160-tool MCP server for real-time global intelligence from free public APIs: markets, liquidity, prediction markets, SEC filings, conflict, military, aircraft-cohort early warning, migration and border flows, cyber, climate, news and more. Dated observation history, 29 research prompts, cited briefs, geofences, SSE dashboard. MIT, no paid API keys.
Python ai-toolsai-watchanthropicclaudecybersecuritydashboardgeofencinggeopoliticalgeospatialintelligencemarketsmcpmigrationmilitarymodel-context-protocolnuclearosintreal-timespace-weatherthreat-intelligence
Prompts: add steps for the merged tools that answer them
Phoenix Agent (mac-studio)5182ef82b08f
diff --git a/CHANGELOG.md b/CHANGELOG.mdindex 0f57d55..d6909c5 100644--- a/CHANGELOG.md+++ b/CHANGELOG.md@@ -372,6 +372,17 @@ source here, and no proxy is substituted for any of these measures.not the file source health first read), the verdict says the numberscome from a file no other process shares, and the scope notes say areplacement made before the first read is not detected.+- Prompts now name the merged tools that answer their questions:+ story-breadth (outlet breadth, ranked situations, per-day hostility+ coverage), election-watch (Kalshi Elections markets and the cross-venue+ comparison), the migration prompts (US border encounters, UK small boat+ arrivals), the AOI prompts (polygon and corridor areas, the all-area+ digest), market-risk-snapshot (liquidity board, net liquidity),+ company-events (news and market reaction), yield-curve-check and+ central-bank-compare (a keyless FRED backfill read back from the history+ store when `FRED_API_KEY` is not set) and history-recall (the history+ store and the collector's run log). The source-health scope points to+ `intel_collection_health` for the run history it does not read.### Added: tropical cyclone forecast tracks- `intel_cyclones` and `intel_jtwc_cyclones` take `include_forecast`diff --git a/src/world_intel_mcp/guide.py b/src/world_intel_mcp/guide.pyindex 275a079..e22608a 100644--- a/src/world_intel_mcp/guide.py+++ b/src/world_intel_mcp/guide.py@@ -484,7 +484,8 @@ _HEALTH_SCOPE = ("keyed_sources says whether each variable is set in this server's ""environment. It never shows values and does not check that a key works.","Not read here: vector store statistics (intel_vector_stats) and the "- "collector's run history. Reading this resource starts no fetches.",+ "collector's run history (intel_collection_health). Reading this "+ "resource starts no fetches.",)diff --git a/src/world_intel_mcp/guide_prompts.py b/src/world_intel_mcp/guide_prompts.pyindex 0bf6c90..3ca9e1c 100644--- a/src/world_intel_mcp/guide_prompts.py+++ b/src/world_intel_mcp/guide_prompts.py@@ -453,6 +453,17 @@ PROMPTS: tuple[ResearchPrompt, ...] = (("intel_sector_heatmap", "which US equity sectors are leading and lagging"),("intel_yield_curve", "curve shape and spreads"),("intel_macro_composite", "the composite score and the inputs it weighs"),+ (+ "intel_liquidity_board",+ "keyless liquidity readings (Fed balance sheet, Treasury General "+ "Account, reverse repo, net liquidity, money supply), each with "+ "its units and observation date",+ ),+ (+ "intel_net_liquidity",+ "only if liquidity matters to the answer: the weekly US net "+ "liquidity series behind the board's row, with its formula",+ ),),limits=("Report current readings and their recent change only. Quote the "@@ -476,6 +487,19 @@ PROMPTS: tuple[ResearchPrompt, ...] = (args=(("series_id", "T10Y2Y"),),),("intel_central_bank_rates", "the policy rate the short end is anchored to"),+ Step(+ "intel_history_backfill",+ "only if FRED_API_KEY is not set: loads the T10Y2Y history from "+ "FRED's keyless graph CSV into the history store (it stores no "+ "unit for this series; a failed fetch is an error capture)",+ args=(("provider", "fred"), ("series", ["T10Y2Y"])),+ ),+ Step(+ "intel_series_history",+ "only after that backfill: the stored T10Y2Y observations with "+ "their observed_at dates",+ args=(("series_key", "backfill:fred:T10Y2Y"),),+ ),),limits=("Spreads are in percentage points. A move from 0.10 to 0.30 is "@@ -510,6 +534,19 @@ PROMPTS: tuple[ResearchPrompt, ...] = (args=(("series_id", "DFF"),),),("intel_major_crosses", "how the major currencies sit against each other today"),+ Step(+ "intel_history_backfill",+ "only if FRED_API_KEY is not set: loads the effective federal "+ "funds rate history from FRED's keyless graph CSV into the "+ "history store (a failed fetch is an error capture)",+ args=(("provider", "fred"), ("series", ["DFF"])),+ ),+ Step(+ "intel_series_history",+ "only after that backfill: the stored observations of that "+ "rate with their observed_at dates",+ args=(("series_key", "backfill:fred:DFF"),),+ ),),limits=("Only the Fed, ECB and BoE rates can be live, read from FRED when "@@ -590,6 +627,13 @@ PROMPTS: tuple[ResearchPrompt, ...] = ("actual against estimated earnings",args=(("symbol", "{ticker}"),),),+ (+ "intel_news_market_reaction",+ "companies named in recent news story clusters with the stock's "+ "move on the story's session against a benchmark. If {ticker} "+ "is not listed, no recent cluster was mapped to it, which is "+ "not the same as no news",+ ),),limits=("A filing date is when the company filed, not when the event "@@ -811,6 +855,13 @@ PROMPTS: tuple[ResearchPrompt, ...] = (args=(("origin", "{country}"),),),("intel_displacement_summary", "UNHCR's global displacement totals, for context"),+ Step(+ "intel_us_border_encounters",+ "only for the origin view: US CBP encounters per month of "+ "{country} citizens, a flow of events (not people, not a "+ "stock) that never adds to the displacement figures above",+ args=(("citizenship", Fill("the country as CBP names it, in capitals, e.g. MEXICO")),),+ ),),limits=("UNHCR figures are year-end stocks (people displaced at the end of "@@ -842,6 +893,18 @@ PROMPTS: tuple[ResearchPrompt, ...] = (),("intel_net_migration", "World Bank net migration for the leading countries"),("intel_humanitarian_summary", "HDX crisis datasets for the leading origins"),+ (+ "intel_us_border_encounters",+ "only if the question turns to arrivals in the US: CBP "+ "encounters per month with same-month prior-year change, a "+ "flow of events, not a stock",+ ),+ (+ "intel_uk_small_boat_arrivals",+ "only if the question turns to arrivals in the UK: Home Office "+ "small boat crossings with year to date against last year, a "+ "flow of arrivals, not a stock",+ ),),limits=("World Bank net migration counts every reason, work included; "@@ -1051,6 +1114,24 @@ PROMPTS: tuple[ResearchPrompt, ...] = ("saves the area",args=(("name", "{name}"), ("lat", "{lat}"), ("lon", "{lon}"), ("radius_km", "{radius_km}")),),+ Step(+ "intel_aoi_define_polygon",+ "only if a circle does not fit the area (a border region or a "+ "strait): define it under another name as a polygon of [lat, "+ "lon] vertices, and use that name in the steps below",+ args=(("name", Fill("a name for the polygon")), ("vertices", Fill("3 to 64 [lat, lon] vertices"))),+ ),+ Step(+ "intel_aoi_define_corridor",+ "only if the area is a route (a shipping lane or cable run): "+ "define it under another name as a corridor of waypoints with "+ "a width, and use that name in the steps below",+ args=(+ ("name", Fill("a name for the corridor")),+ ("waypoints", Fill("2 to 64 [lat, lon] waypoints")),+ ("width_km", Fill("the total width in km, 1 to 500")),+ ),+ ),Step("intel_aoi_brief", "the cited picture inside {name} today", args=(("name", "{name}"),)),Step("intel_aoi_changes",@@ -1088,6 +1169,12 @@ PROMPTS: tuple[ResearchPrompt, ...] = (args=(("name", "{name}"),),),Step("intel_aoi_brief", "cited detail behind any change worth reporting", args=(("name", "{name}"),)),+ (+ "intel_aoi_digest",+ "only if asked about every watched area: the change sweep "+ "across all defined AOIs in one call. It advances each AOI's "+ "snapshot as step 2 does, so never run both for {name}",+ ),),limits=("Changes are relative to the previous snapshot, whenever it was "@@ -1118,6 +1205,24 @@ PROMPTS: tuple[ResearchPrompt, ...] = (Step("intel_gdelt_search", "the articles themselves", args=(("query", "{topic}"), ("mode", "artlist"))),("intel_news_clusters", "the clusters of related stories in the RSS set"),("intel_news_feed", "the latest items from the RSS feeds"),+ (+ "intel_top_stories",+ "stories carried by at least three distinct outlets in the RSS "+ "feeds over the last 24 hours, ranked by outlet count: breadth "+ "across outlets",+ ),+ (+ "intel_situations",+ "situations between and inside countries ranked by how many RSS "+ "headlines cover them",+ ),+ Step(+ "intel_hostility_timeline",+ "only if the story is about one country or a pair: per-day "+ "GDELT article counts for it, where a day GDELT did not return "+ "is null and listed in data_gaps",+ args=(("country", Fill("the country the story is about")),),+ ),),limits=("Report the timelinevol series as GDELT returns it and compare it "@@ -1157,6 +1262,19 @@ PROMPTS: tuple[ResearchPrompt, ...] = ("volume. It has no topic filter, so most will be about other ""things: keep any about this election and say so if there are none",),+ Step(+ "intel_kalshi_markets",+ "Kalshi open markets in its Elections category, ranked by "+ "volume; keep only those about this election",+ args=(("category", "Elections"),),+ ),+ Step(+ "intel_prediction_compare",+ "the same Elections markets on Kalshi and Polymarket side by "+ "side with the price gap; match_confidence ranks pairs and is "+ "not a probability",+ args=(("category", "Elections"),),+ ),),limits=("Proximity risk rises as an election nears; it is a stated "@@ -1252,6 +1370,18 @@ PROMPTS: tuple[ResearchPrompt, ...] = ("earlier stored entries that resemble the strongest match",args=(("text", Fill("the text of the strongest match from step 3")),),),+ (+ "intel_series_history",+ "the numeric series the history store holds (a separate store "+ "from the vector store), with observed_at, published_at and "+ "collected_at for each",+ ),+ (+ "intel_collection_health",+ "the collector's run log per source: last success, consecutive "+ "failures and whether it is overdue, which says how complete "+ "that history is",+ ),),limits=("History exists only for what the collector stored while it was "diff --git a/src/world_intel_mcp/tests/test_guide.py b/src/world_intel_mcp/tests/test_guide.pyindex 8f3b32f..60e7ac9 100644--- a/src/world_intel_mcp/tests/test_guide.py+++ b/src/world_intel_mcp/tests/test_guide.py@@ -1856,3 +1856,34 @@ def test_source_health_marks_a_cache_file_moved_or_replaced_under_it(tmp_path) -other.close()scope = " ".join(guide._HEALTH_SCOPE)assert "cache.moved is true when the file at db_path is gone" in scope+++def test_prompts_point_to_the_tools_that_answer_them_best(registered) -> None:+ """Merged tools that answer a prompt's question are steps in it, so a+ model is not left with a weaker tool or told a key is needed."""+ expected = {+ "story-breadth": {"intel_top_stories", "intel_situations", "intel_hostility_timeline"},+ "election-watch": {"intel_kalshi_markets", "intel_prediction_compare"},+ "displacement-history": {"intel_us_border_encounters"},+ "migration-surges": {"intel_us_border_encounters", "intel_uk_small_boat_arrivals"},+ "aoi-watch-setup": {"intel_aoi_define_polygon", "intel_aoi_define_corridor"},+ "aoi-review": {"intel_aoi_digest"},+ "market-risk-snapshot": {"intel_liquidity_board", "intel_net_liquidity"},+ "company-events": {"intel_news_market_reaction"},+ "yield-curve-check": {"intel_history_backfill", "intel_series_history"},+ "central-bank-compare": {"intel_history_backfill", "intel_series_history"},+ "history-recall": {"intel_series_history", "intel_collection_health"},+ }+ missing = {+ name: sorted(tools - set(guide_prompts.get(name).tools))+ for name, tools in expected.items()+ if tools - set(guide_prompts.get(name).tools)+ }+ assert missing == {}+ assert set().union(*expected.values()) <= registered+ # The keyless route reads the key the backfill stores the series under.+ for name, series in (("yield-curve-check", "T10Y2Y"), ("central-bank-compare", "DFF")):+ calls = {s.tool: c for s, c in guide_prompts.resolved_steps(guide_prompts.get(name), {})}+ assert calls["intel_history_backfill"] == {"provider": "fred", "series": [series]}+ assert calls["intel_series_history"] == {"series_key": f"backfill:fred:{series}"}+ assert "(intel_collection_health)" in " ".join(guide._HEALTH_SCOPE)