coilyco-flight-deck/eco-mcp-app
1
0
Fork 0
Code Issues 18 Pull requests Projects Releases Packages Wiki Activity Actions 2

Fair-Price Advisor #8

New issue
Open
opened 2026-05-23 20:54:10 +00:00 by coilysiren · 0 comments
coilysiren commented 2026-05-23 20:54:10 +00:00
Owner
Copy link

Originally filed by @coilysiren on 2026-05-02T23:10:12Z - https://github.com/coilysiren/eco-mcp-app/issues/17

Task 8 — Fair-Price Advisor

Prereq: read todo/README.md first.

Goal

Build an MCP tool fair_price(item) that returns a short narrative card using real-world commodity data.

Data sources

  • SSM /eco-mcp-app/fred-api-key (SecureString, region us-east-1) — already provisioned. Fetch once at app start.
  • FRED API GET https://api.stlouisfed.org/fred/series/observations?series_id=X&api_key=KEY&file_type=json

Internal mapping: eco item → FRED series

Eco item FRED series Cadence
CopperIngot PCOPPUSDM monthly
Wheat PWHEAMTUSDM monthly
Board (lumber) WPU0811 monthly
IronIngot PIORECRUSDM monthly
Oil DCOILWTICO daily

Cadence matters — don't compute "7-day %" for monthly series

Most of the cheap FRED series for commodities publish monthly. Computing a "7-day percent change" against a monthly series is meaningless — you'll either get zero (same observation repeating) or garbage when the observation flips.

Branch the math on series frequency:

def latest_pct_changes(observations, frequency):
    # observations: list of {date, value}, newest last
    if frequency == "daily":
        return {
            "7d":  pct(obs, lookback_days=7),
            "30d": pct(obs, lookback_days=30),
            "90d": pct(obs, lookback_days=90),
        }
    elif frequency == "monthly":
        return {
            "1m":  pct(obs, lookback=1),   # latest vs prior observation
            "3m":  pct(obs, lookback=3),
            "12m": pct(obs, lookback=12),
        }

Read the frequency from FRED's /series?series_id=X response (frequency_short: "D", "M", etc.).

On each call

  1. Fetch ~90 days (daily series) or ~13 observations (monthly) of the mapped series from FRED.
  2. Compute percent changes using the branching rule above.
  3. Return a narrative card:

    Real copper: $4.12/lb (monthly series, latest Mar 2026). Up 3.2% vs prior month, 12.1% YoY. In-cycle fair price for CopperIngot trending up — if you were pricing against real markets, ask ~Y currency per unit.

  4. "Y currency per unit" uses a calibration: at first-ever call for a given cycle, record the ratio of in-game median price (from your economy data, see task 2) vs the current FRED spot. Multiply by today's real price for subsequent calls. Store calibration in ~/.cache/eco-mcp-app/price-calib.json keyed by cycle id.

Caching

  • FRED observations: SQLite at ~/.cache/eco-mcp-app/fred.sqlite, 6-hour TTL.
  • Calibration: flat JSON, per-cycle — set-once per cycle.

Out of scope

  • Any in-game enforcement. This is advisory flavor only; players opt in by calling the tool.

Acceptance

  • Tool returns a narrative for ["Copper", "Wheat", "Board", "Iron", "Oil"] with non-null percentages.
  • Daily vs monthly series render distinct cadence labels in the narrative (don't call it "7-day %" for a monthly series).
  • Second call for any item hits the SQLite cache.
  • inv smoke passes.
_Originally filed by @coilysiren on 2026-05-02T23:10:12Z - [https://github.com/coilysiren/eco-mcp-app/issues/17](https://github.com/coilysiren/eco-mcp-app/issues/17)_ # Task 8 — Fair-Price Advisor **Prereq**: read `todo/README.md` first. ## Goal Build an MCP tool `fair_price(item)` that returns a short narrative card using real-world commodity data. ## Data sources - **SSM** `/eco-mcp-app/fred-api-key` (SecureString, **region `us-east-1`**) — already provisioned. Fetch once at app start. - **FRED API** `GET https://api.stlouisfed.org/fred/series/observations?series_id=X&api_key=KEY&file_type=json` ### Internal mapping: eco item → FRED series | Eco item | FRED series | Cadence | |---|---|---| | CopperIngot | PCOPPUSDM | **monthly** | | Wheat | PWHEAMTUSDM | **monthly** | | Board (lumber) | WPU0811 | monthly | | IronIngot | PIORECRUSDM | monthly | | Oil | DCOILWTICO | **daily** | ### Cadence matters — don't compute "7-day %" for monthly series Most of the cheap FRED series for commodities publish **monthly**. Computing a "7-day percent change" against a monthly series is meaningless — you'll either get zero (same observation repeating) or garbage when the observation flips. Branch the math on series frequency: ```python def latest_pct_changes(observations, frequency): # observations: list of {date, value}, newest last if frequency == "daily": return { "7d": pct(obs, lookback_days=7), "30d": pct(obs, lookback_days=30), "90d": pct(obs, lookback_days=90), } elif frequency == "monthly": return { "1m": pct(obs, lookback=1), # latest vs prior observation "3m": pct(obs, lookback=3), "12m": pct(obs, lookback=12), } ``` Read the `frequency` from FRED's `/series?series_id=X` response (`frequency_short`: "D", "M", etc.). ## On each call 1. Fetch ~90 days (daily series) or ~13 observations (monthly) of the mapped series from FRED. 2. Compute percent changes using the branching rule above. 3. Return a narrative card: > Real copper: $4.12/lb (monthly series, latest Mar 2026). **Up 3.2% vs prior month, 12.1% YoY.** In-cycle fair price for CopperIngot trending up — if you were pricing against real markets, ask ~Y currency per unit. 4. "Y currency per unit" uses a **calibration**: at first-ever call for a given cycle, record the ratio of in-game median price (from your economy data, see task 2) vs the current FRED spot. Multiply by today's real price for subsequent calls. Store calibration in `~/.cache/eco-mcp-app/price-calib.json` keyed by cycle id. ## Caching - FRED observations: SQLite at `~/.cache/eco-mcp-app/fred.sqlite`, 6-hour TTL. - Calibration: flat JSON, per-cycle — set-once per cycle. ## Out of scope - Any in-game enforcement. This is **advisory flavor only**; players opt in by calling the tool. ## Acceptance - Tool returns a narrative for `["Copper", "Wheat", "Board", "Iron", "Oil"]` with non-null percentages. - Daily vs monthly series render distinct cadence labels in the narrative (don't call it "7-day %" for a monthly series). - Second call for any item hits the SQLite cache. - `inv smoke` passes.
Sign in to join this conversation.
No Branch/Tag specified
Branches Tags
main
claude/climate-parallel-arrays
No results found.
Labels
Clear labels
  
P0
now / blocking / urgent

  
P1
next / important

  
P2
backlog / will do

  
P3
someday / low

  
P4
icebox / frozen / reopen if it becomes real

No labels
P0
P1
P2
P3
P4
Milestone
Clear milestone
No items
No milestone
Projects
Clear projects
No items
No project
Assignees
Clear assignees
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
coilyco-flight-deck/eco-mcp-app#8
No description provided.