Emem Locate And Recall

When to use this skill

Use Emem Locate And Recall when an AI agent needs a reusable SKILL.md workflow for this job: Resolve a free-form place name to an emem cell64 and recall signed Earth-observation facts at that location. Use when the user asks about current weather, vegetation index, elevation, soil properties, or any other geospatial measurement at a named place ("what's the temperature in Bengaluru", "how high is Denali", "what's the NDVI in the Sundarbans"). Returns content-addressed facts with Ed25519 receipts that can be verified offline. Reads are public — no auth required.

When not to use it

Skip Emem Locate And Recall when the task is outside the creativity category, or when a more specific skill in this directory already covers the same workflow with clearer triggers.

How to install

1. Personal install: create ~/.claude/skills/emem-locate-and-recall/SKILL.md (and any bundled scripts) so Claude Code, Claude Desktop, and compatible agents can load it in every project.
2. Project install: commit the same folder at .claude/skills/emem-locate-and-recall/ so teammates get the skill with the repo.
3. Restart the agent session after copying files so it re-scans the skills directory, then ask for the task in words that match the skill description.

Full install guide for Claude, Cursor, and Codex

What this skill does

# emem-locate-and-recall

This skill turns a place name into signed Earth-observation facts in two REST calls.

## When to invoke

The user asks about a measurable geospatial fact at a named place:

- "What's the current 2 m air temperature in Bengaluru?" - "Show me the NDVI in the Sundarbans." - "How does the soil pH look in central Iowa?" - "What's the elevation of Mount Kilimanjaro?"

If the user has a `cell64` already (a string like `defi.zb493.xoso.zcb6a`), skip the locate step and go straight to `/v1/recall`. If they want a *thumbnail* (visual preview) instead of numeric facts, use `GET /v1/cells/{cell64}/scene.png` directly.

## How to invoke

The endpoint is `https://emem.dev`. Reads are public.

### Step 1 — resolve the place name

```sh curl -sf -X POST https://emem.dev/v1/locate \ -H 'content-type: application/json' \ -d '{"q":"Bengaluru, India"}' | jq '.cell64, .place_label, .via' ```

`via` reports how the name was resolved (`embedded`, `cache`, `photon`, `nominatim`, `fallback`). Treat `via=fallback` as low-confidence and ask the user to clarify.

### Step 2 — recall the band(s)

```sh curl -sf -X POST https://emem.dev/v1/recall \ -H 'content-type: application/json' \ -d '{"cell":"<CELL64_FROM_STEP_1>", "bands":["weather.temperature_2m","indices.ndvi"]}' \ | jq '.facts[] | {band, value, unit, signed_at}, .receipt.fact_cids[]' ```

The response carries `.facts[]` (with `value`, `unit`, `signed_at`, `signer_pubkey_b32`) and `.receipt` (Ed25519 signature, fact_cids, schema_cid, registry_cid). The `fact_cids[i]` is a durable handle — re-fetching that CID from any responder in any year returns the same bytes.

## Picking bands

The full band list is at `https://emem.dev/v1/bands` (124 auto-materializable bands, 1792 total dims). Common picks:

- **Weather**: `weather.temperature_2m`, `weather.precipitation_mm`, `weather.relative_humidity_2m`, `weather.wind_speed_10m` - **Air quality**: `cams.pm25`, `cams.no2`, `cams.aod_550` - **Climate**: `era5.t2m`, `era5.precip` (1940→present hourly) - **Vegetation**: `indices.ndvi`, `indices.evi`, `modis.ndvi_mean` - **Elevation**: `copdem30m.elevation_mean`, `gmrt.topobathy_mean` - **Land cover**: `esa_worldcover.lc_2021` - **Embeddings**: `geotessera` (128-D, default 2024 vintage); the responder also ships `geotessera.{2017..2024}` for per-year and `geotessera.multi_year` (1024-D = 8×128 stacked).

If unsure which band fits the user's question, call `/v1/bands` and filter by family (`weather`, `vegetation`, `terrain`, `landcover`, `climate`).

## Recovering from errors

- **HTTP 404 `cid_not_found`** — band has no fact at this cell on this responder. With `EMEM_AUTO_MATERIALIZE=on` (default at emem.dev) the recall already attempted to materialise; the band has no upstream connector wired. Use `/v1/data_availability` to check which bands have a materialiser. - **HTTP 400 `unknown cell64 symbol`** — typo in the cell64. Re-run step 1 with the place name. - **HTTP 422** — request body malformed; the responder's structured error message names the missing field.

## Verifying offline

Pass the returned receipt to the `emem-verify-receipt` skill (or to `POST /v1/verify_receipt` server-side) to confirm the signature without re-trusting the responder.

## Worked example

``` USER: What's the current temperature in Bengaluru, and the NDVI?

CLAUDE invokes this skill: POST /v1/locate {"q":"Bengaluru, India"} → cell64=defi.zb493.xoso.zcb6a POST /v1/recall {"cell":"defi.zb493.xoso.zcb6a", "bands":["weather.temperature_2m","indices.ndvi"]} → facts: weather.temperature_2m = 28.4 degC indices.ndvi = 0.42 → fact_cids: [qi3jo4..., bn7c4d...]

CLAUDE replies: "Bengaluru is at 28.4 °C right now (signed by emem.dev at 2026-05-08T13:22Z). The vegetation index there is 0.42 — middling, consistent with a dry urban core. Both facts are content-addressed: qi3jo4...l2hgjtwm and bn7c4d...kkb73a respectively." ```

Intended uses

• Use Emem Locate And Recall when this documented workflow matches the task.