Version Packages

[github-actions]

This PR was opened by the Changesets release GitHub action. When you're ready to do a release, you can merge this and the packages will be published to npm automatically. If you're not ready to do a release yet, that's fine, whenever you add more changesets to main, this PR will be updated.

Releases

@iqai/mcp-debank@2.2.0

Minor Changes

• 304039f: Fetch whale NFT holdings by fanning out per-chain instead of the slow server-side aggregate.

  getUserAllNftList previously called DeBank's /user/all_nft_list, which aggregates across every chain server-side and takes 10–20 s for active wallets — routinely timing out to nothing for whales (e.g. vitalik.eth). It now discovers the wallet's used chains (used_chain_list) and fans out the fast per-chain /user/nft_list calls in parallel with bounded concurrency (8), so wall-clock is ≈ (chains / 8) × ~1.5 s.

  The method also degrades gracefully instead of failing all-or-nothing: a chain that errors or is too slow is dropped and reported, and the method enforces its own soft deadline (~75% of the call's timeout budget) so it returns the chains that landed before the client's hard abort fires — partial-on-timeout, mirroring the result-size-budget philosophy.

  Return-shape change (guest-facing): getUserAllNftList now resolves to { nfts, chains, partial, chains_skipped } instead of a bare UserNFT[]. Read result.nfts for the holdings (each NFT is stamped with its chain); partial: true with chains_skipped names the chains to retry (narrow via chain_ids, or use getUserNftList per chain). The tool description, cookbook, and instructions are updated accordingly.

Patch Changes

• ca6d059: Bound the execute result size so whale-wallet queries can't overflow the consuming model's context window.

  The execute tool serialized the guest run() return value with no size limit and returned it verbatim to the MCP client. For large wallets (e.g. vitalik.eth), whole-portfolio queries — "all token holdings across all chains", "all NFT holdings", "all approvals on Ethereum" — produced payloads of ~500k tokens, which overflowed the consuming agent's context window before the summarizing model call could even run.

  There is now a generic size budget enforced at the result boundary (DEBANK_MCP_EXECUTE_RESULT_BUDGET_CHARS, default 200,000 chars ≈ 50k tokens). When a result exceeds it, the envelope metadata (ok / error / log_lines / err_lines) is kept intact and the bulk result is truncated — arrays keep the first N elements, objects trim their largest arrays, strings keep a prefix — with a structured _truncated: { shown, total, reason, hint } signal attached so the consumer can answer "showing top N of M — refine to see more" instead of crashing. Results under budget are returned byte-for-byte unchanged.