ecoportal-api-graphql 1.3.13 → 1.3.16

data/.ai-assistance/capabilities/cowork.md

@@ -1,55 +0,0 @@
-# CoWork Native Capabilities
-
-Last verified: 2026-06-04
-Environment: Claude desktop app — CoWork mode
-
-## Can do natively
-
-- Read, write, edit files in connected folders (including all of `.ai-assistance/`)
-- Web search and page fetch
-- Bash sandbox (Ubuntu 22, Python/Node preinstalled, allowlisted network)
-- Create documents: .docx, .pptx, .xlsx, .pdf, .md, .html
-- Render live HTML/React artifacts (persist across sessions)
-- Scheduled tasks (while desktop app is open)
-- Skills: docx, pptx, xlsx, pdf, and all installed plugin skills
-
-## Cannot do
-
-- Write or edit files under `.ai-assistance/` (application-layer guard on that path)
-- Direct git operations (clone, push, pull, branch, commit)
-- Install software on host machine
-- Run commands outside the bash sandbox
-- Access Code sessions or terminal history
-- Use MCP servers defined in `claude_desktop_config.json` — those do not surface as
-  CoWork connectors. CoWork connectors are installed via Settings → Plugins only.
-  Verified 2026-06-04: GitLab and GitKraken written to Desktop config, neither
-  appeared as available connectors in CoWork.
-
-## Connectors available in CoWork
-
-See `connectors.md` for the full registry. Install via Settings → Plugins.
-
-Confirmed available (2026-06-04):
-- Google Workspace (Drive, Gmail, Calendar) — if authenticated
-- Slack — if authenticated
-- Notion — if authenticated
-- HubSpot, Intercom, Asana, Linear, etc. — if plugin installed + authenticated
-
-## When to use the bridge
-
-The bridge is a **connector delegation service** — use it only when you need to call
-a service that CoWork cannot reach. Do not bridge for working tree edits, documentation,
-or anything in the "Can do natively" list above.
-
-Bridge to Code when you need:
-
-| Connector / service | Why Code has it |
-|---------------------|-----------------|
-| Gemini | Gemini MCP configured in Code |
-| GitLab (MRs, pipelines, issues) | GitLab MCP configured in Code |
-| GitKraken (git history, branches) | GitKraken MCP configured in Code |
-| Unrestricted shell (host-level) | Code runs outside the sandbox |
-
-Before writing a bridge task, check `connectors.md` to confirm the connector is
-available and authenticated in Code. If status is `unknown` or `needs-setup`, note
-that in the bridge task so Code can surface the issue.

data/.ai-assistance/code/OVERVIEW.md

@@ -1,155 +0,0 @@
-# Code Spec: ecoportal-api-graphql — Gem Overview
-
-**Scope:** High-level structure, entry points, key patterns, and namespace conventions for the whole gem. Does NOT cover per-area implementation detail — see area-specific docs in `.ai-assistance/code/` for those.
-**Last updated:** 2026-06-02
-**Key files:**
-- `lib/ecoportal/api-graphql.rb` — gem entry point (requires upstream, boots namespace)
-- `lib/ecoportal/api/graphql.rb` — main `Ecoportal::API::GraphQL` class
-- `lib/ecoportal/api/common/graphql/client.rb` — GraphQL HTTP client (wraps graphlient)
-- `ecoportal-api-graphql.gemspec` — gem spec and dependencies
-
----
-
-## Overview
-
-`ecoportal-api-graphql` provides a Ruby client for EcoPortal's GraphQL API. It wraps `graphlient` (which wraps `graphql-client`) and builds a structured layer of models, queries, mutations, fragments, inputs, payloads, and builders on top of it.
-
-The gem is used primarily by `eco-helpers` (downstream). Backwards compatibility with `eco-helpers` must be preserved unless explicitly broken.
-
----
-
-## Dependency Chain
-
-```
-graphql-ruby (rmosolgo)         — schema definition side (not directly used as a dep)
-graphql-client                  — low-level HTTP+parsing
-graphlient                      — friendlier wrapper around graphql-client (~> 0.8)
-ecoportal-api (~> 0.10)         — REST API base (auth, org context)
-ecoportal-api-v2 (~> 3.3)       — REST API v2 (required by entry point)
-        ↓
-ecoportal-api-graphql           ← THIS GEM
-        ↓
-eco-helpers                     — downstream consumer
-```
-
----
-
-## Entry Point
-
-`require 'ecoportal/api-graphql'` loads:
-1. `ecoportal-api-v2` (and transitively `ecoportal-api`)
-2. `ecoportal/api/graphql_version`
-3. `ecoportal/api/common.graphql` — shared common GraphQL utilities
-4. `ecoportal/api/graphql` — the main class + all sub-requires
-
-Load order within `graphql.rb`:
-`helpers` → `concerns` → `logic/base_model` → `error` → `interface` → `base` → `model` → `logic` → `connection` → `payload` → `input` → `fragment` → `query` → `mutation` → `builder`
-
----
-
-## Main Class: `Ecoportal::API::GraphQL`
-
-The public entry point. Instantiated with credentials:
-
-```ruby
-api = Ecoportal::API::GraphQL.new(email: "...", pass: "...", org_id: "...", host: "live.ecoportal.com")
-```
-
-Key public methods:
-- `#currentOrganization` / `#currentOrganizationClass` — returns `Model::Organization` with client set
-- `#contractorEntity` → `Builder::ContractorEntity`
-- `#locationStructure` → `Builder::LocationStructure`
-- `#action` → `Builder::Action`
-- `#createContractorEntity(input:)` — convenience shortcut (delegates to mutation)
-
-Includes `Common::GraphQL::ClassHelpers` for `class_resolver` — allows subclasses to override `client_class`.
-
----
-
-## Namespace Convention
-
-**IMPORTANT:** The module is `Ecoportal::API::GraphQL` (capital Q, capital L) but Ruby casing — NOT `Ecoportal::API::GraphQl`. File paths use lowercase: `graphql/`.
-
-Class nesting mirrors folder structure exactly:
-
-| Folder | Module/Class |
-|--------|-------------|
-| `graphql/model/` | `GraphQL::Model::*` |
-| `graphql/input/` | `GraphQL::Input::*` |
-| `graphql/mutation/` | `GraphQL::Mutation::*` |
-| `graphql/query/` | `GraphQL::Query::*` |
-| `graphql/builder/` | `GraphQL::Builder::*` |
-| `graphql/fragment/` | `GraphQL::Fragment::*` |
-| `graphql/payload/` | `GraphQL::Payload::*` |
-| `graphql/connection/` | `GraphQL::Connection::*` |
-| `graphql/logic/` | `GraphQL::Logic::*` |
-| `graphql/base/` | `GraphQL::Base::*` |
-| `graphql/interface/` | `GraphQL::Interface::*` |
-| `graphql/error/` | `GraphQL::Error::*` |
-
----
-
-## Architectural Layers
-
-```
-Builder           High-level façade — developer-facing API per domain
-    ↓
-Query / Mutation  GraphQL operation definitions
-    ↓
-Logic             Execution layer (base_query, base_model, query, mutation, payload, input, connection)
-    ↓
-Fragment          GraphQL fragment strings
-    ↓
-Model             Response data models (read side)
-Input / Payload   Mutation arguments (write side) and mutation responses
-    ↓
-Base              Shared base classes for models
-Interface         GraphQL interface type definitions
-    ↓
-Common::GraphQL::Client   HTTP layer (graphlient wrapper with auth)
-```
-
----
-
-## Common::GraphQL Layer
-
-Lives in `lib/ecoportal/api/common/graphql/`. Key classes:
-
-- `Client` — extends `Graphlient::Client`. Handles auth (`AuthService`), org_id, timeouts (90s read/write). `no_schema: true` skips schema introspection.
-- `Model` — base model class for response parsing.
-- `ClassHelpers` — `class_resolver` macro for dependency injection of class references.
-- `HashHelpers` — hash utility methods.
-- `QueryIntegration` — mixin for query execution.
-- `Patches` — monkey-patches to upstream libs (review carefully before changing).
-
----
-
-## Domain Objects
-
-The gem covers these EcoPortal GraphQL domains:
-
-| Domain | Builder | Key Model | Key Mutations |
-|--------|---------|-----------|--------------|
-| Location Structure | `Builder::LocationStructure` | `Model::LocationStructure` | draft create/delete/publish, apply_commands |
-| Contractor Entity | `Builder::ContractorEntity` | `Model::ContractorEntity` | create, update, destroy |
-| Action | `Builder::Action` | `Model::Action` | create, update, archive |
-| Organization | — (direct on GraphQL class) | `Model::Organization` | — |
-| Page | — | `Model::Page` | — |
-
----
-
-## Gotchas / Non-Obvious Behaviours
-
-- `no_schema: true` is always passed — schema introspection is disabled. Queries are not validated against a schema at build time.
-- `GraphQL::Fragment` is instantiated at GraphQL init and passed around — it's a container for fragment string definitions.
-- `Builder` classes take a `client` directly, not a `GraphQL` instance.
-- `Logic::BaseModel` inherits from `Common::GraphQL::Model` and adds a class-level `client` accessor. Most model/query classes inherit from this.
-- `class_resolver` (from `ClassHelpers`) allows subclasses to override which concrete class is used — important for testing and extension.
-
----
-
-## TODOs / Open Questions
-
-- [ ] Per-area code spec docs need to be created as analysis progresses.
-- [ ] `Common::GraphQL::Patches` contents not yet reviewed — may contain important behavioural notes.
-- [ ] Spec coverage is minimal — `spec/ecoportal/api_spec.rb` exists but subdirectory coverage is sparse.

data/.ai-assistance/code/data_fields.md

@@ -1,242 +0,0 @@
-# Data Fields — Code Spec
-
-*All 20 DataField types, access patterns, mutation input serialisation, and Collection API.*
-
----
-
-## Overview
-
-Data fields are the form inputs within a page (rich content — not structural metadata).
-Each field is a live instance with a MongoDB ObjectId (`id`) assigned on page creation.
-
-All field types inherit from `Base::Page::DataField < Logic::BaseModel` and live in
-`lib/ecoportal/api/graphql/base/page/data_field/`.
-
----
-
-## All 20 Field Types
-
-| GraphQL `__typename` | Ruby class | Writable | Input key |
-|---|---|---|---|
-| `PlainText` | `DataField::PlainText` | ✓ | `plainText` |
-| `RichText` | `DataField::RichText` | ✓ | `richText` |
-| `Date` | `DataField::DateField` | ✓ | `date` |
-| `Number` | `DataField::Number` | ✓ | `number` |
-| `Gauge` | `DataField::Gauge` | ✓ | `gauge` |
-| `Select` | `DataField::Select` | ✓ | `select` |
-| `Checklist` | `DataField::Checklist` | ✓ | `checklist` |
-| `TagField` | `DataField::TagField` | ✓ | `locationField` |
-| `People` | `DataField::People` | ✓ | `people` |
-| `Geo` | `DataField::Geo` | ✓ | `geo` |
-| `ContractorEntities` | `DataField::ContractorEntities` | ✓ | `contractorEntities` |
-| `CrossReference` | `DataField::CrossReference` | ✓ | `crossReference` |
-| `File` | `DataField::FileField` | ✓* | `file` |
-| `ImageGallery` | `DataField::ImageGallery` | ✓* | `imageGallery` |
-| `Signature` | `DataField::Signature` | — | — |
-| `Mailbox` | `DataField::Mailbox` | — | — |
-| `ActionsList` | `DataField::ActionsList` | — | — |
-| `Law` | `DataField::Law` | — | — |
-| `AiSummary` | `DataField::AiSummary` | — | — |
-| `Table` | `DataField::Table` | — | — |
-
-\* File and ImageGallery require a prior file upload to the REST upload endpoint to obtain
-`fileContainerIds`. See `projects/TODO.md` — "File upload" item.
-
----
-
-## Dispatch
-
-`DataField.from_doc(raw_hash)` reads `__typename` (or `type` if already normalised) and
-returns the right subclass instance. Falls back to base `DataField` for unknown types.
-
-```ruby
-field = Base::Page::DataField.from_doc({ '__typename' => 'PlainText', 'id' => 'f1', 'value' => 'X' })
-# => #<DataField::PlainText ...>
-```
-
----
-
-## Field Access on Page Models
-
-`Interface::BasePage` includes `Concerns::DataFieldAccess`, so all concrete page models have:
-
-```ruby
-page.field_collection      # → DataField::Collection (memoized)
-page.components            # → same (v2 compat alias)
-page.data_fields_updates   # → [DataFieldInput, ...] for dirty fields
-page.data_fields_additions # → [DataFieldInput, ...] for added fields
-page.data_fields_deletions # → [field_id, ...] queued for deletion
-page.data_fields_dirty?    # → Boolean
-```
-
-`Base::Page::Phased::Stage` also exposes `components` (via `sections.components`) for
-stage-level field access.
-
----
-
-## DataField::Collection API
-
-```ruby
-coll = page.components
-
-# Read
-coll.get_by_type(:plain_text)     # → [DataField::PlainText, ...]
-coll.get_by_type('PlainText')     # → same (string also works)
-coll.get_by_name('Summary')       # → DataField::PlainText (case-insensitive)
-
-# Iterate
-coll.each { |f| puts f.label }
-coll.map(&:label)
-
-coll.doc                          # → [raw_hash, ...] (v2 compat)
-coll.empty?                       # → Boolean
-coll.length                       # → Integer
-
-# Mutations — updates (existing fields)
-field = coll.get_by_name('Summary')
-field.value = 'New text'
-coll.dirty_inputs                 # → [{ plainText: { id: 'f1', value: 'New text' } }]
-
-# Mutations — additions (new field instances from buildFromTemplate)
-coll.add(id: 'server_id', label: 'New Field', type: 'PlainText') { |f| f.value = 'X' }
-coll.dirty_additions              # → [{ plainText: { id: 'server_id', value: 'X' } }]
-
-# Mutations — deletions (remove field instances)
-coll.mark_for_deletion('f1')          # by ID string
-coll.mark_for_deletion(field_object)  # by field object (uses field.id)
-coll.dirty_deletions              # → ['f1']
-
-coll.dirty?                       # → true if any of the above have changes
-```
-
----
-
-## Setter API per Type
-
-### PlainText
-```ruby
-field.value = 'text string'
-field.as_input  # → { plainText: { id:, value: } }
-```
-
-### RichText
-```ruby
-field.content = '<p>HTML</p>'
-field.value = '<p>HTML</p>'   # alias
-field.as_input  # → { richText: { id:, content: } }
-```
-
-### Date
-```ruby
-field.value = '2025-06-30T09:00:00+12:00'  # ISO8601
-field.as_input  # → { date: { id:, value: } }
-```
-
-### Number / Gauge
-```ruby
-field.value = 42.5
-field.as_input  # → { number: { id:, value: } }  or  { gauge: { id:, value: } }
-```
-
-### Select
-```ruby
-field.select_option('Active')         # select by name (deselects all others)
-field.select_option('o1')             # select by option id
-field.select_option('opt1', 'opt2')   # multi-select by id
-field.clear_selection
-field.selected_options                # → [{ 'id' =>..., 'name' =>..., 'selected' => true }]
-field.as_input  # → { select: { id:, options: [{ id:, selected: }] } }
-```
-
-### Checklist
-```ruby
-field.check('Step A')                  # check by label
-field.check('i1', checked: false)      # uncheck by id
-field.as_input  # → { checklist: { id:, items: [{ id:, checked: }] } }
-```
-
-### TagField (LocationField)
-```ruby
-field.location_ids = ['loc1', 'loc2']
-field.as_input  # → { locationField: { id:, locationIds: [...] } }
-```
-
-### People
-```ruby
-field.people_ids = ['person1', 'person2']
-field.value = ['person1']   # alias
-field.as_input  # → { people: { id:, peopleIds: [...] } }
-```
-
-### Geo
-```ruby
-field.address = '123 Main St, Auckland'
-field.coordinates = { 'lat' => -36.86, 'lon' => 174.76 }
-field.as_input  # → { geo: { id:, address:, coordinates: } }
-```
-
-### ContractorEntities
-```ruby
-field.contractor_entity_ids = ['ce1', 'ce2']
-field.value = ['ce1']   # alias
-field.as_input  # → { contractorEntities: { id:, contractorEntityIds: [...] } }
-```
-
-### CrossReference
-```ruby
-field.page_ids = ['page1', 'page2']
-field.as_input  # → { crossReference: { id:, referenceIds: [...] } }
-```
-
-### File / ImageGallery
-```ruby
-# Requires prior file upload — see TODO "File upload"
-field.file_container_ids = ['fc1', 'fc2']
-field.as_input  # → { file: { id:, fileContainerIds: [...] } }
-                # or { imageGallery: { id:, fileContainerIds: [...] } }
-```
-
----
-
-## DataFieldOneToManyInput — Wiring to Mutations
-
-`Input::Page::Update.from_model(page)` automatically collects field changes:
-
-```ruby
-input = Input::Page::Update.from_model(page)
-# input[:dataFields] may contain:
-# {
-#   updates:   [{ plainText: { id:, value: } }, ...],  # dirty existing fields
-#   additions: [{ plainText: { id:, value: } }, ...],  # newly added fields
-#   deletions: ['field_id_1', 'field_id_2']            # fields queued for deletion
-# }
-```
-
-`build_data_fields` only includes non-empty keys — if only `updates` exist, only
-`updates` is present in the hash.
-
----
-
-## Stage-Level Field Access
-
-```ruby
-page = graphql.pages.get(page_id)
-stage = page.stages['Risk Assessment']
-
-# All fields in the stage (flat, across sections)
-stage.components.get_by_name('Hazard Description').value = 'Exposure to...'
-
-# Section-by-section
-stage.sections.each do |section|
-  section.components.each { |f| puts "#{f.label}: #{f.value}" }
-end
-```
-
----
-
-## Spec Coverage
-
-| File | Coverage |
-|---|---|
-| `spec/ecoportal/api/graphql/base/page/data_field_spec.rb` | All 20 type dispatch, dirty tracking, PlainText, RichText, Select, People, Checklist, Collection (get_by_type/name, dirty_inputs, add/additions, deletions) |
-| `spec/ecoportal/api/graphql/input/page/update_spec.rb` | from_model with dataFields.updates, additions, deletions |

data/.ai-assistance/code/dependencies.md

@@ -1,151 +0,0 @@
-# Code Spec: Dependency Map
-
-**Scope:** All upstream and downstream gem dependencies for `ecoportal-api-graphql` — remote URLs, fork relationships, collaboration status, and access notes. Local paths are developer-specific; see the section below on local setup.
-**Last updated:** 2026-06-02
-
----
-
-## This Repo
-
-| Property | Value |
-|----------|-------|
-| Gem name | `ecoportal-api-graphql` |
-| Remote (origin) | https://gitlab.ecoportal.co.nz/oscar/ecoportal-api-graphql.git |
-| Relationship | Target repo — all work happens here |
-
----
-
-## Upstream Dependencies
-
-### ecoportal-api
-
-| Property | Value |
-|----------|-------|
-| Gem name | `ecoportal-api` |
-| Version constraint | `~> 0.10, >= 0.10.15` |
-| Remote | https://gitlab.ecoportal.co.nz/ecoportal/ecoportal-api.git |
-| Collaboration | **Team-managed** — we own this gem |
-| Key deps | `dotenv`, `elastic-apm`, `http`, `rate_throttle_client` |
-| Notes | Provides REST API base, authentication, org context, and the `Ecoportal::API::Common` namespace that this gem extends. |
-
-### ecoportal-api-v2
-
-| Property | Value |
-|----------|-------|
-| Gem name | `ecoportal-api-v2` |
-| Version constraint | `~> 3.3, >= 3.3.1` |
-| Remote | https://gitlab.ecoportal.co.nz/oscar/ecoportal-api-v2.git |
-| Collaboration | **Team-managed** — we own this gem |
-| Key deps | `ecoportal-api`, `mime-types` |
-| Notes | REST API v2 layer. Required by the entry point (`api-graphql.rb`). Depends on `ecoportal-api` itself. |
-
-### graphlient
-
-| Property | Value |
-|----------|-------|
-| Gem name | `graphlient` |
-| Version constraint | `>= 0.8.0, < 0.9` |
-| Remote (our fork) | https://github.com/rellampec/graphlient.git |
-| Remote (upstream) | https://github.com/ashkan18/graphlient.git |
-| Collaboration | **Fork (active)** — we drive new features; upstream is `ashkan18/graphlient` |
-| Git workflow | **Feature branches off our fork's working branch** (`feature/directives-dsl-support`). We open PRs upstream when appropriate but are not required to. Do NOT commit directly to `master`. |
-| Key deps | `faraday ~> 2.0`, `graphql-client` |
-| Notes | Friendly wrapper around `graphql-client`. `Common::GraphQL::Client` inherits from `Graphlient::Client`. |
-
-### graphql-client
-
-| Property | Value |
-|----------|-------|
-| Gem name | `graphql-client` |
-| Version constraint | (via graphlient) |
-| Remote (our fork) | https://github.com/rellampec/graphql-client.git |
-| Remote (upstream) | https://github.com/github-community-projects/graphql-client.git |
-| Collaboration | **Fork (upstream-contribution only)** — we do NOT maintain this gem; upstream is `github-community-projects/graphql-client` |
-| Git workflow | **ALWAYS use a feature branch** (e.g. `feature/allow-named-fragment-access`). Changes are intended as upstream PRs to `github-community-projects/graphql-client`. NEVER commit to `master`. If unsure about the branch name or PR strategy, ask the developer before committing. |
-| Key deps | `activesupport >= 3.0`, `graphql >= 1.13.0` |
-| Notes | Low-level GraphQL HTTP + parsing library. Pulled in transitively via `graphlient` — not a direct dependency. Check `Common::GraphQL::Patches` for any monkey-patches applied on top. |
-
-### graphql-ruby
-
-| Property | Value |
-|----------|-------|
-| Gem name | `graphql` |
-| Version constraint | `>= 1.13.0` (transitive, via graphql-client) |
-| Remote | https://github.com/rmosolgo/graphql-ruby |
-| Collaboration | **No collaboration** — external, read-only reference |
-| Notes | Schema definition and execution engine. Used server-side by EcoPortal; client-side usage here is indirect. API reference: https://graphql-ruby.org |
-
----
-
-## Downstream Dependencies
-
-### eco-helpers
-
-| Property | Value |
-|----------|-------|
-| Gem name | `eco-helpers` |
-| Requires this gem as | `ecoportal-api-graphql ~> 1.3, >= 1.3.4` |
-| Remote | https://gitlab.ecoportal.co.nz/oscar/script_api_helpers.git |
-| Collaboration | **Team-managed** — we own this gem |
-| Notes | Primary consumer of this gem. **Backwards compatibility with `eco-helpers` must be preserved.** When a change could break it, cross-reference its usage before committing. |
-
----
-
-## Dependency Chain Summary
-
-```
-graphql-ruby          (rmosolgo — external, read-only)
-        ↓
-graphql-client        (our fork ← github-community-projects/graphql-client)
-        ↓
-graphlient            (our fork ← ashkan18/graphlient)
-        ↓
-ecoportal-api         (team-managed — gitlab.ecoportal.co.nz)
-ecoportal-api-v2      (team-managed — gitlab.ecoportal.co.nz)
-        ↓
-ecoportal-api-graphql ← THIS REPO
-        ↓
-eco-helpers           (team-managed — gitlab.ecoportal.co.nz)
-```
-
----
-
-## Local Development Setup
-
-Local paths are **developer-specific** — do not hardcode them in shared docs.
-
-Each developer who works across multiple gems should clone the relevant repos and record their local paths in `.claude/local_paths.md` (git-ignored). See `.claude/local_paths.example.md` for the format.
-
-### Using local gem overrides with Bundler
-
-To point Bundler at a local clone instead of the published gem:
-
-```bash
-bundle config local.ecoportal-api      /your/path/to/ecoportal-api
-bundle config local.ecoportal-api-v2   /your/path/to/ecoportal-api-v2
-bundle config local.graphlient         /your/path/to/graphlient
-```
-
-Or add a `path:` entry in `Gemfile` during development (remove before committing):
-
-```ruby
-gem 'graphlient', path: '/your/path/to/graphlient'
-```
-
-### For AI agents working with local clones
-
-1. Check `.ai-assistance/local_paths.md` for this developer's local paths.
-2. If the file doesn't exist, the repos are not locally available — use the remote URLs for reference only (read via web fetch or ask the developer to clone).
-3. Never assume a specific local path exists — always check first.
-
----
-
-## Upstream Change Decision Guide
-
-If a task requires changes in an upstream gem, the options are (confirm with developer before proceeding):
-
-1. **Local monkey-patch** — apply the fix from within `ecoportal-api-graphql` (see `Common::GraphQL::Patches`). Fast, contained, but temporary debt.
-2. **Fork change** — commit to our fork (for `graphlient` or `graphql-client`). Appropriate for lasting fixes we own.
-3. **New version** — for `ecoportal-api` or `ecoportal-api-v2` (team-owned), cut a new version. For external forks, optionally contribute back upstream.
-
-Document the decision in the active project's `DECISIONS.md`.