account-info

Files (1)

• SKILL.md

SKILL.md (first lines)

# Account Info (Read-Only)

Use the **Omnimatch MCP** codemode tool to read the current user's account data ([Cloudflare Codemode](https://developers.cloudflare.com/agents/api-reference/codemode/)). The server exposes only **codemode**: call it with a `code` parameter (async arrow function body) that runs `codemode.list_my_markets()`, `codemode.list_inbox_matches({ ... })`, etc., and returns the result. This skill is **read-only**: list and view only. For updating profiles, filters, plan changes, or match actions, use the **manage-account** skill.

## When to use

- User asks **what markets am I in?** → `codemode.list_my_markets()`.
- User asks **what entities do I have?** → `codemode.list_my_entities()` (across markets) or `codemode.list_market_entities({ marketId })` for one market.
- User asks **show me details for entity X** → `codemode.get_entity({ entityId })` (returns title, type, externalLink, description, criteria, files with summaries, inbox filters, entity query answers if plan allows).
- User asks **what are my matches** or **show matches** → first `codemode.list_inboxes({ entityId })` to see inboxes (priority, liked, all, plus each inbox filter), then `codemode.list_inbox_matches({ entityId, inboxKey })` to load the table for the chosen inbox.
- User asks about **a specific match** → `codemode.get_match({ matchId, entityId })` returns a concise summary (names, contact, suitability summaries, insight answers, overall scores) that explains why you matched; pass `detail: "full"` when you also need the per-category score breakdown with justifications. The actual reachable contact (LinkedIn URL, email) lives in `otherEntityContact` on the response, not in the `contact` column of the inbox row. Use `codemode.view_match_link({ matchId, entityId })` for just the externalLink.
- User asks **why did we match** for one inbox filter, or **what does filter X say for match Y** → `codemode.list_match_inbox_filter_answers({ matchId, entityId })`.
- User asks about **inbox filters** they have → `codemode.list_inbox_filters({ entityId })`.
- User asks about **subscription**, **plans**, or **what features do I have** → `codemode.list_plans({ marketId })`.
- User asks **how many credits do I have**, **what's my balance**, or **am I out of credits** → `codemode.get_my_credit_balance({ marketId })`. For markets that don't charge credits, the tool returns `creditsEnabled: false`—say so briefly.
- User asks **what credit packs can I buy** or **how do I top up** → `codemode.list_credit_packs({ marketId })`. Present packs by name and credit amount; never expose Stripe price ids in user-facing copy. Direct the user to the **subscription** skill (or `topup_credits`) when they want to actually buy.
- User asks **when is my next search**, **when do new matches arrive**, or **when does discovery run** → `codemode.get_next_match_discovery({ entityId })`. Returns `{ scheduled: false }` or `{ scheduled: true, nextRunAt }` (UTC ISO-8601). This only reports the **automatic** schedule; to **trigger a search now**, that's a manage-account action (`discover_matches`), which adds new matches (it does not re-score existing ones).

**When talking to the user:** Always use **real titles**, never IDs. Refer to entities by their **title**, markets by their **market title**, matches by the other side's **title**, types by the **entity type title** (e.g. "job", "candidate"), and plans by **plan name**. Do not show market ID, entity ID, match ID, or Stripe price ID in user-facing messages—use IDs only inside codemode parameters.

## Codemode read-only tools

Call the **codemode** tool with `code` set to an async arrow function body. Inside it, you can use:

- **codemode.list_my_markets()** → `[{ marketId, title, description }]`.
- **codemode.list_my_entities()** → flat `[{ entityId, marketId, marketTitle, title, type, shortDescription }]`.
- **codemode.list_market_entities({ marketId })** → entities in one market.
- **codemode.get_entity({ entityId })** → full profile.
- **codemode.list_inboxes({ entityId })** → built-in inboxes (priority, liked, all) plus filter inboxes (key `query:{filterId}`), each with a count.
- **codemode.list_inbox_matches({ entityId, inboxKey, offset, limit, sortBy, includeRejected, includeLowScores })** → TUI-friendly table: `{ inboxKey, total, offset, columns, rows: [{ matchId, title, yourScore, theirScore, status, contact, updatedAt, locked }] }`. **`contact` is misleading**: it's a tri-state (`"shared" | "pending" | "none"`) that only reflects owner-to-owner mutual contact sharing via `share_my_contact_with_match`. It stays `"none"` even when the matched entity is externally sourced with a public contact (e.g. a `Liaison` carrying a LinkedIn URL). To get the actual reachable contact, call `get_match` and read `otherEntityContact.externalLink` / `.email` / `.name`.
- **codemode.get_match({ matchId, entityId, detail })** → match detail from your entity's perspective. `detail` defaults to `"summary"` (names, contact, suitability summaries, insight answers, overall scores — small, safe to fetch across many matches); pass `detail: "full"` for the per-category score breakdown with justifications (large, fetch one at a time). Score / insight / contact fields are plan-gated; gated values come back as an upgrade-message string. When a match is freshly created or has just been re-evaluated, `yourCriteriaScore`/`theirCriteriaScore` may be `null` and inbox-filter answers may show `computed: false`; **scoring runs asynchronously and can take up to 5 minutes**, so tell the user it's still computing rather than treating null as an error, and don't aggressively poll.
- **codemode.view_match_link({ matchId, entityId })** → just the externalLink (or `{ externalLink: null, locked: true }`).
- **codemode.list_match_inbox_filter_answers({ matchId, entityId })** → per-filter answers with `computed`, `answer`, `outdated`. `computed: false` means the workflow has not produced an answer yet.
- **codemode.list_inbox_filters({ entityId })** → filter list with prompt, strictness, outputFormat.
- **codemode.list_plans({ marketId })** → plans with `planId`, name, description, isFree, isCurrentPlan, price (live from Stripe), features, limits.
- **codemode.get_my_credit_balance({ marketId })** → `{ creditsEnabled: false, message }` when the market doesn't charge credits, else `{ creditsEnabled: true, subscriptionBalance, topupBalance, nextFreeGrantAt }`. `subscriptionBalance` refreshes monthly from the plan; `topupBalance` is purchased credits that don't expire.
- **codemode.list_credit_packs({ marketId })** → `{ creditsEnabled, packs: [{ priceId, name, creditsAmount, price?: { amount, currency } }] }`. `price.amount` is in the smallest unit of the currency (e.g. cents for USD).
- **codemode.get_next_match_discovery({ entityId })** → `{ scheduled: false }` or `{ scheduled: true, nextRunAt }` (UTC ISO-8601 of the next automatic discovery for this entity).

All ids are UUIDs. Do not invent or assume endpoints.

From omnimatch/omnimatch-mcp, MIT.

More from omnimatch/omnimatch-mcp