git-checkout + fs-directory), DAG metadata all done. |
+| **C** — Actors, Tools, Skills | Actor YAML schema/loader/compiler, tool registry/runtime, skill framework, MCP adapter, validation runner | Actor compiler, agent skills discovery, actor CLI, skill registry flattening, validation pipeline all DONE. Actor YAML hierarchy merged (#157). **MCP adapter** (#159, Aditya) and **agent skills loader** (#160, Aditya) remain. Skill persistence (#166) and skill CLI (#167, #168) remain. |
+| **D** — Decisions & Apply | Decision tree model, change tracking, diff review, apply pipeline, correction engine | **Major progress.** Decision persistence merged (#171). Validation-gated apply merged (#176). DoD gating merged (#178). Diff review artifacts merged (#303). Error recovery patterns merged (#186). Remaining: decision recording (#172), invariant enforcement (#173), plan explain CLI (#174). |
+| **E** — Subplans | Subplan domain model, orchestration, parallel execution, three-way merge | Domain model done; orchestrator and merge logic remaining. E2E M4 correction + subplan suites merged (#405). |
 | **F** — ACMS & Context | UKO ontology, CRP protocol, context strategies, assembly pipeline, hot/warm/cold tiers | Not yet started; scheduled for M5-M6 milestones |
-| **Q** — Quality & CI | Pre-commit hooks, Forgejo CI/CD, coverage enforcement (97%), security scanning | Complete. Database parallelism fix merged today (#418). |
-| **T** — Testing | Behave unit tests, Robot integration tests, ASV benchmarks, end-to-end smoke suites | Ongoing; CLI extension tests in review (Brent PR #420). E2E milestone suites (#402-#407) pending. Rui returns Feb 27. |
+| **Q** — Quality & CI | Pre-commit hooks, Forgejo CI/CD, coverage enforcement (97%), security scanning | Complete. CI concurrency groups and job timeouts added (#452). Security hardening: template rendering (#319) and exception handling (#320) merged. Async resource cleanup merged (#321). |
+| **T** — Testing | Behave unit tests, Robot integration tests, ASV benchmarks, end-to-end smoke suites | **Strong progress.** M1 E2E suite (#156), M2 E2E suite (#169, #403), M3 E2E suite (#179, #180, #404), M4 E2E suite (#405) all merged via Brent QA batches. M5 (#406) and M6 (#407) E2E suites remain. Rui returns Feb 27. |
 
 ### What Has Been Completed
 
@@ -1990,129 +1992,128 @@ The following major areas are fully or substantially implemented:
 - **Security hardening**: All `eval()`/`exec()` calls replaced with named operation registries, config security scanner detecting 15 disallowed patterns, AST-based code validation, Jinja2 sandboxed environments, secrets masking and validation.
 - **Agent skills discovery**: Filesystem scanning for Agent Skills Standard folders (SKILL.md with YAML front-matter), registering discovered skills as tools in ToolRegistry with configurable conflict handling (skip/error/replace), refresh hooks, and CLI integration (#161 merged 2026-02-24).
 - **CLI extensions**: Plan use command extended with --automation-profile, --invariant (repeatable), --strategy-actor/--execution-actor/--estimation-actor/--invariant-actor flags. Plan status/list and action show commands extended with profile, invariant, and actor override rendering across all output formats (#325 merged 2026-02-24).
-- **Quality automation**: Pre-commit hooks (12 hooks across 5 categories), Forgejo CI/CD pipeline with 10 nox-routed jobs, nightly quality monitoring, ADR compliance checking, 97% test coverage enforcement, comprehensive security scanning (Bandit + Semgrep + Vulture), and per-scenario temp database isolation for parallel Behave execution (#418 merged 2026-02-24).
+- **Apply pipeline with validation gating**: Sandbox merge/apply pipeline with conflict resolution (#155 merged 2026-02-25). Validation-gated apply (#176 merged 2026-02-25) with required/informational mode enforcement. Diff review artifacts (#303 merged 2026-02-25). DoD gating enforcement (#178 merged 2026-02-25). Error recovery patterns with CLI hints (#186 merged 2026-02-25).
+- **Decision persistence**: Decision persistence layer with database tables, repositories, and CRUD operations (#171 merged 2026-02-26, Hamza). Decision domain model, context snapshots, and influence DAG now fully persisted.
+- **Changeset persistence**: ChangeSet and diff artifact persistence with database migration and repository (#163 merged 2026-02-26, Jeff).
+- **Concurrency and resilience**: Plan and project concurrency locks with advisory locking (#327 merged 2026-02-25). Plan resume with state recovery and re-execution (#328 merged 2026-02-25). Provider cost controls and fallback routing (#324 merged 2026-02-25).
+- **Security hardening (Days 17-18)**: Template rendering hardened (#319 merged 2026-02-25). Explicit exception handling enforced (#320 merged 2026-02-25). Async resource cleanup and leak prevention (#321 merged 2026-02-26).
+- **Actor YAML hierarchy**: Extended hierarchical actor YAML schema with namespace inheritance, multi-level includes, and loader improvements (#157 merged 2026-02-26, Aditya).
+- **E2E milestone suites**: M1 source-code plan lifecycle suite (#156), M2 actor + tool source suite (#169), M1/M2 success criteria verification (#402, #403), M3 decision + validation suites (#179, #180, #404), M4 correction + subplan suite (#405) — all merged via Brent QA batches (2026-02-25 and 2026-02-26).
+- **Quality automation**: Pre-commit hooks (12 hooks across 5 categories), Forgejo CI/CD pipeline with 10 nox-routed jobs, nightly quality monitoring, ADR compliance checking, 97% test coverage enforcement, comprehensive security scanning (Bandit + Semgrep + Vulture), per-scenario temp database isolation for parallel Behave execution (#418 merged 2026-02-24), and CI concurrency groups with job timeouts (#452 merged 2026-02-26).
 
 ### What Remains To Be Done
 
 The majority of the specification's domain models, services, CLI commands, and infrastructure modules are now implemented. The following areas still require completion or have only partial implementations:
 
-- **MCP runtime adapter**: Live connection to MCP servers for external tool discovery and invocation at runtime. Aditya has active work on `develop-aditya-2` branch with PR #382 in review. Schema and domain models exist; the runtime bridge is pending. **This is the remaining M2 critical-path blocker** (actor compiler was merged 2026-02-24).
-- **Decision persistence**: Hamza has active work on `feature/m4-decision-persistence`. The domain model and context snapshots are now merged; the database persistence layer (tables, repositories, CRUD) is next.
-- **Decision recording service**: Wire decision recording into the strategize/execute phases so decisions are automatically captured during plan execution.
-- **Validation runner pipeline**: A dedicated module for executing required and informational validations attached to resources, with pass/fail gating on the apply phase.
+- **MCP runtime adapter**: Live connection to MCP servers for external tool discovery and invocation at runtime. Aditya's previous PR #382 merged the foundation; the agent skills loader (#160) and MCP adapter runtime (#159) remain. **This is the remaining M2 critical-path blocker.**
+- **Decision recording service**: Wire decision recording into the strategize/execute phases so decisions are automatically captured during plan execution (#172, Hamza).
+- **Invariant enforcement**: Runtime enforcement of invariant constraints with scoped precedence and reconciliation (#173, Jeff).
+- **Plan explain CLI**: Plan tree and explain commands for decision tree visualization (#174, Hamza).
+- **Skill persistence and CLI**: Persist flattened tool sets (#166, Jeff), skill CLI commands (#167, Aditya), MCP refresh hooks (#168, Aditya).
 - **Estimation actor logic**: The `estimation_actor` field is plumbed through Plan, Action, CLI, and service layers, but no actual estimator implementation exists to compute cost/risk estimates before execution.
 - **ACMS v1 full pipeline**: Context policy with phase-scoped views and the context service exist. Still pending: Universal Knowledge Ontology (UKO) with RDF-based resource representation, Context Request Protocol (CRP), pluggable context strategies, the full ten-component Context Assembly Pipeline, and hot/warm/cold tiered storage with per-actor scoped views.
 - **Large-project context management**: Indexing 10,000+ file codebases with bounded memory context assembly, skeleton compression for parent-to-child context propagation, and performance validation at scale.
-- **Sandbox merge/apply with validation gating**: The apply service exists but needs validation gating hooks and conflict resolution for the sandbox merge pipeline.
-- **E2E smoke test suites**: Per-milestone E2E verification suites (#402-#407) need to be written to validate milestone success criteria.
+- **E2E smoke test suites**: M5 (#406) and M6 (#407) E2E verification suites remain. M1-M4 suites are complete.
+- **FakeListLLM default removal**: Provider defaults cleanup (#323, Jeff).
 
 The following areas are substantially implemented and are no longer blocking:
 
 - Resource abstraction (types, registry, handlers with git-checkout and fs-directory, DAG metadata, CLI)
 - Sandbox strategies (git_worktree, copy-on-write, no-sandbox, factory, merge)
 - Tool system (runner, registry, built-in file/git/changeset tools, schema validation, lifecycle, router)
-- Security hardening (config scanner, sandboxed Jinja2 templates, input validation, credential redaction, secrets masking)
-- Apply pipeline (plan_apply_service with diff review, merge, changeset capture)
-- Decision domain model (12 decision types, context snapshots, influence DAG, frozen model helpers)
+- Security hardening (config scanner, sandboxed Jinja2 templates, input validation, credential redaction, secrets masking, template rendering hardening, exception handling enforcement, async resource cleanup)
+- Apply pipeline (plan_apply_service with validation gating, diff review, merge, changeset capture, DoD enforcement, error recovery patterns)
+- Decision system (12 decision types, context snapshots, influence DAG, frozen model helpers, database persistence)
 - Correction engine (revert/append modes, BFS impact analysis, dry-run reporting)
 - Invariant reconciliation (domain model with scopes, merge with precedence, enforcement service)
 - Subplan orchestration (domain models for config, status, merge strategies, failure handling)
 - Phase reversion (constrained apply -> back to strategize state machine)
 - Checkpointing (CheckpointScope, profile flags, cleanup pruning, integrated throughout)
 - Automation profiles (8 built-in profiles, guard logic, 4-level precedence service, CLI commands)
-- Agent Skills loader (registry, protocol, built-in skills, SKILL.md schema support)
+- Agent Skills discovery (registry, protocol, built-in skills, SKILL.md schema support)
+- Changeset persistence (database tables, repository, diff artifact storage)
+- Concurrency controls (plan/project advisory locks, plan resume with state recovery)
+- Provider controls (cost controls, fallback routing, rate limiting)
 - LSP integration (stub modules with registry, models, tool adapter — runtime deferred to server mode)
 - Server connectivity stubs (ACP HTTP transport stub, ACP facade, local-mode routing)
 
 ### Milestone Roadmap
 
-#### Milestone 1: Minimal Local Source-Code Workflow (Target: Feb 15, ETA: ~Feb 27)
+#### Milestone 1: Minimal Local Source-Code Workflow (Target: Feb 15, ETA: Done)
 
 **Goal**: A minimally usable local-mode flow where a user can register an action from YAML, link a git repository resource to a project, and run a plan end-to-end (`plan use` -> `plan execute` -> `plan diff` -> `plan apply`) with a sandboxed workspace and tool-based change capture.
 
-**Current status**: ~96% of M1 tasks complete (6 open out of ~153 total M1 issues). Resource handler runtime (git-checkout, fs-directory) done. Actor runtime and tool-calling partially operational. Database parallelism fix merged today.
+**Current status**: **~99% complete** (1 open out of ~154 total M1 issues). Apply pipeline merged (#155). E2E M1 suite merged (#156, #402). Provider cost controls merged (#324).
 
 **Key remaining work**:
 
-- Sandbox merge/apply pipeline with conflict resolution and validation gating hooks (#155, Luis — active branch)
-- E2E source-code plan lifecycle smoke test suite (#156, Brent)
-- M1 success criteria verification (#402, Brent)
-- FakeListLLM default removal and provider cost controls (#323, #324 — reassigned to Jeff)
+- FakeListLLM default removal (#323, Jeff)
 
 #### Milestone 2: Actor Compiler + Full LLM Integration (Target: Feb 22, ETA: ~Mar 1)
 
 **Goal**: Actor YAML files compile into live LangGraph graphs. Custom actors (strategy, execution, estimation) are fully operational. The tool router normalizes calls across providers, the validation runner enforces resource-attached validations, and the MCP adapter connects to external tool servers.
 
-**Current status**: ~63% of M2 tasks complete (17 closed out of ~48 total M2 issues). Actor compiler (#158) merged today. MCP adapter (#159) in review (PR #382). Agent skills loader (#160) in review (PR #416). Actor YAML hierarchy (#157) active.
+**Current status**: **~95% complete** (5 open out of ~97 total M2 issues). Actor compiler (#158), actor YAML hierarchy (#157), changeset persistence (#163), concurrency locks (#327), plan resume (#328), validation pipeline (#175, #176), M2 E2E suite (#169, #403), CI concurrency (#451) all merged. MCP adapter and agent skills loader remain as critical-path items.
 
 **Key remaining work**:
 
-- MCP adapter runtime (#159, Aditya — PR #382 in review)
-- Agent skills loader (#160, Aditya — PR #416 in review)
-- Actor YAML hierarchy extensions (#157, Aditya — active branch)
-- Changeset persistence and diff review (#163, #303 — Jeff after reassignment, #303 deferred to M3)
-- Validation runner with required/informational mode enforcement (#175, #176 — Jeff after reassignment)
-- M2 E2E actor integration suite (#403, Brent)
+- MCP adapter runtime (#159, Aditya)
+- Agent skills loader (#160, Aditya)
+- Skill persistence — persist flattened tool sets (#166, Jeff)
+- Skill CLI — tools and refresh commands (#167, Aditya)
+- MCP refresh hooks (#168, Aditya)
 
-#### Milestone 3: Decisions + Validations + Invariants (Target: Feb 26, ETA: ~Mar 5)
+#### Milestone 3: Decisions + Validations + Invariants (Target: Feb 26, ETA: ~Mar 2)
 
 **Goal**: Decisions are recorded during Strategize and Execute phases and persisted to the database. Users can view the decision tree (`plan tree`), inspect individual decisions (`plan explain`), manage invariants (`invariant add/list/remove`), and correct decisions (`plan correct --mode revert|append`) with selective subtree recomputation.
 
-**Current status**: ~33% of M3 tasks complete (5 closed out of ~15 total M3 issues). Decision domain model and context snapshots merged. Decision persistence (#171) in review (PR #131, Hamza).
+**Current status**: **~75% complete** (3 open out of ~12 total M3 issues). Decision persistence merged (#171). Validation pipeline and gating merged (#175, #176). DoD gating merged (#178). Diff review artifacts merged (#303). M3 E2E suites merged (#179, #180, #404).
 
 **Key remaining work**:
 
-- Decision persistence service wiring (#171, Hamza — PR #131 in review)
 - Decision recording during strategize/execute phases (#172, Hamza)
+- Invariant enforcement and reconciliation (#173, Jeff)
 - Plan tree and explain CLI commands (#174, Hamza)
-- Diff review artifacts (#303, Luis — active branch, deferred from M2 to M3)
-- Invariant enforcement and reconciliation (#319-#322, Hamza after reassignment)
-- Validation runner and gating (#175, #176, Jeff after reassignment)
 
-#### Milestone 4: Corrections + Subplans + Checkpoints (Target: Mar 2, ETA: ~Mar 8)
+#### Milestone 4: Corrections + Subplans + Checkpoints (Target: Mar 2, ETA: ~Mar 6)
 
 **Goal**: Plans can spawn child plans (subplans) during execution. Subplans execute in parallel with configurable concurrency limits. Results are merged back using three-way merge strategies. Correction engine supports revert and append modes. Checkpointing enables rollback to previous plan states.
 
-**Current status**: ~59% of M4 tasks complete (13 closed out of ~22 total M4 issues). Phase reversion landed. Subplan domain model done.
+**Current status**: **~86% complete** (3 open out of ~22 total M4 issues). DoD gating (#178), error recovery (#186), security hardening (#319, #320, #321), M4 E2E suites (#405, #187) all merged and closed.
 
 **Key remaining work**:
 
-- Correction engine — revert mode (#178, Jeff after reassignment)
-- Correction engine — append mode (#186, Jeff after reassignment)
-- Subplan orchestrator and scheduler (#201, Jeff after reassignment)
-- Parallel execution with max_parallel and fail-fast (#259, Jeff after reassignment)
-- Three-way merge of subplan changesets (#312, Jeff after reassignment)
-- Checkpoint persistence and rollback (#313, Jeff after reassignment)
-- CLI extension tests (PR #420, Brent — in review)
-- M4 E2E correction + subplan suite (#405, Brent)
+- Subplan orchestrator and scheduler (#183, Jeff)
+- Correction service runtime wiring (#184, Luis)
+- Invariant template hardening (#322, Hamza)
 
-#### Milestone 5: ACMS v1 + Context Scaling (Target: Mar 6, ETA: ~Mar 11)
+#### Milestone 5: ACMS v1 + Context Scaling (Target: Mar 6, ETA: ~Mar 10)
 
 **Goal**: The Advanced Context Management System v1 is operational. Projects with 10,000+ files can be indexed and queried. The context assembly pipeline produces scoped, budget-constrained context views for actors. Hot/warm/cold storage tiers manage context lifecycle.
 
-**Current status**: ~21% of M5 tasks complete (4 closed out of ~19 total M5 issues). UKO/RDF groundwork started (Hamza). Config service foundations exist.
+**Current status**: ~7% of M5 tasks complete (14 open out of ~15 total M5 issues). UKO/RDF groundwork started (Hamza). Config service foundations exist.
 
 **Key remaining work**:
 
-- UKO ontology with RDF-based resource representation (#338, #342, #343, Jeff after reassignment)
-- CRP protocol for actor context requests (#199, Jeff after reassignment)
-- Context strategies — keyword, semantic, graph, temporal (#208, #209, Jeff after reassignment)
+- UKO ontology with RDF-based resource representation (#338, #342, #343, Jeff)
+- CRP protocol for actor context requests (#199, Jeff)
+- Context strategies — keyword, semantic, graph, temporal (#208, #209, Jeff)
 - Ten-component context assembly pipeline (Hamza)
 - Hot/warm/cold tiered storage with skeleton compression (Hamza)
 - M5 E2E ACMS integration suite (#406, Brent)
 
-#### Milestone 6: Autonomy Hardening + Server Stubs (Target: Mar 10, ETA: ~Mar 13)
+#### Milestone 6: Autonomy Hardening + Server Stubs (Target: Mar 10, ETA: ~Mar 12)
 
 **Goal**: The system can autonomously execute a large-scale task (e.g., porting a substantial codebase) using hierarchical plan decomposition with 4+ levels of subplans, decision correction with selective subtree recomputation, parallel execution scaling to 10+ concurrent subplans, and validation-gated apply. Client-side server stubs are in place for future connectivity.
 
-**Current status**: ~26% of M6 tasks complete (5 closed out of ~19 total M6 issues). Foundation work from M3-M5 will accelerate M6.
+**Current status**: ~7% of M6 tasks complete (14 open out of ~15 total M6 issues). Foundation work from M3-M5 will accelerate M6.
 
 **Key remaining work**:
 
-- Hierarchical decomposition — recursive subplan planning (#335, #336, Jeff after reassignment)
-- Autonomous estimation — cost/risk/time (#337, Jeff after reassignment)
-- Performance tuning for large-project context assembly (#340, Jeff after reassignment)
-- Server connectivity client stubs (#323, #324, Jeff after reassignment)
+- Hierarchical decomposition — recursive subplan planning (#335, #336, Jeff)
+- Autonomous estimation — cost/risk/time (#337, Jeff)
+- Performance tuning for large-project context assembly (#340, Jeff)
+- Server connectivity client stubs (Jeff)
 - End-to-end porting run with validation gates (all)
 - M6 E2E autonomy integration suite (#407, Brent)
 
@@ -2140,9 +2141,9 @@ Final refinements to achieve the complete specification:
 
 ### Schedule Risk Summary
 
-The project is currently running approximately **7-9 days behind** the original aggressive schedule. The primary causes are: (1) the scope of v2-to-v3 architectural transition was larger than anticipated, requiring significant model realignment and test rewriting; (2) CI/CD pipeline stabilization consumed more effort than planned due to container resource constraints and cross-environment compatibility issues; and (3) the execute/apply pipeline — the critical vertical slice — was deferred in favor of breadth-first foundation work.
+The project is currently running approximately **5-7 days behind** the original aggressive schedule, narrowed from 7-9 days at Day 16. The primary causes of the original delay were: (1) the scope of v2-to-v3 architectural transition was larger than anticipated; (2) CI/CD pipeline stabilization consumed more effort than planned; and (3) the execute/apply pipeline was deferred in favor of breadth-first foundation work.
 
-The gap is expected to **narrow** in later milestones because: (a) foundational infrastructure (persistence, CLI patterns, quality gates, domain models) is now solid and reusable; (b) aggressive workload redistribution on Day 16 moved 25 issues from slower developers (Luis, Hamza) to Jeff (highest velocity), reducing bottleneck risk; and (c) each subsequent milestone builds on proven patterns rather than establishing new ones. Actor compiler merging today unblocks actor-dependent work across M2-M4.
+The gap has been **actively narrowing** since Day 16: (a) the workload redistribution moved 25 issues to Jeff and paid off with a 65-issue closure burst over Days 16-18; (b) the validation pipeline, apply pipeline, security hardening, and decision persistence all landed in a concentrated 2-day window; (c) M1 is essentially complete and M2/M3 are close behind with only 6 and 3 issues remaining respectively; (d) Brent's QA batch merges (M1-M4 E2E suites) eliminated the testing backlog through M4.
 
 !!! note "Server Mode"
 
@@ -2979,3 +2980,94 @@ Note: Each cell is **completed/total story points** for that milestone/owner per
 | M6 | 13/92 | - | 0/39 | - | 24/30 | - | - | 37/161 |
 | M6+ | 0/64 | - | 0/24 | 0/29 | 0/5 | - | 8/8 | 8/130 |
 | **Total** | **403/681** | **177/222** | **66/163** | **29/145** | **125/176** | **0/0** | **108/108** | **908/1495** |
+
+### 2026-02-26 (Day 18)
+
+**Summary**
+
+Status: Behind ~5-7d  
+Deadline risk: HIGH (narrowing)  
+Overall completion: 74.5% (~168/225 milestone issues closed, excluding Epics/Legendaries)
+
+**Notes**
+
+- **65 issues closed in 3 days** (Days 16-18) — the largest productivity burst of the project
+- M1 essentially complete: only #323 (FakeListLLM defaults) remains open
+- Apply pipeline + validation gating + diff review + DoD gating landed in a single batch (Luis PR #417)
+- Decision persistence merged (Hamza PR #131)
+- Actor YAML hierarchy merged (Aditya PR #432)
+- Security hardening batch: template rendering (#319), exception handling (#320), async cleanup (#321) — all merged
+- Brent QA batches merged: M1-M4 E2E suites (#156, #169, #179, #180, #326, #402-#405, #187)
+- Concurrency controls: locks (#327), plan resume (#328), provider cost controls (#324) merged
+- Changeset persistence (#163) and CI concurrency groups (#452) merged
+- Rui returns from leave tomorrow (Feb 27)
+- M3 target date is today; 3 issues remain (#172, #173, #174)
+
+
+#### Milestone forecast 
+
+| Milestone | Target -> ETA | Delta | Risk |
+|-----------|---------------|-------|------|
+| M1 (v3.0.0) | (2026-02-15) -> ETA Done | +11d actual | LOW — 1 issue remaining (#323) |
+| M2 (v3.1.0) | (2026-02-22) -> ETA 2026-03-01 | +7d | MED-HIGH — 5 open issues; MCP adapter + agent skills loader are blockers |
+| M3 (v3.2.0) | (2026-02-26) -> ETA 2026-03-02 | +4d | MEDIUM — 3 open issues; decision recording + invariant enforcement + plan explain |
+| M4 (v3.3.0) | (2026-03-02) -> ETA 2026-03-06 | +4d | MEDIUM — 3 open issues; subplan orchestrator + correction wiring |
+| M5 (v3.4.0) | (2026-03-06) -> ETA 2026-03-10 | +4d | MEDIUM — 14 open issues; ACMS pipeline + context scaling |
+| M6 (v3.5.0) | (2026-03-10) -> ETA 2026-03-12 | +2d | MEDIUM — 14 open issues; autonomy hardening + server stubs |
+
+#### Track forecast 
+
+| Track | Status | ETA | Risk | Blocking |
+|-------|--------|-----|------|----------|
+| Track A (Plan lifecycle) | **essentially complete** | Done (config service deferred to M5) | LOW | none |
+| Track B (Resources + sandbox) | **complete** | Done | LOW | none |
+| Track C (Actors/tools/skills/MCP) | behind ~4-5d | ETA 2026-03-01 | MED-HIGH | MCP adapter (#159) + agent skills loader (#160) + skill persistence (#166) + skill CLI (#167, #168) |
+| Track D (Decisions/validations) | behind ~4d | ETA 2026-03-02 | MEDIUM | decision recording (#172) + invariant enforcement (#173) + plan explain (#174) |
+| Track E (Corrections/subplans) | behind ~4d | ETA 2026-03-06 | MEDIUM | subplan orchestrator (#183) + correction wiring (#184) |
+| Track F (ACMS/context) | not started | ETA 2026-03-10 | MEDIUM | dependent on M3 completion |
+| Track Q (Quality automation) | **complete** | Done | LOW | maintenance only |
+| Track T (Testing) | strong progress | ETA ongoing | LOW-MED | M1-M4 E2E suites done; M5/M6 remain; Rui returns Feb 27 |
+
+#### Developer forecast 
+
+| Name | Days Ahead/Behind | Availability | Risk | Focus |
+|------|-------------------|--------------|------|-------|
+| Jeff | +2d ahead (exceptional throughput; absorbing 25+ reassigned issues successfully) | available | MEDIUM | Invariant enforcement (#173), skill persistence (#166), FakeListLLM fix (#323), M5/M6 backlog |
+| Luis | -3d behind (8/19 closed = 42%; but validation/apply batch was substantial contribution) | available | MEDIUM | Correction wiring (#184), remaining M5/M6 items |
+| Hamza | on track (decision persistence landed; security batch landed) | available | MEDIUM | Decision recording (#172), plan explain CLI (#174), M5 ACMS items |
+| Aditya | -1d behind (MCP adapter and agent skills loader are M2 critical path) | available | MED-HIGH | MCP adapter (#159), agent skills loader (#160), skill CLI (#167, #168) |
+| Brent | +1d ahead (M1-M4 E2E suites all merged in 2 QA batches) | available | LOW | M5 E2E suite (#406), M6 E2E suite (#407), coverage maintenance |
+| Rui | N/A (returns Feb 27) | returning tomorrow | MEDIUM | Behave/Robot test suites after return |
+| Mike/Brian | N/A | standby | LOW | ASV runner fix merged (#388) |
+
+#### Task inventory 
+
+Milestone per developer as of the current date.
+Note: Each cell is **completed/total issues** for that milestone/owner per Forgejo tracker. Epics and Legendaries excluded.
+
+| Milestone | Jeff | Aditya | Luis | Hamza | Brent | Rui | Unassigned | Total |
+|-----------|------|--------|------|-------|-------|-----|------------|-------|
+| M1 | 26/27 | 12/12 | 2/2 | 4/4 | 10/10 | 0/0 | 97/97 | 151/152 |
+| M2 | 22/22 | 8/12 | 4/4 | 1/1 | 5/5 | 0/0 | 52/52 | 92/96 |
+| M3 | 5/6 | 0/0 | 2/2 | 1/3 | 3/3 | 0/0 | 0/0 | 11/14 |
+| M4 | 7/8 | 0/0 | 3/4 | 1/2 | 7/7 | 0/0 | 1/1 | 19/22 |
+| M5 | 1/8 | 0/1 | 0/1 | 0/5 | 1/3 | 0/0 | 2/2 | 4/20 |
+| M6 | 2/10 | 0/0 | 0/4 | 0/0 | 3/5 | 0/0 | 0/0 | 5/19 |
+| M6+ | 0/8 | 0/0 | 0/3 | 0/4 | 0/1 | 0/0 | 1/1 | 1/17 |
+| **Total** | **63/89** | **20/25** | **11/20** | **7/19** | **29/33** | **0/0** | **153/153** | **283/339** |
+
+#### Story point allocation 
+
+Story points per developer per milestone.
+Note: Each cell is **completed/total story points** for that milestone/owner per Forgejo tracker. Issues without a `Points/X` label or milestone are excluded. Epics and Legendaries (organizational items) are excluded.
+
+| Milestone | Jeff | Aditya | Luis | Hamza | Brent | Rui | Unassigned | Total |
+|-----------|------|--------|------|-------|-------|-----|------------|-------|
+| M1 | 180/188 | 128/128 | 16/16 | 19/19 | 71/71 | - | 95/95 | 509/517 |
+| M2 | 147/176 | 49/86 | 29/29 | 5/5 | 21/21 | - | - | 251/317 |
+| M3 | 45/58 | - | 21/21 | 13/23 | 11/11 | - | - | 90/113 |
+| M4 | 42/47 | - | 29/34 | 8/32 | 24/24 | - | 1/1 | 104/138 |
+| M5 | 5/61 | 0/8 | - | 0/37 | 8/14 | - | 4/4 | 17/124 |
+| M6 | 13/92 | - | 0/39 | - | 24/30 | - | - | 37/161 |
+| M6+ | 0/64 | - | 0/24 | 0/29 | 0/5 | - | 8/8 | 8/130 |
+| **Total** | **432/686** | **177/222** | **95/163** | **45/145** | **159/176** | **0/0** | **108/108** | **1016/1500** |