diff --git a/.github/copilot-instructions.md b/.github/copilot-instructions.md
new file mode 100644
index 000000000000..f3d8958f6628
--- /dev/null
+++ b/.github/copilot-instructions.md
@@ -0,0 +1,5 @@
+# GitHub Copilot instructions for PyFluent
+
+Read @AGENTS.md first and follow it as the canonical setup, repo map, commands, and operating rules. `src/ansys/fluent/core` is the runtime package; generated modules are generation-managed unless the task requires otherwise.
+
+Load only relevant sections: @devel/agents/architecture.md (source owners), @devel/agents/testing.md (tests/requirements), @devel/agents/workflows.md (commands/maintenance). Keep this file a short pointer; do not duplicate those maps or rules.
diff --git a/AGENTS.md b/AGENTS.md
new file mode 100644
index 000000000000..0eaa5bece806
--- /dev/null
+++ b/AGENTS.md
@@ -0,0 +1,66 @@
+# AGENTS.md
+
+PyFluent is the Python interface for Ansys Fluent.
+
+## Setup
+
+```bash
+python -m venv .venv
+# Windows
+.venv\Scripts\activate
+# Linux/macOS
+source .venv/bin/activate
+pip install -e ".[tests]"
+# optional extras
+pip install -e ".[reader,search,ui,ui-jupyter]"
+```
+
+## Primary objectives
+
+1. Good answers with minimal token consumption.
+2. Accurate answers using the correct repo map and architecture.
+
+## Commands
+
+```bash
+python -m pytest tests
+pre-commit run --all-files
+```
+
+## Repo map
+
+- `src/ansys/fluent/core` — runtime package
+- `src/ansys/fluent/core/execution` — launchers, base-session context, file sessions, containers, schedulers
+- `src/ansys/fluent/core/connectivity` — Fluent connections and file/data transfer
+- `src/ansys/fluent/core/meshing` — meshing sessions and pre-set meshing workflows
+- `src/ansys/fluent/core/solver` — solver sessions and solver APIs, including settings objects
+- `src/ansys/fluent/core/fields` / `services` — field APIs and backend transports
+- `src/ansys/fluent/core/diagnostics` — search, logging, journaling, exceptions
+- `src/ansys/fluent/core/generated` — generated API code
+- `codegen` — generation workflow
+- `tests` — behavior and regression tests
+- `doc` / `examples` / `devel` — docs, examples, engineering notes
+
+## Agent rules
+
+1. Start with one targeted search and the smallest relevant reads.
+2. Prefer existing tests and project conventions over guessing.
+3. Diagnose the root cause and fix it in the owning abstraction; avoid symptom patches or workarounds. Keep changes narrow and unrelated behavior untouched.
+4. Do not hand-edit generated API files unless required.
+5. Validate with the smallest relevant test target.
+6. Prefer cheap validation first: import checks, syntax checks, and nearby non-Fluent tests.
+7. Avoid large integration or active-Fluent-session tests unless the change truly requires them.
+8. If imports, package boundaries, or public APIs are affected, check import paths before escalating.
+9. If the subsystem, feature mapping, or test target is unclear, ask the user instead of guessing.
+10. Use current implementation paths, not legacy import aliases, to locate owners; check public compatibility aliases in `src/ansys/fluent/core/__init__.py` when imports change.
+11. Follow standard software design principles: separation of concerns, cohesive responsibilities, explicit contracts, and minimal coupling. Prefer existing patterns, simple solutions, and justified abstractions over duplication or speculative complexity.
+12. Preserve user changes and existing instruction intent; confirm before removing instructions. When new concepts, anomalies, or stale guidance suggest an improvement, explain the evidence and proposed update, ask the user, and modify `AGENTS.md` or the relevant guide only after approval. Keep updates concise and verified; remove duplication, not unique facts, constraints, or safety rules.
+
+## Deeper guidance
+
+- `devel/agents/architecture.md` — source owners, feature entry points, runtime relationships
+- `devel/agents/testing.md` — test map, cheap targets, dependencies and skip rules
+- `devel/agents/workflows.md` — contribution, generation, CI/docs commands and maintenance
+- `.github/copilot-instructions.md` — short repo-specific pointer
+
+Load only the relevant section when routing or validation is unclear, or the task touches lifecycle, meshing/solver APIs, generation, packaging, or CI. Do not preload all guides or duplicate their maps here.
diff --git a/devel/agents/architecture.md b/devel/agents/architecture.md
new file mode 100644
index 000000000000..59033a662b79
--- /dev/null
+++ b/devel/agents/architecture.md
@@ -0,0 +1,137 @@
+# Architecture map
+
+Follow `AGENTS.md`. Source paths below are relative to `src/ansys/fluent/core`; test targets and execution requirements live in `devel/agents/testing.md`.
+
+## Project orchestration
+
+Read the diagrams for relationships, then use the owner table for exact edit locations. Arrows show conceptual control/data flow, not class inheritance or every import; dashed arrows show supporting inputs. Fluent itself is an external process, not Python code in this repository.
+
+### Runtime paths
+
+```mermaid
+flowchart TB
+	user["User scripts / examples / integrations"] --> public["Public exports: ansys.fluent.core"]
+	public --> launch["execution/launcher: launch or connect"]
+	platforms["Standalone / containers / PIM / Slurm"] --> launch
+	resources["execution/docker + scheduler: containers / machine allocation"] -.-> launch
+	launch --> connection["connectivity: FluentConnection / channel / cleanup"]
+	launch --> sessions["execution/session: BaseSession / active context"]
+	connection --> factories["Version-aware gRPC and high-level service factories"]
+	factories --> sessions
+	sessions --> meshing["meshing/session: Meshing / PureMeshing"]
+	sessions --> solver["solver/session: Solver and variants"]
+	meshing --> workflows["Pre-set meshing workflows + shared workflow wrappers"]
+	meshing --> model["Datamodel / TUI wrappers + datamodel cache"]
+	workflows --> model
+	solver --> settings["solver/flobject: settings objects / built-ins"]
+	solver --> model
+	sessions --> fields["fields: field data / solver reductions and solution variables"]
+	sessions --> streaming["Events / monitors / transcript / data streaming"]
+	settings --> services["services: abstractions and high-level wrappers"]
+	model --> services
+	fields --> services
+	streaming --> services
+	services --> grpc["_grpc_services: backend implementations / protocol versions"]
+	proto["External ansys-api-fluent: generated protobuf / gRPC schemas"] -.-> grpc
+	grpc --> channel["gRPC channel managed by FluentConnection"]
+	channel --> fluent["External Fluent server"]
+	user --> http["REST client + independent HttpSolver"]
+	http --> restsettings["services/rest_settings + shared flobject settings runtime"]
+	restsettings --> transport["rest: client / HTTP transport"]
+	transport --> web["External Fluent web server: 27.1+"]
+	user --> offline["FileSession + file_reader: case/data-backed APIs"]
+	files["Case / data assets"] --> offline
+	transfer["connectivity: file/data transfer strategies"] -.-> sessions
+	transfer -.-> files
+```
+
+- `solver.settings`, generated `object_model` and `session.tui` are distinct service-backed APIs with their own wrappers and schemas.
+- `execution.session` wires these services into the session API to expose Fluent capabilities.
+- The `fields` namespace provides field-data, reduction and solution-variable APIs.
+- The `expressions` namespace constructs and evaluates expressions.
+- `_variable_strategies` handles descriptor naming for expressions, fields and solution variables.
+- `rpvars` provides RP-variable helpers.
+- `services.scheme_interpreter` provides the Scheme interface.
+- `local_parametric_study` orchestrates parametric studies across sessions.
+- `system_coupling` integrates solver operations with external coupled workflows.
+- `ui` integrates session web and Jupyter presentation.
+- `module_config` provides runtime configuration and environment defaults.
+- `diagnostics` provides logging, journaling and exception support.
+- `utils` provides utilities used across runtime layers.
+- `_types` defines shared types used across runtime layers.
+- `examples` provides example assets and download helpers.
+- `diagnostics.search` reads the generated API index; it is not a transport.
+
+### Generation and repository tooling
+
+Paths in this diagram are repository-relative. Detailed commands and prerequisites remain in `devel/agents/workflows.md` rather than being duplicated here.
+
+```mermaid
+flowchart LR
+	driver["codegen/allapigen.py: live generation driver"] --> live["Meshing + solver-icing sessions: static info / workflow tasks"]
+	live --> generators["src/ansys/fluent/core/codegen: schema generators"]
+	generators --> output["generated: versioned settings / datamodel / TUI / task stubs + shared built-ins / API index"]
+	output -.-> runtime["Runtime wrappers / built-in settings / search"]
+	output -.-> docs["doc: API RST generators + Sphinx sources / gallery"]
+	examples["examples: end-to-end user workflows"] --> runtime
+	examples --> docs
+	tests["tests: unit / live Fluent / version-mode / external integration"] --> runtime
+	tests --> generators
+	packaging["pyproject.toml + requirements: dependencies / extras / build and tool config"] -.-> runtime
+	packaging -.-> tests
+	ci[".github/workflows + .ci + Makefile: build / generate / test / docs / release lanes"] --> driver
+	ci --> tests
+	ci --> docs
+	ci --> packaging
+	guidance["AGENTS.md + devel/agents: routing / validation / operating rules"] -.-> work["Agent / contributor work"]
+	notes["devel: engineering notes / investigations"] -.-> work
+	work --> runtime
+	work --> generators
+	work --> tests
+```
+
+## Source owners and entry points
+
+| Feature / public surface | Source owner |
+| --- | --- |
+| Public exports, legacy aliases; configuration/environment defaults | `__init__.py`; `module_config.py` |
+| `launch_fluent()`, `connect_to_fluent()`; standalone/container/PIM/Slurm | `execution/launcher/launcher.py`; `execution/launcher/standalone_launcher.py`, `execution/launcher/container_launcher.py`, `execution/launcher/pim_launcher.py`, `execution/launcher/slurm_launcher.py` |
+| Base session, `using()`, file-backed `FileSession` | `execution/session/session.py`; `execution/session/file.py` |
+| gRPC connection lifetime, health checks, cleanup; file/data transfer | `connectivity/fluent_connection.py`; `connectivity/file_transfer_service.py`, `connectivity/data_transfer.py` |
+| `Meshing`, `PureMeshing`; pre-set workflows and solver transitions | `meshing/session`; `meshing/meshing_workflow.py`, `meshing/meshing_workflow_old.py`; shared `workflow.py`, `workflow_old.py` |
+| `Solver`, `SolverAero`, `SolverIcing`, `SolverLite`, `PrePost`; solve-mode APIs | `solver/session`; settings runtime in `solver/flobject.py` |
+| `solver.settings`; built-in settings exports from `ansys.fluent.core.solver` | `solver/flobject.py`, `solver/settings_builtin_bases.py`, `solver/settings_builtin_data.py`; `services/settings.py` |
+| Meshing datamodel roots; `session.tui` | `services/object_model.py`, `_data_model_cache.py`; `services/text_interface.py` |
+| `session.fields.field_data`, `.new_batch()`; solver fields: `.reduction`, `.solution_variable_data`, `.solution_variable_info` | `fields/field_data/live_field_data.py`; `fields/reduction/reduction.py`; `fields/solution_variables/solution_variables.py` |
+| Service abstractions/wrappers; events, monitors, transcripts and datamodel/field streaming | `services`; `services/streaming_services`; gRPC implementations in `_grpc_services` |
+| Case/data readers and file-backed fields | `file_reader`; `execution/session/file.py` |
+| `search(...)`, logging, journaling, exceptions | `diagnostics`; search implementation `diagnostics/search.py`, API index generator `codegen/api_tree.py` |
+| Parametric studies; coupled simulation | `local_parametric_study.py`; `system_coupling.py` |
+| Batch service calls; separately, queued/remote execution | `services/batch_ops.py`; `execution/scheduler`, `execution/docker`, launchers above |
+| Expression construction/evaluation, RP variables, Scheme, file S-expressions | `expressions`; `rpvars.py`; `services/scheme_interpreter.py`; `file_reader/lispy.py` |
+| `rest.connect_to_webserver()`, `FluentRestClient`, `HttpSolver` | `rest/client.py`, `rest/transport.py`, `services/rest_settings.py`, `solver/session/http_solver.py` |
+| Web/Jupyter UI; utilities/setup; example downloads/assets | `ui`; `utils`; `examples` |
+| Generated settings/datamodel/TUI/built-ins/search index; generation implementation | `generated`; `codegen/settingsgen.py`, `codegen/datamodelgen.py`, `codegen/tuigen.py`, `codegen/builtin_settingsgen.py`, `codegen/api_tree.py` (repository-root `codegen/allapigen.py` is the live-Fluent driver) |
+| Shared types/launcher arguments; descriptor naming for expressions/fields/solution variables | `_types.py`; `_variable_strategies` |
+| Legacy standalone datamodel-server helper, not a normal session entry point | `_stand_alone_datamodel_client/_datamodel_client.py`; verify its old imports/dependencies before use |
+
+### Runtime contracts
+
+- `FluentMode` in `execution/launcher/launch_options.py` accepts `meshing`, `pure_meshing`, `solver` (default), `solver_icing`, `solver_aero`, `pre_post`. The first five route to their corresponding sessions; `pre_post` currently maps to `Solver`. Do not infer launch support from the existence of a session class.
+- `BaseSession` and `using()` live in execution; mode-specific sessions live in meshing/solver. `FileSession` is file-backed, not a live solver subclass; `HttpSolver` is independent of `BaseSession` and gRPC infrastructure.
+- Solver settings objects use `solver/flobject.py` over a settings service. `get_root()` builds classes from static info when `config.use_runtime_python_classes` is enabled or the generated settings file is missing; otherwise it loads version-specific generated classes.
+- Generated output uses version directories under `generated`, plus shared `generated/solver` and `generated/api_tree` data. Versions present in a local checkout depend on generation/install state; do not assume every supported version is generated locally.
+- Legacy names registered in `__init__.py` are compatibility aliases, not current source locations. Start from the implementation path and check exports separately.
+
+## Architectural rules
+
+- The runtime package and session layer are the primary user-facing entry points.
+- Generated code is authoritative for many schema-driven APIs; do not hand-edit it lightly.
+- Field-data, reduction, settings, and datamodel access are higher-value feature areas than TUI-only command wrappers.
+- For any version-specific path, prefer the generated or compatibility-aware implementation and confirm with the closest tests.
+- If a feature is unclear, ask the user before assuming the route or test target.
+- Fix runtime navigation/behavior in its owning wrapper or service; change generation logic for schema-output defects rather than hand-editing generated output.
+
+Session, workflow, service and schema behavior varies by Fluent version, mode, packaging and environment. Check the closest compatibility-aware implementation and test rather than assuming one path.
+
+File-backed APIs need no live server themselves; reader tests may need assets/downloads or Fluent. Search consumes generated API-index data and semantic search needs NLTK data. `HttpSolver` targets Fluent 27.1+; REST settings can use runtime classes without generated files. UI needs `ui` / `ui-jupyter` extras; Python-console tests are not UI-rendering coverage.
diff --git a/devel/agents/testing.md b/devel/agents/testing.md
new file mode 100644
index 000000000000..4e13e92ad588
--- /dev/null
+++ b/devel/agents/testing.md
@@ -0,0 +1,58 @@
+# Testing strategy
+
+Follow `AGENTS.md`: validate imports/syntax, then the nearest behavior-scoped node/file, then subsystem/integration tests only as needed. Prefer real behavior over mock-only assertions; validate downstream contracts for generated/schema changes and add a focused public-API regression test. Never replace relevant validation with unrelated cheap tests.
+
+## Commands
+
+`python -m pytest tests` is the broad default, not the starting point for a narrow fix:
+
+```bash
+python -m pytest tests/test_rest.py
+python -m pytest tests/test_search.py::test_match_whole_word
+python -m pytest tests/test_settings_api.py -k relevant_behavior --fluent-version=26.1
+python -m pytest tests/test_solvermode/test_models.py --solvermode --fluent-version=26.1
+```
+
+Use the actual target Fluent version. `python -m pytest --collect-only <target>` checks imports/node IDs without ordinary fixture execution; it does not prove tests will run rather than skip.
+
+## Requirements and skips
+
+The source of truth is `pyproject.toml` (`tool.pytest.ini_options`) and `tests/conftest.py`:
+
+- Discovery starts in `tests`; generated-on-demand `tests/fluent` and `tests/journals` are ignored. `--write-fluent-journals` regenerates output and clears the collected run; not routine validation.
+- `nightly` tests skip without `--nightly`. This flag enables nightly tests in addition to ordinary tests.
+- `tests/test_solvermode` runs only with `--solvermode`; that flag also skips tests outside that directory. `tests/test_meshingmode` has no corresponding mode flag.
+- `fluent_version(spec)` checks `--fluent-version` if supplied. `latest` means the current release in ordinary runs and release-or-newer in nightly/solvermode runs; `dev` means the current development version. Omitting the version option does not automatically filter unsupported tests.
+- `standalone` means incompatible with container execution, not mock-only or server-free. `settings_only` means settings without mesh/solve work, not no Fluent session.
+- `rest_server` identifies live web-server tests. Read `rest_server_connection` and `solver_session_grpc_rest` fixtures for launch, endpoint, and authentication requirements; the marker alone does not start or skip a server.
+- Inspect fixture dependencies in `tests/conftest.py` before classifying a test as offline. `tests/fluent_fixtures.py` supplies helpers for generated Fluent journals, not the ordinary pytest fixture registry.
+- The autouse fixture changes the working directory to `tmp_path`, sets test configuration, and copies local certificates when present.
+- Install the package and `tests` extra for pytest; readers need `reader` (`h5py`), semantic search needs `search` (`nltk`), and UI needs `ui` / `ui-jupyter`. Case-data downloads, generated APIs, licenses, Docker, and live Fluent are separate requirements; test extras alone do not provide them.
+
+## Feature-to-test map
+
+Repository-relative candidates, not guarantees that whole files are cheap/server-free. Source owners are in `devel/agents/architecture.md`.
+
+- Launch and session lifecycle: `tests/test_launcher.py`, `tests/test_launcher_remote.py`, `tests/test_fluent_session.py`, `tests/test_session.py`, `tests/test_pre_post_session.py` (mixed unit/live coverage).
+- Connections/security/errors: `tests/test_grpc_security_options.py`, `tests/test_error_handling.py`; config: `tests/test_config.py`.
+- Meshing workflows: `tests/test_meshing_workflow.py`, `tests/test_new_meshing_workflow.py`, `tests/test_meshing_utilities.py`, `tests/test_pure_mesh_vs_mesh_workflow.py`, `tests/test_server_meshing_workflow.py` (live/version-sensitive).
+- Meshing launch/end-to-end: `tests/test_meshingmode/test_meshing_launch.py`, `tests/test_cad_to_post_wtm.py`, `tests/test_cad_to_post_ftm.py` (live).
+- Solver and solver-mode behavior: `tests/test_solution_variables.py`, `tests/test_solvermode`, `tests/test_tui_api.py`, `tests/test_public_api.py`; variants: `tests/test_aero_session.py`, `tests/test_icing_session.py`, `tests/test_pre_post_session.py` (live; solvermode requires its flag).
+- Fields and reduction: `tests/test_field_data.py`, `tests/test_reduction.py`, `tests/test_solution_variables.py`, `tests/test_physical_quantities.py` (live paths; select mock checks where present).
+- Settings and datamodel API: `tests/test_settings_api.py`, `tests/test_settings_reader.py`, `tests/test_datamodel_api.py`, `tests/test_datamodel_service.py`, `tests/test_builtin_settings.py`, `tests/test_mapped_api.py`, `tests/test_creatable.py`, `tests/test_preferences.py` (mostly live/model-dependent).
+- Settings runtime/cache/generation: `tests/test_flobject.py`, `tests/test_data_model_cache.py`, `tests/test_codegen.py`, `tests/test_type_stub.py` (mixed mock/live/generated contracts).
+- File and transfer behavior: `tests/test_file_session.py`, `tests/test_file_transfer_service.py`, `tests/test_datareader.py`, `tests/test_casereader.py` (mixed live, file assets/downloads, reader extras).
+- Search and API browsing: `tests/test_search.py` (mocked lookup, generated index, NLTK/downloads, and live-session selections).
+- API and object surfaces: `tests/test_flobject.py`, `tests/test_rp_vars.py`, `tests/test_public_api.py` (mixed; do not assume object-model tests are offline).
+- Batch/remote/scheduler: `tests/test_batch_ops.py`, `tests/test_launcher_remote.py`, `tests/test_slurm_future.py`; machine allocation: `tests/test_scheduler.py` (no Fluent).
+- Events/streaming/monitors: `tests/test_events_manager.py`, `tests/test_streaming_services.py`, `tests/test_solver_monitors.py` (live).
+- Parametric/coupling: `tests/parametric/test_local_parametric_setup.py` (reader/data download), `tests/parametric/test_local_parametric_run.py`, `tests/parametric/test_parametric_workflow.py`, `tests/test_systemcoupling.py`; optiSLang: `tests/integration/test_optislang/test_optislang_integration.py` (external integration).
+- Expressions: `tests/expressions` (parser/builder unit tests and live smoke/integration tests); Scheme/RP/console: `tests/test_scheme_eval.py`, `tests/test_rp_vars.py`, `tests/test_pyconsole.py`; file S-expressions: `tests/test_lispy.py` (no Fluent).
+- REST: `tests/test_rest.py` (mocked client/transport); parity: `tests/test_settings_grpc_rest.py` (live gRPC/web-server fixtures). No dedicated Panel/Jupyter rendering test file is currently mapped.
+- Utilities/compatibility/docs: `tests/test_utils.py`, `tests/test_fluent_version.py`, `tests/test_fluent_version_marker.py`, `tests/test_deprecate.py`, `tests/test_logging.py`, `tests/test_fix_doc.py`, `tests/test_topy.py`, `tests/test_fluent_fixes.py` (check individual fixtures).
+
+### Cheap starting targets
+
+No Fluent server: `tests/test_config.py`, `tests/test_fluent_version.py`, `tests/test_fluent_version_marker.py`, `tests/test_lispy.py`, `tests/test_scheduler.py`, `tests/test_rest.py`. Package/pytest dependencies still apply.
+
+Mixed-file unit nodes: `tests/test_search.py::test_match_whole_word`, `tests/test_data_model_cache.py::test_data_model_cache`. Inspect bodies/transitive fixtures before expanding.
diff --git a/devel/agents/workflows.md b/devel/agents/workflows.md
new file mode 100644
index 000000000000..832a3cdea3e1
--- /dev/null
+++ b/devel/agents/workflows.md
@@ -0,0 +1,29 @@
+# Contributor workflow
+
+Follow the root rules in `AGENTS.md`; this guide adds command and maintenance details.
+
+## Default flow
+
+1. Find the owner and nearest test with targeted searches/reads; ask if uncertain.
+2. Diagnose and fix the root cause within that feature, following existing patterns and runtime/generated boundaries.
+3. Validate the behavior cheaply, check changed imports/exports before heavier tests, and summarize changes with evidence. Preserve user changes and instruction intent as required by the root rules.
+
+## Ownership and commands
+
+- Packaging, extras, Python support, pytest options, and tool configuration live in `pyproject.toml`; build dependencies are in `requirements/requirements_build.txt`.
+- API generation logic lives in `src/ansys/fluent/core/codegen`. `codegen/allapigen.py` launches live meshing and solver-icing sessions to gather static info, then generates versioned APIs, workflow task stubs, and search index data into `config.codegen_outdir`.
+- Generated output is not an ordinary runtime edit target. For a schema-output defect, update the responsible generator and validate its output contract with `tests/test_codegen.py` and the nearest downstream API test. Full regeneration needs suitable Fluent/server resources.
+- Setup: editable install from `AGENTS.md`; extras/test requirements from `devel/agents/testing.md`.
+- Style: `pre-commit run --files <changed-files>` for a focused check; `pre-commit run --all-files` remains the canonical full check. Hook configuration is in `.pre-commit-config.yaml`; some hooks explicitly scan whole trees even with a file selection.
+- CI entry points are in `.github/workflows`; helper scripts and the type-check baseline live in `.ci`. `Makefile` defines versioned `unittest-dev-*`, `unittest-all-*` (nightly), and `unittest-solvermode-*` lanes, plus `api-codegen`, `test-import`, and documentation targets.
+- Treat make targets as CI/platform workflows, not portable local defaults: `make install` runs `git clean -fd`, versioned test targets clear example data with `sudo rm`, and API/docs targets assume Bash/Linux tools and may remove generated output. Inspect the target and confirm destructive cleanup before running it.
+- Documentation sources are in `doc/source`; API RST generators are in `doc`, examples in `examples/00-fluent`, and changelog fragments in `doc/changelog.d` (Towncrier configuration in `pyproject.toml`). With doc dependencies installed, `make -C doc html` builds Sphinx; `BUILD_ALL_DOCS` controls API RST regeneration. Full builds can import generated APIs and execute gallery examples, so they are not a cheap default for guidance-only edits.
+
+## Keeping maps accurate
+
+- After user approval under the root maintenance rule, update moved source owners, public/legacy import routing, and test paths together. Verify paths/node IDs, not inferred names.
+- Keep one home per fact: root rules, architecture source owners, testing targets/requirements, workflow commands. Load only relevant sections; keep `.github/copilot-instructions.md` a pointer.
+- Classify cost from bodies/transitive fixtures and actual skip rules, not filenames or markers.
+- For guidance-only changes, check referenced paths (allow explicitly generated-on-demand outputs), command definitions, retained rules, and `git diff --check`; runtime integration tests are unnecessary unless the change also affects executable behavior.
+
+Review the relevant compatibility/schema/build/CI definitions when work touches Fluent versions, generation, packaging/releases, or platform-specific execution; escalate unresolved constraints rather than guessing.
diff --git a/doc/changelog.d/5413.maintenance.md b/doc/changelog.d/5413.maintenance.md
new file mode 100644
index 000000000000..1a449738e45d
--- /dev/null
+++ b/doc/changelog.d/5413.maintenance.md
@@ -0,0 +1 @@
+Include agents md