Skip to content
BranchesTags

Repository files navigation

GW2AccountMCP

Local, read-only Guild Wars 2 account facts over stateless Streamable HTTP MCP.

Current scope

The server exposes seventeen tools:

  • find_items resolves a bounded English item-name fragment from the generated local public cache. It returns exact matches before contains matches with canonical ID, name, type, rarity, level, and match kind. It does not call the GW2 API at request time or choose among ambiguous names.
  • get_items accepts 1-100 distinct positive canonical item IDs and makes one unauthenticated public /v2/items request at runtime. It returns caller-ordered item facts—nullable name, type, rarity, level, vendor value, flags, game types, and restrictions—with explicit unavailable-resource rows and nullable source facts when a requested public item resource is absent. Optional includeMaterialCategories: true adds one unauthenticated /v2/materials?ids=all request and compact per-item public category memberships sorted by category order then ID. NotRequested and Unavailable use explicit null category data; Available uses non-null lists, where [] means the complete material response reported no membership. Material failure preserves item facts and item completeness. Item asOf and nullable materialCategoriesAsOf are independent response-completion observations, and isAtomicSnapshot: false discloses that the sources are not one atomic snapshot. The tool does not use account data, caches, recipes, or Trading Post prices.
  • get_recipes returns bounded public player-discovered crafting recipe facts in ByIds, InputItem, or OutputItem mode. ByIds accepts 1-100 distinct positive recipe IDs in caller order. Selector modes accept one positive item ID, sort the complete bounded selector result, and apply local offset/limit paging before one definition lookup. Optional includeAccountUnlocks: true requires only account and unlocks, reads one complete bounded /v2/account/recipes list, and adds a boolean accountUnlockListContainsRecipe to every requested or selected ID, including missing public definitions; omitted, null, or false remains key-independent and emits explicit null membership and accountUnlocksAsOf. true means only list presence, while false means only absence from that accepted complete list and never inability to craft. Public selector/definition times and nullable account completion time remain separate, top-level asOf is the final required completion, and isAtomicSnapshot: false discloses the non-atomic join. Results preserve missing definitions, Item versus GuildUpgrade outputs, raw source output IDs, and source-order typed ingredients. Input search indexes Item ingredients only. Output search follows raw source output_item_id; that value can be bogus for guild recipes and must not be treated as their semantic Item output. The tool excludes Mystic Forge, vendors, item enrichment, caches, Trading Post data, prices, costs, and recommendations.
  • get_item_prices accepts 1-100 distinct positive canonical item IDs and returns caller-ordered factual Trading Post price summaries from the immutable local prices.manifest.json generation. It reports nullable English names from the independently published item cache, quote-side aggregate quantities and best unit prices in copper, explicit zero-side and absent-resource semantics, source/cache timestamps, and manual-cache freshness disclosure. It makes no runtime network request, does not return depth or executable volume, and does not recommend an action. Prices may have changed; run tp --fresh to collect a newer source snapshot.
  • value_items accepts 1-100 caller-ordered distinct positive canonical item IDs with quantities from 1 through Int32.MaxValue. From exactly one local price snapshot, it returns nullable best-price factual arithmetic for each row: highest-buy immediate-sale gross, independent 5% listing and 10% exchange fees, and net; lowest-sell buyer cost; and lowest-sell hypothetical-listing gross, fees, and net. Each positive-gross fee is independently rounded half up on that row's aggregate gross and is at least one copper. The three aggregate views return numeric totals only when every requested row has its required quote; otherwise they are explicitly incomplete with null totals and caller-order missing IDs. Best-price values are quote extrapolations without best-level volume, depth, liquidity, or execution guarantees. It does not calculate recipe costs and makes no buy, sell, craft, keep, profit, flip, trend, forecast, velocity, history, or other recommendation.
  • get_account validates the configured key through /v2/tokeninfo, requires account, and returns basic account facts with an asOf timestamp.
  • get_achievement_progress accepts 1-20 caller-ordered distinct positive achievement IDs, requires account and progression for the complete account progress list, then makes one separate optional public English /v2/achievements explicit-ID lookup. It returns requested rows only, explicit account-absence and public-definition states/nulls, completed account bit indexes with compact resolved public bit details where available, independent source completion times, bounded warnings, and isAtomicSnapshot: false. Public definition failure retains account facts as PublicDefinitionsUnavailable; it does not scan the achievement catalog or return categories, groups, tiers, rewards, or recommendations.
  • get_mastery_progress takes no arguments, requires account and progression, and sequentially reads complete bounded account mastery tracks and mastery-point totals. When account tracks are nonempty, it makes one optional public English /v2/masteries lookup for those ascending explicit IDs. It returns canonical track and ordinal-region point-total order, explicit metadata states/nulls, separate source completion observations, bounded warnings, and isAtomicSnapshot: false. sourceLevel uses this tool's product-owned zero-based highest-trained-level interpretation; omitted account rows or levels do not infer unstarted state. Public metadata failure preserves authenticated facts. It excludes raw unlocked point IDs, public catalog rows, point labels, full level catalogs, region mappings, recommendations, and planning.
  • get_character_build accepts one exact character name from get_characters, requires account, characters, and builds, verifies that name against the complete roster, and returns only that character's active build tab. It preserves fixed specialization, trait, terrestrial/aquatic skill, Ranger pet, and Revenant legend slots; resolves referenced public metadata to compact names where available; and retains unresolved IDs with deterministic warnings and isMetadataComplete: false. It does not return inventory, equipment, inactive tabs, or ownership quantities.
  • get_character_equipment accepts one exact character name from get_characters, requires account, characters, builds, and inventories, and returns that character's active PvE/WvW combat-equipment references. It uses the active equipment tab plus a conditional current-equipment lookup for the API's missing Relic, resolves compact item, prefix, upgrade, infusion, and skin metadata, preserves unresolved canonical IDs with warnings, and explicitly marks the result as non-ownership data. It includes terrestrial and aquatic combat slots but excludes PvP, dyes, gathering, fishing, Jade Bot equipment, inventory, inactive tabs, quantities, and Legendary Armory ownership.
  • get_character_equipment_tabs accepts one exact character name from get_characters, requires account, characters, builds, and inventories, and returns every reported PvE/WvW combat equipment tab with an explicit active marker. It reports equipment references rather than ownership, preserves unresolved compact metadata as explicit nulls with warnings, and does not alter the active-equipment, holdings, inventory, or Legendary Armory surfaces.
  • get_character_inventory accepts one exact character name from get_characters, requires account, characters, and inventories, and returns that character's complete bounded physical equipped-bag layout. It preserves zero-based bag and slot positions, absent bags, empty slots, per-slot stack counts and charges, binding, selected/default stats, upgrades, infusions, and skins, with compact public names where available and deterministic metadata warnings otherwise. Its physical stack counts are already represented by get_account_holdings character-bag contributions and must not be added as a second ownership source; it returns no per-item totals and excludes account storage, equipment references, inactive tabs, and Legendary Armory ownership.
  • get_characters requires account and characters, retrieves the complete character list and each character's core record, and returns name-ordered summaries with name, race, gender, profession, level, playtime seconds, creation and last-modified timestamps, and deaths. The result is complete or the whole operation fails; it does not include inventory, equipment, builds, guild, or title data.
  • get_legendary_armory takes no arguments, requires account, inventories, and unlocks, and returns the complete bounded set of entries reported by /v2/account/legendaryarmory, ordered by item ID. armoryCount means the count available for use in one equipment template; it is reusable account Legendary Armory ownership, not a physical stack, onHand quantity, equipped occurrence, or extra physical copy, and must not be added to get_account_holdings. Compact English item names, types, subtypes, and armor weight classes come from /v2/items; unresolved public metadata leaves the authenticated ID/count intact with explicit nulls and deterministic warnings. The tool does not return unowned catalog rows or public Armory capacity limits.
  • get_wallet requires account and wallet, retrieves /v2/account/wallet, and joins each canonical currency ID to its English /v2/currencies name. It returns long values, one asOf timestamp, and warnings that retain the ID and value when metadata is unavailable. An empty wallet is returned as an empty balance list.
  • get_account_holdings accepts separate optional itemIds and currencyIds arrays, requires at least one ID, and permits at most 20 combined positive IDs with no duplicates within either array. It preserves caller order and treats item and currency IDs as separate canonical namespaces, so the same numeric ID may appear once in each array.
  • get_trading_post_activity supports CurrentBuys and CurrentSells and requires account and tradingpost. It returns one complete caller-selected page with canonical item ID, unit price in copper, unfulfilled quantity, and creation time. CurrentBuys adds checked reserved coin per row and a selected-page subtotal; those quantities are pending orders, not owned items. CurrentSells quantities are currently unfulfilled listed assets already represented by get_account_holdings as listedForSale; they are not on-hand and must not be added as another ownership source. page defaults to 0, pageSize defaults to 50, and page size is limited to 1-200. Results preserve source order but do not claim chronological or global ordering or cross-call pagination stability; transaction data is documented as cached for about five minutes, and asOf is response completion rather than a source timestamp. Upstream transaction IDs are not returned. Model-facing Delivery and previous-90-day history modes are deferred pending demonstrated need; get_account_holdings continues to include TP-delivery item quantities in ownership results.

get_account_holdings queries only the sources relevant to its inputs. Currency requests require account and wallet. Item requests use bank, material storage, and shared inventory (account, inventories); every character bag (account, characters, inventories); and Trading Post delivery and current sells (account, tradingpost). English item names are requested only for the supplied item IDs from the public /v2/items endpoint.

Holdings results distinguish onHand, inTradingPostDelivery, listedForSale, and ownedTotal. A successfully exhausted source contributes an authoritative value, including zero. A failed source or absent wallet balance produces a nullable quantity, explicit unavailableLocations or warning evidence, and isComplete: false; a known partial subtotal is never presented as a complete total. Item metadata failure leaves canonical IDs and known quantities intact with a null name and warning.

No tool returns token metadata or the key. The accepted v1 key scopes remain account, wallet, inventories, characters, builds, progression, unlocks, and tradingpost; each enabled tool validates only its required scopes.

Configure and run

Copy no secret into a repository file. Set GW2_API_KEY with either development user-secrets (replace the placeholder locally):

dotnet user-secrets set "GW2_API_KEY" "<your-GW2-key>" --project src/GW2AccountMCP

or an environment variable for the current PowerShell session:

$env:GW2_API_KEY = "<your-GW2-key>"

appsettings.example.json has only non-secret configuration. GW2_API_BASE_URL defaults to https://api.guildwars2.com and may point to a local fake server in tests. GW2_PUBLIC_CACHE_PATH defaults to data/public-cache, and GW2_API_BUDGET_LOCK_PATH defaults to data/gw2-api-budget.lock; both paths are resolved from the repository-root working directory used by the commands below.

Selected-character inventory has these server-owned safety limits:

Setting Default
GW2_CHARACTER_INVENTORY_MAX_BAG_POSITIONS 20
GW2_CHARACTER_INVENTORY_MAX_SLOTS_PER_BAG 40
GW2_CHARACTER_INVENTORY_MAX_TOTAL_SLOTS 640
GW2_CHARACTER_INVENTORY_MAX_ITEM_REFERENCES 1024
GW2_CHARACTER_INVENTORY_MAX_STAT_ATTRIBUTES 2048

The defaults cover the current documented maximum of 16 bags with 32 slots each and leave modest headroom. Raise them only after verifying an ArenaNet capacity change. Values must be positive 32-bit integers; total slots must be at least slots per bag and no greater than bag positions multiplied by slots per bag, and item references must be at least bag positions plus total slots. Changes take effect after a server restart. Higher values can materially increase response size, metadata request count, latency, and model-context use. No hard upper ceiling is imposed because these are trusted server-owner settings; per-stack structural limits remain fixed.

Refresh the public item cache

Stop MCP and close or pause Excel refreshes before running either refresh command. MCP and the updater each cap all GW2 API request attempts, including retries, at four starts per second (240/minute) and use the same exclusive lock file. A lease error means another MCP/updater process is active. The lock file persists after release; its existence is normal.

Start or resume a production snapshot from the repository root:

dotnet run --project tools/GW2AccountMCP.DataRefresh -- items --output data/public-cache

The full public refresh makes hundreds of bounded requests and can take multiple minutes. Validated 200-item batches are committed under the updater-owned .items-staging directory as they complete. If an ordinary network or batch failure occurs, rerun the same command without --fresh; it verifies the current sorted root-ID count and hash, reuses matching staged batches, and downloads only what remains. The incomplete result reports aggregate staged progress and safe timeout, transport, HTTP, and invalid-response counts without item IDs or response contents.

The updater validates every source ID but excludes blank or whitespace-only names from the searchable CSV. It publishes one immutable items.<sha256>.csv generation and atomically updates items.manifest.json only after every batch is validated; manifest rowCount is the number of published named rows. A failed or incomplete refresh leaves the prior referenced generation intact. MCP never refreshes this cache automatically.

After publication, the validated shards and exact publishing state remain in .items-staging as a bounded repair source. Running the plain command again validates or reconstructs the exact generation and manifest without GW2 API calls. Use --fresh to discard that retained snapshot and deliberately download a new one:

dotnet run --project tools/GW2AccountMCP.DataRefresh -- items --output data/public-cache --fresh

Use --fresh after a known Guild Wars 2 patch, a deliberate catalog update, or incompatible staged data even when the root item count and ID hash appear unchanged, because definitions for existing IDs can change without changing that root list. --fresh removes only the retained staging after a complete ownership preflight; it never removes the currently published manifest or generation before the replacement is ready. It refuses to delete anything if .items-staging contains an unrecognized entry; resolve that entry before retrying.

Use the isolated one-page command for a fast updater-to-reader test:

dotnet run --project tools/GW2AccountMCP.DataRefresh -- items-test --output data/public-cache-test
$env:GW2_PUBLIC_CACHE_PATH = "data/public-cache-test"

items-test validates the complete root catalog but downloads only the first 200 sorted item definitions in memory, excludes blank names, and publishes no resumable staging. Its normalized output directory name must end in -test; it cannot target the production directory. The standard manifest/CSV pair supports a real MCP round trip, but it is deliberately incomplete and must not be used as the production cache. Clear the temporary environment override before a production launch:

Remove-Item Env:GW2_PUBLIC_CACHE_PATH

Production and test cache directories, retained staging, lock files, and the local workbook are ignored by Git.

The index loads lazily on the first find_items call. It checks for a changed cache only after a no-match, then reloads once and reruns that search. find_items, get_item_prices, and value_items use only local published caches at runtime; get_items is the separate public request-time catalog lookup.

Refresh the public price cache

Publish or resume the independent production price cache from the repository root:

dotnet run --project tools/GW2AccountMCP.DataRefresh -- tp --output data/public-cache

The command publishes an immutable prices.<sha256>.csv and manifest-last prices.manifest.json. A normal rerun resumes retained .prices-staging work or repairs a completed publication without re-fetching it. Use --fresh only to discard recognized price staging and collect a newer source snapshot; it never deletes the published generation:

dotnet run --project tools/GW2AccountMCP.DataRefresh -- tp --output data/public-cache --fresh

For an isolated structural test cache only, use an output directory ending in -test:

dotnet run --project tools/GW2AccountMCP.DataRefresh -- tp-test --output data/public-cache-test

tp-test deliberately publishes only the first bounded page and is not a production price cache.

Excel GetItemNames query

Replace the existing Excel Power Query named GetItemNames with the following query. Adjust CacheDirectory only if the production cache is elsewhere. This reads the manifest's committed generation and preserves the downstream Id and Name columns.

let
    CacheDirectory = "D:\Code\GW2AccountMCP\data\public-cache",
    Manifest = Json.Document(File.Contents(CacheDirectory & "\items.manifest.json")),
    CsvFileName = Text.From(Record.Field(Manifest, "csvFileName")),
    Source = Csv.Document(
        File.Contents(CacheDirectory & "\" & CsvFileName),
        [Delimiter = ",", Columns = 5, Encoding = 65001, QuoteStyle = QuoteStyle.Csv]
    ),
    PromotedHeaders = Table.PromoteHeaders(Source, [PromoteAllScalars = true]),
    TypedRows = Table.TransformColumnTypes(
        PromotedHeaders,
        {{"id", Int64.Type}, {"name", type text}, {"type", type text}, {"rarity", type text}, {"level", Int64.Type}}
    ),
    SelectedColumns = Table.SelectColumns(TypedRows, {"id", "name"}),
    RenamedColumns = Table.RenameColumns(SelectedColumns, {{"id", "Id"}, {"name", "Name"}})
in
    RenamedColumns

Excel GetListings query

Replace the existing Excel Power Query named GetListings with the following query. Adjust CacheDirectory only if the production cache is elsewhere. This reads the price manifest's committed generation and preserves the downstream id, Highest Buy Price, Demand, Lowest Sell Price, and Supply columns. GetListings no longer invokes the paged Listings function or NumListings.

let
    CacheDirectory = "D:\Code\GW2AccountMCP\data\public-cache",
    Manifest = Json.Document(File.Contents(CacheDirectory & "\prices.manifest.json")),
    CsvFileName = Text.From(Record.Field(Manifest, "csvFileName")),
    Source = Csv.Document(
        File.Contents(CacheDirectory & "\" & CsvFileName),
        [Delimiter = ",", Columns = 6, Encoding = 65001, QuoteStyle = QuoteStyle.Csv]
    ),
    PromotedHeaders = Table.PromoteHeaders(Source, [PromoteAllScalars = true]),
    SelectedColumns = Table.SelectColumns(
        PromotedHeaders,
        {"id", "buyUnitPrice", "buyQuantity", "sellUnitPrice", "sellQuantity"}
    ),
    TypedRows = Table.TransformColumnTypes(
        SelectedColumns,
        {
            {"id", Int64.Type},
            {"buyUnitPrice", Int64.Type},
            {"buyQuantity", Int64.Type},
            {"sellUnitPrice", Int64.Type},
            {"sellQuantity", Int64.Type}
        }
    ),
    RenamedColumns = Table.RenameColumns(
        TypedRows,
        {
            {"buyUnitPrice", "Highest Buy Price"},
            {"buyQuantity", "Demand"},
            {"sellUnitPrice", "Lowest Sell Price"},
            {"sellQuantity", "Supply"}
        }
    )
in
    RenamedColumns

After a successful production cache refresh, use Excel Refresh All and save the workbook. The cache updater and Excel read the generation selected by the manifest; the updater does not automate or modify the workbook. Any remaining direct GW2 API queries in the workbook must not overlap MCP, the updater, or another bulk GW2 client.

From the repository root:

dotnet restore
dotnet run --project src/GW2AccountMCP

The documented default listener is http://127.0.0.1:5288; the stateless MCP endpoint is http://127.0.0.1:5288/mcp.

After the gw2-account tunnel profile is configured with a persistent file-backed runtime key, launch the server and tunnel together from the repository root:

.\start.ps1

The script builds the project, starts both processes, waits for MCP and tunnel readiness, opens the tunnel UI, and stops both owned process trees when you press Ctrl+C. It does not read or store either API key.

If either default port is occupied, choose alternate loopback ports for that launch:

.\start.ps1 -McpPort 6288 -TunnelHealthPort 9080

-McpPort controls the MCP server and tunnel target. -TunnelHealthPort controls the tunnel health endpoint and admin UI. The ports must differ.

Persistent runtime-key setup is documented in D:\Obsidian\Code\GW2AccountMCP\GW2 Account MCP - Persistent Tunnel Key Setup.md. The runtime key remains outside the repository; do not put it in an environment variable for routine launches.

Test

dotnet build --no-restore
dotnet test --no-build

Tests use in-process fake HTTP responses and do not call Guild Wars 2.

Local MCP Inspector

With the server running, use the current official Inspector CLI flow to list tools:

npx @modelcontextprotocol/inspector --cli http://127.0.0.1:5288/mcp --transport http --method tools/list

Then invoke a tool through Inspector. Account-backed tools need a locally configured valid GW2 key with the required scopes; do not paste it into Inspector arguments or chat. find_items, get_item_prices, and value_items use only local public caches; get_items and default get_recipes calls are public and request-time, while get_recipes with includeAccountUnlocks: true additionally requires account and unlocks. get_achievement_progress requires account and progression and accepts canonical achievement IDs. get_mastery_progress takes no arguments and requires account and progression. get_trading_post_activity accepts CurrentBuys or CurrentSells and requires account and tradingpost. Holdings still accepts canonical IDs, so resolve names first when needed.

npx @modelcontextprotocol/inspector --cli http://127.0.0.1:5288/mcp --transport http --method tools/call --tool-name find_items --tool-args-json '{"query":"Mystic Coin"}' --format json
npx @modelcontextprotocol/inspector --cli http://127.0.0.1:5288/mcp --transport http --method tools/call --tool-name get_items --tool-args-json '{"itemIds":[19976]}' --format json
npx @modelcontextprotocol/inspector --cli http://127.0.0.1:5288/mcp --transport http --method tools/call --tool-name get_recipes --tool-args-json '{"mode":"InputItem","itemId":19976}' --format json
npx @modelcontextprotocol/inspector --cli http://127.0.0.1:5288/mcp --transport http --method tools/call --tool-name get_item_prices --tool-args-json '{"itemIds":[19976]}' --format json
npx @modelcontextprotocol/inspector --cli http://127.0.0.1:5288/mcp --transport http --method tools/call --tool-name value_items --tool-args-json '{"items":[{"itemId":19976,"quantity":3}]}' --format json
npx @modelcontextprotocol/inspector --cli http://127.0.0.1:5288/mcp --transport http --method tools/call --tool-name get_account --format json
npx @modelcontextprotocol/inspector --cli http://127.0.0.1:5288/mcp --transport http --method tools/call --tool-name get_achievement_progress --tool-args-json '{"achievementIds":[1,2]}' --format json
npx @modelcontextprotocol/inspector --cli http://127.0.0.1:5288/mcp --transport http --method tools/call --tool-name get_mastery_progress --format json
npx @modelcontextprotocol/inspector --cli http://127.0.0.1:5288/mcp --transport http --method tools/call --tool-name get_character_build --tool-args-json '{"characterName":"<exact name returned by get_characters>"}' --format json
npx @modelcontextprotocol/inspector --cli http://127.0.0.1:5288/mcp --transport http --method tools/call --tool-name get_character_equipment --tool-args-json '{"characterName":"<exact name returned by get_characters>"}' --format json
npx @modelcontextprotocol/inspector --cli http://127.0.0.1:5288/mcp --transport http --method tools/call --tool-name get_character_equipment_tabs --tool-args-json '{"characterName":"<exact name returned by get_characters>"}' --format json
npx @modelcontextprotocol/inspector --cli http://127.0.0.1:5288/mcp --transport http --method tools/call --tool-name get_character_inventory --tool-args-json '{"characterName":"<exact name returned by get_characters>"}' --format json
npx @modelcontextprotocol/inspector --cli http://127.0.0.1:5288/mcp --transport http --method tools/call --tool-name get_characters --format json
npx @modelcontextprotocol/inspector --cli http://127.0.0.1:5288/mcp --transport http --method tools/call --tool-name get_legendary_armory --format json
npx @modelcontextprotocol/inspector --cli http://127.0.0.1:5288/mcp --transport http --method tools/call --tool-name get_wallet --format json
npx @modelcontextprotocol/inspector --cli http://127.0.0.1:5288/mcp --transport http --method tools/call --tool-name get_account_holdings --tool-args-json '{"itemIds":[101,202],"currencyIds":[3]}' --format json
npx @modelcontextprotocol/inspector --cli http://127.0.0.1:5288/mcp --transport http --method tools/call --tool-name get_trading_post_activity --tool-args-json '{"mode":"CurrentBuys","page":0,"pageSize":50}' --format json
npx @modelcontextprotocol/inspector --cli http://127.0.0.1:5288/mcp --transport http --method tools/call --tool-name get_trading_post_activity --tool-args-json '{"mode":"CurrentSells","page":0,"pageSize":50}' --format json

Secure MCP Tunnel and ChatGPT smoke test

The holdings IDs above are placeholders only. Confirm the current Inspector argument syntax with npx @modelcontextprotocol/inspector --help if the installed CLI changes.

Create and associate the tunnel in the OpenAI Platform/ChatGPT UI. Do not create tunnel resources until local checks pass. Download tunnel-client from the current official latest release; do not pin a stale client version. Start with tunnel-client help quickstart, then initialize a named profile using the tunnel ID shown by Platform:

tunnel-client init --sample sample_mcp_remote_no_auth --profile gw2-account --tunnel-id <tunnel-id> --mcp-server-url http://127.0.0.1:5288/mcp

Configure the profile's reusable runtime key using the persistent-key guide, then run .\start.ps1. In ChatGPT web Developer Mode, create a read-only draft app using the tunnel connection, verify that exactly sixteen tools are discovered: find_items, get_items, get_recipes, get_item_prices, value_items, get_account, get_account_holdings, get_achievement_progress, get_mastery_progress, get_trading_post_activity, get_character_build, get_character_equipment, get_character_inventory, get_characters, get_legendary_armory, and get_wallet. Keep the launcher running while using the app.

About

No description, website, or topics provided.

Resources

Readme
MIT license
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

Packages

Contributors

Languages