chore(skill): refresh solana-add-idl for post-#228 VisualizerContext API

[shahan-khatchadourian-anchorage]

Summary

Stacked on top of #255. Refreshes the solana-add-idl scaffolding skill so newly-generated presets compile against the wire-data VisualizerContext introduced in #255.

Changes

• Replaces the broken squads_multisig template reference (preset doesn't exist in this repo) with dflow_aggregator, a real working example of the post-feat: attested transaction lint diagnostics in SignablePayload #228 pattern.
• Adds explicit guidance for the new accessor pattern at the top of visualize_tx_commands: context.resolve_program_id()? / context.resolve_accounts()? / context.data(). These replace the removed context.current_instruction() and surface unresolved indices with precise error messages.
• Updates the "Required imports" list to match real preset code (BTreeMap for deterministic named-accounts ordering, AccountMeta, OnceLock).
• Drops redundant Some(hex::encode(...)) second arg in create_raw_data_field calls — the helper falls back to the same lowercase hex by default.
• Verify step exercises both feature configs (diagnostics on/off) so newly scaffolded presets work for parser_cli and parser_app alike.

Test plan

• cargo fmt --check, make lint clean
• Scaffold a fresh preset using the updated skill on a sample IDL and verify it compiles + tests cleanly in both feature configs

🤖 Generated with Claude Code

[Copilot]

Pull request overview

Updates the solana-add-idl Claude skill documentation so newly scaffolded Solana IDL-based presets match the post-#228/#255 VisualizerContext wire-data API and compile under both diagnostics feature configurations.

Changes:

• Switches the recommended template preset from the nonexistent squads_multisig to dflow_aggregator and adds guidance for the wire-data accessor pattern (resolve_program_id / resolve_accounts / data).
• Refreshes the “required imports” list to align with current preset implementations (e.g., BTreeMap, AccountMeta, OnceLock) and documents deterministic named-account ordering.
• Expands the verify commands to run clippy/tests with diagnostics both enabled and disabled.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[prasanna-anchorage]

Minor accuracy nit: of the five helpers listed here, dflow_aggregator (the template you point at on line 89) actually defines only build_named_accounts / build_parsed_fields / build_fallback_fields. append_raw_data and format_arg_value live in other IDL presets — e.g. chain_parsers/visualsign-solana/src/presets/kamino_vault/mod.rs:209,217 or jupiter_earn/mod.rs:218,229. Two ways to fix, either is fine:

(a) Trim the helper list to the three that dflow_aggregator actually has, and let users grow their preset to need append_raw_data / format_arg_value only when their IDL has byte-blob args, or
(b) Keep all five but add "(for append_raw_data / format_arg_value, see e.g. kamino_vault)" so the first-time user knows where to copy from.

Not a blocker — newcomers can grep, but the current wording suggests dflow_aggregator alone is a sufficient template, and it is not for any preset that needs those two helpers.

[prasanna-anchorage]

Approved.

Reviewed in a worktree at d9b8f05. Verified the template's structural claims against main:

• chain_parsers/visualsign-solana/src/presets/dflow_aggregator/ exists with the imports list this skill prescribes byte-for-byte (BTreeMap, OnceLock, AccountMeta, etc. — mod.rs:1-21).
• VisualizerContext::resolve_program_id, resolve_accounts, data, instruction_index are all real methods on core/mod.rs (151, 163, 130, 104).
• build_named_accounts / build_parsed_fields / build_fallback_fields exist in dflow_aggregator/mod.rs:118,151,197.
• The "raw-data-field rule" added in d9b8f05 matches what landed in core/mod.rs after #228 (the helper's None branch already does lowercase byte-by-byte hex).
• The Step 6 verify block now exercises both feature configs, which is the right shape since parser_app builds without diagnostics and parser_cli builds with it.

One non-blocking inline nit on line 99 about the helper list (append_raw_data / format_arg_value live in other presets, not in dflow_aggregator). The PR's own "scaffold a fresh preset using the updated skill on a sample IDL" test-plan checkbox is still pending — worth doing once before merge, but the static review of the template against main is clean.