Macro Data Service

Institutional-grade macro-data ingestion, resolution, and observability platform. Ingests 86 economic concepts from 11 core sources, supports a 25-source capability registry, and includes release-calendar-aware scheduling, availability verification, cross-source fallback, and a ClickHouse-backed EODHD market-data layer for US active plus delisted instruments.

Last updated: 2026-05-03 (issue #123 calendar scraper retirement: source-owned calendar APIs, te_api, and EODHD corporate calendar remain.)

Architecture

Sources (macro + EODHD)                           Ingestion              Storage              Resolution
────────────────────────────────────────────     ──────────             ─────────            ───────────
BLS, FRED, EIA                                   Fetchers + SDMX        SQLite               resolve_indicator()
NYFed, Treasury                                  Normalization           concept_map (86)     source-priority ranking
IMF, Eurostat                                    Date alignment          obs_family           cross-source alternates
BIS, ECB, OECD                                   Deduplication           release_schedule     get_market_history()
World Bank                                                               release_status       get_market_history()
EODHD         (US symbols, EOD, corp actions)    EODHDClient             ClickHouse market.*

                     ┌─────────────────────────────────────────┐
                     │  Release Schedule → Availability Check  │
                     │  → Retry (1m/5m/15m/1h/4h) → Fallback  │
                     │  → Status: CONFIRMED / STALE / FAILED   │
                     └─────────────────────────────────────────┘
                     ┌─────────────────────────────────────────┐
                     │  Health Dashboard    │  3 Alert Types    │
                     │  per-source status   │  DELAY            │
                     │  freshness + retries │  FAILED           │
                     │  provenance          │  MISMATCH         │
                     └─────────────────────────────────────────┘
                     ┌─────────────────────────────────────────┐
                     │ Source Capability Layer                 │
                     │  catalog-crawlable / discovery-rich     │
                     │  fixed-scope-complete                   │
                     │  discovery / latest-sync / status       │
                     └─────────────────────────────────────────┘
                     ┌─────────────────────────────────────────┐
                     │ Vintage + PIT Layer                     │
                     │  indicator_vintages canonical store     │
                     │  indicators SQL view over latest rows    │
                     │  as_of reads in resolve/history          │
                     └─────────────────────────────────────────┘
                     ┌─────────────────────────────────────────┐
                     │ Bronze Replay Layer                     │
                     │  obs_raw + cal_*_raw + fundamentals_raw  │
                     │  raw payload → canonical silver rows     │
                     │  per-source replay round trips           │
                     └─────────────────────────────────────────┘
                     ┌─────────────────────────────────────────┐
                     │ Market-Data Layer (#118/#119)           │
                     │  EODHD US active + delisted universe    │
                     │  ClickHouse bars_1d/dividends/splits    │
                     │  Daily bulk refresh + DELETE/refetch    │
                     │  Self-heal + realtime spot checks       │
                     └─────────────────────────────────────────┘

Layout

src/
  ingestion/
    timeseries/       FRED, BLS, EIA, Treasury, NYFed, WorldBank + SDMX providers
      scrapers/       Source-specific API clients (fred.py, bls.py, eia.py, ...)
      fetchers/       RawSeries adapters (Fetcher protocol)
      clients/        Ingestion clients (FREDIngestionClient, ...)
      sdmx/           Unified SDMX engine + providers/ (IMF, ECB, BIS, OECD, ...)
      _config.py      Series configs (MACRO_SERIES, BLS_SERIES, OECD_SERIES, ...)
    news/             Reuters, FT, WSJ, Bloomberg
      scrapers/       Article listing scrapers
      client.py       NewsIngestionClient
      _config.py      RSS feed registry (140 feeds)
    documents/        Government reports, Fed communications
      scrapers/       gov_report.py
      clients/        GovReportIngestionClient, FedIngestionClient
      _config.py      FED_FEEDS, FED_SPEAKERS
    trends/           Reddit + Weibo social trend tracking
      scrapers/       reddit.py, weibo.py upstream readers
      clients/        _trends.py ingestion clients
    sentiment/        X (Twitter) sentiment lane
      x_api/          X API client, scrapers, discovery
    market/           EODHD US universe, EOD bars, corporate actions
      scrapers/       Provider HTTP clients
        _eodhd.py        EODHDClient — symbol lists, EOD, bulk, realtime
        _eodhd_identity.py  EODHD search / delisted / symbol-change / exchanges
        _openfigi.py     OpenFIGIClient — /v3/mapping
      clients/        High-level providers
        _eodhd.py        EODHDMarketDataProvider (ClickHouse)
        _macro_market.py MacroMarketProvider (FRED/EIA/ECB projection)
        _identity_repair.py Historical identity-repair prototype
      _quality.py          Provider-neutral bar checks
      _macro_map.py        13-entry rates/FX/commodity mapping
      _config.py           MACRO_WATCHLIST (real-time watchlist)
    calendar/         Economic + corporate calendars
      *_api/          Source-owned packages (abs_api, banxico_api, bcb_api,
                      bls_api, boc_api, boe_api, boj_api, ecb_api, ...)
      te_api/         TradingEconomics historical bootstrap + updates
      eodhd_api/      EODHD earnings, IPOs, splits, dividends
      scheduler.py    Recurring schedule/value refresh driver
      agency_registry.py  Agency attribution map for parity filings
      parity.py       TE-vs-official parity harness
      parity_daily.py Daily parity comparator
      parity_filer.py Parity issue filer
      evidence_archive.py  Wayback evidence archive helper
    _shared/          Cross-cutting: http_transport, url_canon, selector versioning
    validation/       Data quality checks, cross-source consistency
    sources.py        IngestionOrchestrator — scheduling, retry, health
    source_capabilities.py  Unified capability registry + discovery/sync adapters
    release_schedule.py     Date-math resolvers, availability checks, alert logic
    scrapers/         Facade — backward-compatible re-exports
    clients/          Facade — backward-compatible re-exports
    fetchers/         Facade — backward-compatible re-exports
    sdmx/             Facade — backward-compatible re-exports
  storage/            SQLite macro schema/query mixins + ClickHouse market backend
    clickhouse/       market.bars_1d/dividends/splits/instruments store
  macro_data/         CLI, HTTP server, service layer, factory
  contracts.py        Date normalization, epoch utilities

Concepts covered (86)

Category	Examples
US Inflation	CPI_US, CORE_CPI_US, PPI_US, CORE_PCE_US, breakevens
US Employment	NFP_US, UNEMP_US, JOLTS, initial/continuing claims, ECI
US Growth	GDP_REAL_US, GDP_NOMINAL_US, retail sales, industrial production
US Rates	POLICY_RATE_US, SOFR, OBFR, 2Y/10Y/30Y treasuries, yield curve
US Liquidity	Fed balance sheet, M2, reverse repo, TGA
US Credit/Fiscal	HY OAS, credit gap, debt outstanding, avg interest rate
US Energy	WTI, Brent, natural gas, crude stocks, petroleum supply
Euro Area	CPI_EU, GDP_EU, UNEMP_EU, ECB policy rate, M1/M2/M3, EURUSD
China	CPI_CN, GDP_REAL_CN, FX reserves, PBOC rate, credit gap
Japan	CPI_JP, GDP_REAL_JP, BOJ rate
Cross-country	OECD CLI (US/CN/JP/EU), BIS EER/property, World Bank development

CLI

Core operations

macro-data-service serve                              # start HTTP server
macro-data-service refresh                            # refresh all sources
macro-data-service refresh-source --source fred       # refresh single source
macro-data-service refresh-indicator CPI_US           # refresh single concept

# Market-data operations:
PYTHONPATH=src python3 scripts/seed_market_universe.py
PYTHONPATH=src python3 scripts/backfill_market_history.py --status active
PYTHONPATH=src python3 scripts/market_daily_refresh.py --date 2026-05-01

Writer and source-diagnostic commands run on the VPS with MACRO_DATA_PROFILE=prod. Local fixture/debug writer runs use --allow-local-write and a temporary --db-path.

Resolution

macro-data-service resolve CPI_US                     # latest resolved value
macro-data-service resolve CPI_US --date 2024-01-01   # specific date
macro-data-service resolve CPI_US --history --limit 12  # resolved time series
macro-data-service resolve CPI_US --as-of 2024-03-01  # point-in-time vintage view
macro-data-service resolve CPI_US --json              # JSON with provenance

Release schedule

macro-data-service schedule --show                    # all 86 concepts + next_expected
macro-data-service schedule --due                     # concepts due within 2h window
macro-data-service schedule --status                  # availability status per concept
macro-data-service schedule --run                     # start release-aware scheduler
macro-data-service schedule --show --json             # JSON output

Health dashboard and alerts

macro-data-service health                             # per-source status table
macro-data-service health --indicator CPI_US          # single concept
macro-data-service health --alerts                    # active alerts only
macro-data-service health --json                      # JSON output
macro-data-service source-health                      # customer-facing source matrix
macro-data-service source-health --all               # include internal/coming-soon sources

Output:

  indicator                 source       status       freshness  retries  note
  -------------------------------------------------------------------------------------
  CPI_US                    bls          CONFIRMED    fresh      0        primary
  CPI_US                    fred         CONFIRMED    fresh      0        secondary
  GDP_REAL_US               fred         WAITING      stale      2        primary
  SOFR_US                   nyfed        CONFIRMED    fresh      0        primary

Three alert types:

DELAY — data not confirmed 30min after expected release
FAILED — fetch retries exhausted (5 attempts)
MISMATCH — cross-source value divergence exceeds threshold

Validation

macro-data-service validate                           # full validation suite
macro-data-service validate-concept CPI_US            # single concept
macro-data-service validate-concept --all             # all concepts
macro-data-service launch-gate --json                 # production data-quality gate

Catalog exploration (OECD / World Bank)

macro-data-service oecd-dataflows --query inflation
macro-data-service oecd-structure --dataflow QNA
macro-data-service wb-sources
macro-data-service wb-indicators --query gdp

Source capabilities and catalog sync

macro-data-service list-sources                                # every registered source as {name, family}
macro-data-service list-sources --family market_price          # filter to one family
macro-data-service sources-capabilities
macro-data-service catalog-list --source oecd --refresh --limit 10
macro-data-service catalog-structure --source ecb --entity ECB_EA_DEPOSIT_RATE
macro-data-service catalog-sync-discovery --source worldbank --limit 10
macro-data-service catalog-sync-latest --source worldbank --entity SP.POP.TOTL --limit 1
macro-data-service catalog-status --source worldbank

Source modes:

catalog-crawlable — source supports discovery plus catalog-based latest sync
discovery-rich — source exposes discovery/structure metadata, but latest sync still maps to curated or configured paths
fixed-scope-complete — source has a complete supported-entity surface, but no meaningful upstream full-catalog crawl

Operational notes:

Capability structure output may be a live provider structure summary or a config-backed summary, depending on the source mode.
Capability latest-sync is best-effort and now surfaces upstream HTTP/timeouts instead of silently returning false-green zero results.
Some scaffolded sources can validly return zero discovered entities when the upstream provider or local configured series set is empty.
Customer-facing health surfaces hide ILO and UNSD until they return non-empty entity catalogs.
news disables known bad feeds and drops low-content articles.
gov_reports falls back to RSS description when full content extraction fails; metadata-only records are stored for PDF/asset sources.
eia prefers live EIA data, then recent local cache, then selected FRED fallback series. Timeout is configurable via EIA_TIMEOUT env var (default 60s). Failed series degrade gracefully instead of crashing the batch.
fred_nondaily source refreshes weekly/monthly/quarterly FRED series (ICSA, CCSA, WALCL, PCEPILFE, INDPRO, RSAFS, M2SL, GDP, GDPC1) every 6h, complementing fred_daily which handles daily series only.

Revision and replay layers

indicator_vintages is the canonical macro time-series write target. indicators is a SQL view over the latest vintage per (source, series_id, observation_date), and resolve_indicator / resolve_indicator_history accept as_of=YYYY-MM-DD to reconstruct point-in-time projections from vintages.

Routine time-series fetchers follow a bronze-to-silver dual-write contract: raw HTTP payloads land in obs_raw before normalization writes parsed observations to indicator_vintages. Canonicalizers in ingestion/timeseries/canonicalize.py hash stable observation content, while per-source _parse_* replay helpers re-project stored payloads through the same parsers used by live fetches.

Calendar and fundamentals use the same replay shape. cal_econ_raw, cal_corp_raw, and fundamentals_raw preserve provider payloads; cal_econ_event, cal_corp_event, and fundamentals_* tables expose parsed projections.

HTTP API

The API is served by FastAPI under uvicorn. GET /healthz is the liveness probe and is open for uptime checks. Other routes require X-API-Key; tokens are loaded at startup from /etc/macro-data/api_tokens.json unless ANALYST_MACRO_DATA_API_TOKENS_PATH is set.

GET /health customer-facing source health matrix
GET /healthz lightweight liveness probe
GET /v1/manifest data availability manifest for consumer routing
GET /v1/calendar unified calendar feed
GET /v1/calendar/revisions corporate calendar revision feed
GET /v1/fundamentals/{ticker} fundamentals projection
POST /v1/ops/<operation>

Key operations:

Operation	Description
`get_data_manifest`	Dataset availability manifest with status, row counts, freshness, and quality status
`resolve_indicator`	Highest-priority observation for a concept; accepts `as_of` for PIT vintage reads
`resolve_indicator_history`	Resolved time series with best source per date; accepts `as_of`
`get_release_schedule`	Release calendar for all/single concept
`get_release_status`	Availability tracking status
`get_health`	Per-source health dashboard
`get_alerts`	Active alerts (DELAY, FAILED, MISMATCH)
`list_sources`	Registered ingestion sources as `{name, family}` rows, optional `family` filter
`list_source_capabilities`	Capability registry for all known sources
`get_source_health_dashboard`	Customer-facing per-source health summary
`sync_catalog_discovery`	Persist discovered provider entities for a source
`list_catalog_entities`	List stored catalog entities for a source
`get_catalog_structure`	Source-specific structure/config summary for an entity
`sync_catalog_latest`	Run source-level latest sync through the capability layer
`get_catalog_status`	Checkpoints and recent runs for capability jobs
`refresh_indicator`	Trigger ingestion for a single concept
`validate_concept`	Run validation checks on a concept
`get_indicator_ontology`	Structural metadata for an indicator
`get_recent_releases`	Recent economic data releases
`get_upcoming_calendar`	Upcoming calendar events
`get_market_snapshot`	Latest market prices
`list_items`	Cross-type feed for a subject: documents (news / gov reports / notes) unioned with indicator observations (via `concept_map` + direct `subject_aliases`) and market rows from the service layer. Every row carries a `family` + `kind` tag; optional `family` filter narrows to one bucket. Still accepts `q` (FTS5, documents only), `document_type`, `country_code`, `min_confidence`
`get_document`	Single document by `document_id` or `hash_sha256` (17-field summary + markdown body + subject tags)
`list_subjects`	Subject vocabulary (auto-synced from `src/storage/subjects.yaml`)
`backfill_document_indexes`	One-shot FTS + subject-tag backfill for DBs whose documents predate the new sidecars (idempotent)

Information layer

Non-numeric sources (news, gov reports, research notes) land in a shared document surface keyed by a canonical subject vocabulary, so downstream callers can pull them alongside the macro timeseries through one API. The schema additions live in src/storage/sqlite.py:

subjects / subject_aliases / item_subjects — canonical subject IDs (econ.cpi, rate.us.sofr, …) and the aliases that map source-native identifiers (FRED series, calendar indicators, title regex) back to them. Seeded from src/storage/subjects.yaml.
document carries the 17-field information surface (institution, authors, asset_class, impact_level, contains_commentary, confidence, …), with full-text search in documents_fts.
obs_enrichment sidecar — derived labels keyed by (obs_family_id, date, key). Currently used for VIX regime classification.

Ingestion paths:

news — NewsIngestionClient mirrors each article into document (document_type='report', source_id='news') + documents_fts + item_subjects, alongside the legacy news_articles row. News rows carry scraper-supplied metadata.
gov_reports — GovReportIngestionClient merges scraper metadata with optional LLM extraction into the 17-field document surface.
notes — python -m ingestion.notes.ingest --input <dir> parses YAML-frontmatter markdown into document (source_id='notes', document_type='report', confidence=1.0).
FEDWATCH_US — CME-equivalent midpoint persisted daily from rateprobability.com; bridges to the rate.us.fedwatch subject.
VIX_US — FRED VIXCLS close; refresh_vix_regime writes a low/elevated/stressed label into obs_enrichment per date.

Optional gov-report LLM enrichment of institution, asset_class, impact_level, and the rest of the 17-field surface is controlled by DOCUMENT_EXTRACT_API_KEY (or OPENAI_API_KEY) plus DOCUMENT_EXTRACT_MODEL / DOCUMENT_EXTRACT_BASE_URL. Unset → ingestion uses scraper-supplied metadata.

Source families (issue #5)

Every IngestionSourceDefinition carries a family tag applied at registration via SOURCE_FAMILIES in src/ingestion/sources.py. The tag flows through IngestionRunReport.to_dict(), the list_sources op and the list_items envelope so downstream callers can group sources and rows by type without reflecting on names:

Family	Sources
`economic_data`	`fred_daily`, `fred_nondaily`, `fred_full`, `fred_vintages`, `bls`, `eia`, `treasury_fiscal`, `nyfed_rates`, `imf`, `imf_vintages`, `eurostat`, `bis`, `ecb`, `oecd`, `worldbank`, `worldbank_catalog`
`market_price`	ClickHouse market read ops and EODHD maintenance scripts
`release_report`	`gov_reports`, `fed`
`news`	`news`
`calendar`	`calendar`
`trend`	`reddit_trends`, `weibo_trends`
`signal`	`rate_probability`

list_items(subject="econ.cpi") returns the merged envelope: documents (via item_subjects), indicator observations (via subject_aliases → concept_map → indicators, pivoted through concept_id so cross-source alternates surface), and market rows from the service layer. Each row carries a family + kind discriminator; an optional family filter narrows the result.

Market-data layer

The market lane uses src/storage/clickhouse/ and ClickHouse tables under the market database: bars_1d, dividends, splits, and instruments. EODHD is the provider. Universe seeding discovers US active symbols and active plus delisted symbols from /api/exchange-symbol-list/US; is_active is derived from set membership. SQLite stays the macro store for calendar, indicators, documents, news, and fundamentals.

Historical backfill uses per-ticker /api/eod/{T}.US, /api/div/{T}.US, and /api/splits/{T}.US with the resumable cursor .macro-data/market_backfill_cursor.json. Daily refresh uses /api/eod-bulk-last-day/US for bars, dividends, and splits. New corp actions append to market.dividends / market.splits, then the ticker's bars are deleted and re-fetched from EODHD so adjusted_close stays provider-trusted.

Operational entrypoints:

scripts/seed_market_universe.py
scripts/backfill_market_history.py
scripts/market_daily_refresh.py
scripts/detect_missing_bars.py
scripts/spot_check_close.py

Quality checks

The market maintenance scripts check stale active tickers, delisting set membership, and sampled realtime close drift.

Local ClickHouse storage is booted by docker-compose.yml and persists under .macro-data/clickhouse/{data,tmp,user_files,logs}. The get_market_history service op reads daily bars from this ClickHouse lane.

Data model

┌─────────────────┐     ┌──────────────┐     ┌─────────────────┐
│  concept_map    │────▶│  indicators  │     │ release_schedule │
│  86 concepts    │     │  observations│     │  86 rules        │
│  11 sources     │     │  per source  │     │  8 rule types    │
│  priority ranks │     └──────┬───────┘     └────────┬────────┘
└─────────────────┘            │                      │
                               │ (macro projection)   │
                               ▼                      ▼
                   ┌───────────────────────┐ ┌───────────────────┐
                   │  ClickHouse market    │ │ release_status    │
                   │  instruments          │ │  PENDING / WAITING│
                   │  bars_1d/div/splits   │ │  FETCHED / CONFIRM│
                   │  US active+delisted   │ │  STALE / FAILED   │
                   └───────────────────────┘ └───────────────────┘
┌─────────────────┐
│  obs_family     │
│  obs_source     │
│  calendar_*     │
│  documents      │
│  fundamentals_* │
└─────────────────┘

┌─────────────────────┐    ┌──────────────────────┐
│ source_capability   │───▶│ catalog_entity       │
│ source mode + flags │    │ discovered entities  │
└─────────────────────┘    └──────────┬───────────┘
                                      │
                         ┌────────────▼────────────┐
                         │ catalog_sync_checkpoint │
                         │ catalog_sync_run        │
                         └─────────────────────────┘

Shadow mode

Continuous ingestion runner with structured logging and daily digest.

python shadow_runner.py --interval 6    # run full cycle every 6 hours
python shadow_status.py                 # show latest digest + recent errors

Logs: .macro-data/logs/shadow.log, .macro-data/logs/daily_digest.jsonl

Production status (2026-03-22): 86/86 concepts covered (100%), 12 sources active, cycle time ~16 min.

Local run

python -m venv .venv
. .venv/bin/activate
pip install -e .
macro-data-service serve --host 127.0.0.1 --port 8765

SQLite path: .macro-data/engine.db

API token file shape:

{
  "tokens": {
    "secret-token": {
      "consumer_id": "analyst-project",
      "rate_limit_per_minute": 60
    }
  }
}

Environment variables

src/env.py reads from process env first, then .env at the repo root, then ~/.macro-data/dev.env.

Variable	Used by	Required?
`MACRO_DATA_PROFILE`	Deployment profile: `dev` local client, `prod` VPS writer/API	optional; default `dev`
`FRED_API_KEY`	FRED series, FX, and macro-market rate/FX projection	yes for any FRED data
`BLS_API_KEY`	BLS series	yes for BLS
`BEA_API_KEY`	BEA series	yes for BEA
`CENSUS_API_KEY`	Census series	yes for Census
`EIA_API_KEY`	EIA + macro-market commodity projection	yes for EIA
`IMF_API_KEY`	IMF SDMX	yes for IMF
`EODHD_API_KEY`	EODHD market universe, EOD bars, corp actions, fundamentals	yes for market/fundamentals
`OPENFIGI_API_KEY`	historical identity-repair prototype	optional (anonymous tier works)
`DOCUMENT_EXTRACT_API_KEY` / `OPENAI_API_KEY`	Gov-report 17-field LLM extraction (`make_extractor_from_env`, reads `os.environ` only — not `.env`)	optional; unset → scraper metadata
`DOCUMENT_EXTRACT_MODEL`, `DOCUMENT_EXTRACT_BASE_URL`, `DOCUMENT_EXTRACT_CONTEXT_CHARS`	Model / endpoint / chunk-size overrides for the gov-report extractor (same process-env-only loading)	optional
`ANALYST_MACRO_DATA_API_TOKENS_PATH`	FastAPI public API token mapping path	optional; default `/etc/macro-data/api_tokens.json`
`ANALYST_MACRO_DATA_RATE_LIMIT_DB`	shared SQLite fixed-window rate-limit state	optional; default `/tmp/macro-data-api-rate-limits.sqlite3`
`ANALYST_MACRO_DATA_WORKERS`	uvicorn worker count for `macro-data-service serve`	optional; default `2`
`ANALYST_MACRO_DATA_BASE_URL`	dev-profile client base URL for prod API reads	yes for local `resolve`
`ANALYST_MACRO_DATA_API_TOKEN`	client token sent as `X-API-Key`; server also accepts it as a local inline token for CLI launches	optional

Agent wiring

Point analyst-project at this service:

ANALYST_MACRO_DATA_BASE_URL=http://127.0.0.1:8765
ANALYST_MACRO_DATA_API_TOKEN=

Tests

python -m pytest tests/ -v                            # all tests
python -m pytest tests/test_release_schedule.py -v    # schedule + alerts (58 tests)
python -m pytest tests/test_macro_data_cli.py -v      # CLI smoke tests
python -m pytest tests/test_fetcher_adapters.py -v    # fetcher adapter tests
python -m pytest tests/test_source_capabilities.py -v # capability registry + catalog ops

Name		Name	Last commit message	Last commit date
.github		.github
docs		docs
scripts		scripts
src		src
tests		tests
tools		tools
.coderabbit.yaml		.coderabbit.yaml
.gitignore		.gitignore
README.md		README.md
docker-compose.yml		docker-compose.yml
pyproject.toml		pyproject.toml
shadow_runner.py		shadow_runner.py
shadow_status.py		shadow_status.py

Navigation Menu

Repository navigation

About

Repository files navigation

Macro Data Service

Architecture

Layout

Concepts covered (86)

CLI

Core operations

Resolution

Release schedule

Health dashboard and alerts

Validation

Catalog exploration (OECD / World Bank)

Source capabilities and catalog sync

Revision and replay layers

HTTP API

Information layer

Source families (issue #5)

Market-data layer

Quality checks

Data model

Shadow mode

Local run

Environment variables

Agent wiring

Tests

About

Releases

Packages

Contributors

Languages

Latest commit

History

Folders and files

Repository files navigation

Macro Data Service

Architecture

Layout

Concepts covered (86)

CLI

Core operations

Resolution

Release schedule

Health dashboard and alerts

Validation

Catalog exploration (OECD / World Bank)

Source capabilities and catalog sync

Revision and replay layers

HTTP API

Information layer

Source families (issue #5)

Market-data layer

Quality checks

Data model

Shadow mode

Local run

Environment variables

Agent wiring

Tests

About

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages