SVM alpha runtime: instructions, typed handlers, any_of filters

[JasoonS]

Summary

Stacks the SVM alpha work onto main in a single PR. Most of this has been brewing on the svm-alpha-bugfixes integration branch; landing it here so the SVM surface has a coherent base.

Headline pieces:

• SVM runtime (Stages 4-6): HyperSyncSolanaSource driver, instruction event router + per-program discriminator probing, onInstruction dispatch plumbing, envio init SVM Metaplex template, TUI slot labels, end-to-end test.
• Stage 7b: decoded Borsh args + named accounts surfaced on handler events; codegen emits typed handler signatures per instruction.
• account_filters any_of (newest commit): second YAML shape for filters expressing OR across positions. Outer = OR of AND-groups, inner = the existing flat shape. Maps 1:1 to HyperSync's array<instructionSelection> so no wire-level change is needed. Validation tightened: positions are 0..=5 (6..=9 reserved), duplicate positions in a single group now hard-error.
• CI/correctness fixes: clippy type-complexity + workspace allowBuilds cleanup, pnpm-lock.yaml regen on pnpm 10, missing svmInstructionEventConfig fields backfilled in test fixtures, 3 RpcSource tests skipped (hit external eth.rpc.hypersync.xyz), Hasura metadata reload race fixed.

any_of example:

account_filters:
  any_of:
    - - position: 1
        values: [pkA]
    - - position: 3
        values: [pkB]

EventRouter, dispatch, and existing flat-shape configs are unchanged.

Test plan

• CI green
• Run svm_metaplex_demo scenario end-to-end against devnet/mainnet and verify event throughput
• Smoke a config using any_of with at least two AND-groups and confirm instructionSelection count on the wire matches branch count
• Smoke a config using the legacy flat account_filters shape and confirm behavior is unchanged
• Validation negative cases: position: 6, duplicate position within a group, empty any_of, empty any_of branch

🤖 Generated with Claude Code

Summary by CodeRabbit

Release Notes

• New Features

  • Added Solana instruction event indexing with automatic Borsh schema decoding
  • New Metaplex Token Metadata starter template
  • Integrated HyperSync Solana endpoint support for efficient data fetching
  • Instruction filtering capabilities (by discriminator and account positions)
  • Enhanced Hasura metadata reloading workflow

• Bug Fixes

  • Simulator now respects startBlock/endBlock configuration overrides

• Chores

  • Updated build dependencies for Solana support

[coderabbitai]

📝 Walkthrough

Walkthrough

Introduces full Solana (SVM) instruction indexing: YAML schema/types and validation, public/system config translation, HyperSync Solana client and Borsh decoder, runtime routing and handler API, codegen/typings, templates, and demo/tests. Also updates EVM/Fuel schemas and HyperSync EVM addon interfaces.

Changes

SVM instruction indexing and codegen

Layer / File(s)	Summary
Design notes, CLI deps, template options STAGE_7B_DECISIONS.md, packages/cli/Cargo.toml, packages/cli/CommandLineHelp.md, packages/cli/src/cli_args/init_config.rs, packages/cli/src/template_dirs.rs, packages/cli/templates/dynamic/init_templates/shared/package.json.hbs, packages/cli/templates/static/pnpm-workspace.yaml	Adds Stage 7b/7c decisions, CLI dependencies, template enum/help and dir mapping, and pnpm config.
CLI human/public/system config and validation packages/cli/src/config_parsing/*.rs	Adds SVM human-config (programs/instructions/args/filters), validation, public JSON (svm programs/abi), and system translation (Abi::Svm/EventKind::Svm/DataSource::Svm).
JSON schema updates (EVM/Fuel/SVM) packages/envio/*.schema.json	Renames EVM/Fuel contract $defs and introduces SVM schema for programs, hypersync_config, instruction args/accounts filters.
HyperSync EVM N-API addon and types packages/cli/src/hypersync_source/*	Adds EVM HyperSync client/decoder N-API, query/type conversions, and mapping helpers.
HyperSync Solana (SVM) N-API addon and types packages/cli/src/hypersync_source_svm/*, packages/cli/src/lib.rs	Adds Solana client/query/types and Borsh decoder bridge; wires modules in CLI crate.
Codegen changes, TS typings, and templates packages/cli/src/hbs_templating/*, packages/envio/index.d.ts, packages/envio/package.json, packages/cli/templates/static/svm_metaplex_template/**/*	Generates SVM instruction modules and config.svm.programs, adds TS indexer typings and starter template files.
Envio runtime wiring, sources, and TUI packages/envio/src/**/*	Adds SVM onInstruction surface, event config/builder/router, HyperSync Solana source with decoding, config parsing with hypersync URL, RPC fallback, UI “Slots” labeling, and client bindings updates.
Metaplex scenario, configs, and tests scenarios/**/*, packages/cli/test/configs/svm-metaplex-config.yaml	Adds Metaplex demo (config, handlers, schema), bounded test config, E2E live test, and SVM router/client tests.

Sequence Diagram(s)

sequenceDiagram
  participant YAML as Config (YAML)
  participant CLI as SystemConfig (CLI)
  participant Envio as Envio Runtime
  participant HSSrc as HyperSyncSolanaSource
  participant HS as HyperSyncSolanaClient
  participant API as HyperSync API
  participant Hdl as User Handler

  YAML->>CLI: Parse+validate SVM programs/instructions
  CLI->>Envio: Build svmInstructionEventConfig (+ordering)
  Envio->>HSSrc: Provide event configs and schemas
  HSSrc->>HS: get(query with selections/filters)
  HS->>API: Fetch instructions/tx/logs
  API-->>HS: Response (slots/instr/tx/logs)
  HS-->>HSSrc: Typed response
  HSSrc->>Envio: Route by programId+discriminator
  Envio->>Hdl: onInstruction(event with optional decoded args/accounts)

Loading

Estimated code review effort

🎯 5 (Critical) | ⏱️ ~120 minutes

Possibly related PRs

• enviodev/hyperindex#1105 — Shifts SVM handler surface toward instruction-based registration similar to this PR.
• enviodev/hyperindex#1212 — Continues HyperSync N-API integration and narrows client interfaces leveraged here.
• enviodev/hyperindex#1036 — Touches the same codegen path for EventTemplate and field typing expanded in this PR.

Suggested reviewers

• JonoPrest
• DZakh

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

⚔️ Resolve merge conflicts

• Resolve merge conflict in branch svm-alpha-bugfixes

[coderabbitai]

Actionable comments posted: 8

Caution

Some comments are outside the diff and can’t be posted inline due to platform limitations.

⚠️ Outside diff range comments (2)

packages/cli/src/config_parsing/system_config.rs (2)

1090-1101: ⚠️ Potential issue | 🟠 Major | ⚡ Quick win

Don't hardcode every SVM chain to id 0.

Every SVM network is inserted into chains with the same key/id, so a second configured chain will fail unique_hashmap::try_insert instead of producing a usable multichain config. Either reject multi-chain SVM configs up front or carry a real chain identifier through this model.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@packages/cli/src/config_parsing/system_config.rs` around lines 1090 - 1101,
The code currently hardcodes Chain.id = 0 causing
unique_hashmap::try_insert(&mut chains, chain.id, chain) to fail for multiple
SVM networks; instead propagate a real identifier from the source network (use
network.id or compute a stable unique id) when constructing the Chain struct so
each Chain has a distinct chain.id, or explicitly detect and reject multi-chain
SVM configs before insertion; update the Chain construction site and the call
sites that expect chain.id (the Chain struct, the chain creation block, and the
unique_hashmap::try_insert invocation) accordingly so duplicate-key insertion no
longer occurs.

---

1104-1123: ⚠️ Potential issue | 🟠 Major | ⚡ Quick win

Preserve top-level handlers for SVM configs.

This branch hardcodes handlers: None, so BaseConfig.handlers is ignored only for SVM projects.

Suggested fix

-                    handlers: None,
+                    handlers: svm_config.base.handlers.clone(),

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@packages/cli/src/config_parsing/system_config.rs` around lines 1104 - 1123,
The SystemConfig builder for SVM projects is forcing handlers to None, which
drops any top-level handlers from the incoming SVM base config; change the
handlers assignment in the SystemConfig construction to preserve the original
handlers (e.g., use svm_config.base.handlers.clone() or otherwise map
svm_config.base.handlers into the SystemConfig.handlers field) instead of
hardcoding handlers: None so BaseConfig.handlers are honored for SVM configs.

🧹 Nitpick comments (5)

scenarios/svm_metaplex_demo/src/handlers/TokenMetadataHandlers.ts (2)

40-40: ⚡ Quick win

Consider using undefined instead of empty string for unknown mint.

Using "" for an unknown mint makes it impossible to distinguish between "mint is genuinely an empty string" and "mint is unknown/missing." This could complicate downstream queries and analytics. Consider using undefined or null to represent missing data explicitly, or document this convention clearly in the schema/README.

📝 Suggested alternative

-    const mint = accounts.mint ?? "";
+    const mint = accounts.mint;

And line 93:

-        mint: "",
+        mint: undefined,

Ensure the GraphQL schema marks mint as nullable if taking this approach.

Also applies to: 93-93

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@scenarios/svm_metaplex_demo/src/handlers/TokenMetadataHandlers.ts` at line
40, The code currently uses an empty string sentinel for missing mints (const
mint = accounts.mint ?? "") which obscures whether a mint is genuinely empty or
absent; change both occurrences (the mint assignment at accounts.mint and the
other instance around line 93) to return undefined (e.g., const mint =
accounts.mint ?? undefined) or otherwise leave the value unset when missing, and
update the GraphQL schema/type for the mint field to be nullable so downstream
consumers can distinguish absent vs empty values; ensure any code that consumes
TokenMetadataHandlers' mint variable handles undefined accordingly.

---

97-97: ⚡ Quick win

createdAtSlot is misleading when the account pre-existed start_block.

Setting createdAtSlot: event.slot (line 97) when the metadata account existed before indexing started creates misleading data. While the comment on lines 89-90 explains the limitation, storing an incorrect creation slot could confuse users analyzing account history. Consider renaming the field to firstObservedSlot or storing null/a sentinel value to signal "creation time unknown."

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@scenarios/svm_metaplex_demo/src/handlers/TokenMetadataHandlers.ts` at line
97, The field createdAtSlot currently set to event.slot is misleading for
accounts that pre-existed indexing; change its semantics to reflect observation
rather than creation by either (A) renaming createdAtSlot to firstObservedSlot
across TokenMetadataHandlers.ts and any related types/interfaces and migrations,
and set firstObservedSlot = event.slot when the account was observed, or (B)
keep createdAtSlot name but store null (or a clearly documented sentinel value)
when the account existed before start_block; update any code that reads/writes
createdAtSlot/firstObservedSlot (constructors, serializers, DB mappings, and
consumers) so type definitions accept null/sentinel and tests/consumers are
adjusted accordingly to avoid implying a false creation time.

scenarios/svm_metaplex_demo/src/indexer.test.ts (1)

32-42: ⚡ Quick win

Improve type safety by avoiding any and guarding array access.

The use of any[] types (lines 32-33) and unguarded array indexing (line 42) reduces type safety. Consider:

1. Defining proper types for TokenMetadataAccount and ProgramStats changes based on the envio package exports.
2. Checking statsChanges.length > 0 before accessing the last element, or using optional chaining.

🛡️ Suggested improvements

-      const tokenChanges: any[] = [];
-      const statsChanges: any[] = [];
+      const tokenChanges: Array<{ id: string; mint: string; /* ... */ }> = [];
+      const statsChanges: Array<ProgramStats> = [];
       let totalInstructionsAcrossBatches = 0;
       for (const change of result.changes) {
-        const tma = (change as any).TokenMetadataAccount;
+        const tma = change.TokenMetadataAccount;
         if (tma?.sets) tokenChanges.push(...tma.sets);
-        const ps = (change as any).ProgramStats;
+        const ps = change.ProgramStats;
         if (ps?.sets) statsChanges.push(...ps.sets);
-        totalInstructionsAcrossBatches += (change as any).eventsProcessed ?? 0;
+        totalInstructionsAcrossBatches += change.eventsProcessed ?? 0;
       }
-      const finalStats = statsChanges[statsChanges.length - 1];
+      const finalStats = statsChanges.at(-1);

Note: Adjust the type definitions to match the actual shape returned by createTestIndexer().process().

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@scenarios/svm_metaplex_demo/src/indexer.test.ts` around lines 32 - 42,
Replace the unsafe any[] usage and unguarded index access: define proper types
for the entries returned in result.changes (use the exported types for
TokenMetadataAccount and ProgramStats from the envio package) and type the
arrays tokenChanges and statsChanges accordingly; in the loop that reads (change
as any).TokenMetadataAccount / ProgramStats, cast to those concrete types
instead of any, extract sets into tokenChanges and statsChanges with correct
element types, and before computing finalStats use a guard (e.g., if
(statsChanges.length > 0) or optional chaining) to avoid accessing
statsChanges[statsChanges.length - 1] when the array is empty. Ensure
totalInstructionsAcrossBatches logic keeps using the typed
change.eventsProcessed field.

packages/envio/src/sources/HyperSyncSolanaSource.res (1)

306-311: 💤 Low value

Consider computing instruction selections once at source construction.

buildInstructionSelections(eventConfigs) is called on every getItemsOrThrow invocation, but eventConfigs is fixed at source creation time. Moving this computation to make and capturing the result would avoid redundant work on each fetch.

♻️ Suggested refactor

 let make = ({chain, endpointUrl, apiToken, eventConfigs, clientMaxRetries, clientTimeoutMillis}: options): t => {
   let name = "HyperSyncSolana"
   let chainId = chain->ChainMap.Chain.toChainId
 
   let client = HyperSyncSolanaClient.make(
     ~url=endpointUrl,
     ~apiToken=?apiToken,
     ~httpReqTimeoutMillis=clientTimeoutMillis,
     ~maxNumRetries=clientMaxRetries,
   )
 
   let (eventRouter, programOrderings) =
     EventRouter.fromSvmEventConfigsOrThrow(eventConfigs, ~chain)
 
+  // Compute once at startup - eventConfigs doesn't change between fetches
+  let instructionSelections = buildInstructionSelections(eventConfigs)
+
   // ...
 
   let getItemsOrThrow = async (...) => {
     // ...
-    let instructionSelections = buildInstructionSelections(eventConfigs)
     let query: HyperSyncSolanaClient.query = {
       fromSlot: fromBlock,
       toSlot: ?toBlock,
       instructions: instructionSelections,
     }

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@packages/envio/src/sources/HyperSyncSolanaSource.res` around lines 306 - 311,
The code recomputes instructionSelections every time getItemsOrThrow runs;
instead, call buildInstructionSelections(eventConfigs) once during source
construction (in make) and capture the result (e.g., closure variable or a
property on the source) so getItemsOrThrow uses that precomputed
instructionSelections when building the HyperSyncSolanaClient.query
(fromSlot/toSlot/instructions). Update make to compute and store
instructionSelections, remove the repeated call from getItemsOrThrow, and ensure
the stored value is referenced where instructionSelections is currently used.

packages/cli/src/hypersync_source/types.rs (1)

160-178: 💤 Low value

expect() calls can panic on malformed authorization data.

The try_from_be_slice calls with .expect() will panic if the byte slices have unexpected lengths. While the format should be fixed for blockchain data, consider using context() with ? to propagate errors gracefully instead of crashing the Node process.

♻️ Proposed change to return Result

If Authorization::from could return a Result, these could use ?. Alternatively, if the struct must impl From, consider wrapping the conversion in a TryFrom impl and calling that from a fallible context.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@packages/cli/src/hypersync_source/types.rs` around lines 160 - 178, The impl
From<&format::Authorization> currently uses expect() on
ruint::aliases::U256::try_from_be_slice and
alloy_primitives::I64::try_from_be_slice which can panic; replace this with a
fallible conversion by implementing TryFrom<&format::Authorization> for
Authorization (returning Result<Authorization, Error>) and use the ? operator
(and add context() or map_err to produce meaningful errors) when calling
ruint::aliases::U256::try_from_be_slice,
alloy_primitives::I64::try_from_be_slice and any other conversions (e.g.,
convert_bigint_unsigned), then update callers to handle the Result or provide a
separate infallible From that calls the TryFrom and maps errors appropriately.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@packages/cli/src/config_parsing/system_config.rs`:
- Around line 1503-1543: resolve_instruction_layout can't use parsed Anchor IDL
schemas because lookup_program_schema only returns schemas for
SvmSchemaSource::Bundled; update the ABI to keep the parsed program schema and
make lookup_program_schema return it: add a program_schema field
(Option<&'static SvmProgramSchema> or owned SvmProgramSchema as appropriate) to
SvmAbi and populate it when creating the ABI from schema_from_anchor_idl_json
and when using bundled_program_schemas (use the getter result), then change
lookup_program_schema to first return abi.program_schema if present and
otherwise fall back to bundled_program_schemas() like it currently does;
reference SvmAbi, SvmSchemaSource, lookup_program_schema,
resolve_instruction_layout, bundled_program_schemas, and
schema_from_anchor_idl_json when applying the changes.
- Around line 1500-1504: The code currently resolves idl_path against
project_paths.project_root making config-relative paths fail; change resolution
to be relative to the config file directory instead: when computing abs for
reading the IDL, if idl_path is absolute keep it, otherwise join it with the
directory containing the config (e.g. the parent of project_paths.config or
equivalent) rather than project_paths.project_root; update the use site where
abs is computed (the block that creates abs, reads body, and calls
schema_from_anchor_idl_json) to use this config-relative base so nested configs
load their IDL correctly.

In `@packages/cli/src/config_parsing/validation.rs`:
- Around line 216-233: The current is_valid_solana_pubkey only checks
Base58-like chars and length, allowing strings that decode to non-32-byte
values; update is_valid_solana_pubkey to actually decode the input using a
Base58 decoder (e.g., bs58::decode) and return true only if decoding succeeds
and the resulting byte vector is exactly 32 bytes, otherwise return false; keep
the existing fast reject on gross length if desired but remove relying solely on
character checks and handle decode errors gracefully in the function.

In `@packages/cli/src/hypersync_source_svm/query.rs`:
- Around line 203-213: The conversion in impl TryFrom<SolanaQuery> for
net::SolanaQuery currently silently drops negative max_num_* values and uses
unchecked casts (v as usize); instead validate each optional limit
(max_num_blocks, max_num_transactions, max_num_instructions, max_num_logs) and
return an Err when a value is negative or cannot be losslessly converted to
usize. Locate the fields in the TryFrom implementation and replace the
.filter(...).map(|v| v as usize) logic with explicit checks (e.g., match or if
let) that reject negative inputs and use a safe conversion (v.try_into() or
similar) with error mapping so invalid or out-of-range inputs produce a
descriptive conversion error rather than being silently dropped or truncated.

In `@packages/cli/src/hypersync_source/query.rs`:
- Around line 454-456: The cast of query.from_block and query.to_block into u64
(in the net_types::Query construction) can silently wrap negative i64 values;
replace the direct casts with fallible conversions that validate non-negativity
(e.g., use TryFrom/TryInto or explicit checks) and propagate an error if a
negative value is encountered: validate query.from_block >= 0 before converting
to u64 and convert query.to_block with map(|b| b.try_into().map_err(|_| ...)?),
returning or mapping a descriptive error instead of producing u64::MAX. Ensure
the error type matches the surrounding function's Result so callers get a clear
failure instead of wrapped values.

In `@packages/envio/svm.schema.json`:
- Around line 333-366: The schema currently allows invalid CLI values: tighten
AccountFilter.position by lowering "maximum" from 255 to 5 (keep integer/format
but enforce max 5) and make AnyOfAccountFilters.any_of a non-empty array with
non-empty groups by adding "minItems": 1 on the any_of array and also add
"minItems": 1 to its inner items arrays (the arrays of AccountFilter groups) so
any_of: [] and any_of: [[]] are disallowed; refer to the AccountFilter
definition (properties.position) and the AnyOfAccountFilters definition
(properties.any_of) to apply these changes.

In `@scenarios/svm_metaplex_demo/README.md`:
- Around line 18-20: The optional copy command should be guarded so it only runs
when CARGO_TARGET_DIR is set; wrap the cp -f
"$CARGO_TARGET_DIR/debug/libenvio.so" target/debug/libenvio.so command in a
shell conditional (e.g., check [ -n "$CARGO_TARGET_DIR" ] or use
${CARGO_TARGET_DIR:+...}) to skip the copy when CARGO_TARGET_DIR is unset and
optionally print a short message; update the README.md instruction around the
CARGO_TARGET_DIR usage to show this guarded form and keep the step optional.

In `@scenarios/test_codegen/test/RpcSource_test.res`:
- Around line 39-44: Remove the "DO NOT MERGE WITH THESE TESTS SKIPPED" comments
in RpcSource_test.res and either (A) re-enable the tests by addressing the CI
ENVIO_API_TOKEN access to eth.rpc.hypersync.xyz so the it_skip usages (search
for it_skip or the specific skipped test blocks) are no longer needed, or (B) if
you must keep the RPC tests skipped, replace the DO-NOT-MERGE text with a clear
skip explanation that references a tracking issue (e.g., "Skipped pending
`#ISSUE_NUMBER`: CI ENVIO_API_TOKEN lacks access to eth.rpc.hypersync.xyz (403)")
and apply the same change to the other occurrences noted (around the other
comment ranges).

---

Outside diff comments:
In `@packages/cli/src/config_parsing/system_config.rs`:
- Around line 1090-1101: The code currently hardcodes Chain.id = 0 causing
unique_hashmap::try_insert(&mut chains, chain.id, chain) to fail for multiple
SVM networks; instead propagate a real identifier from the source network (use
network.id or compute a stable unique id) when constructing the Chain struct so
each Chain has a distinct chain.id, or explicitly detect and reject multi-chain
SVM configs before insertion; update the Chain construction site and the call
sites that expect chain.id (the Chain struct, the chain creation block, and the
unique_hashmap::try_insert invocation) accordingly so duplicate-key insertion no
longer occurs.
- Around line 1104-1123: The SystemConfig builder for SVM projects is forcing
handlers to None, which drops any top-level handlers from the incoming SVM base
config; change the handlers assignment in the SystemConfig construction to
preserve the original handlers (e.g., use svm_config.base.handlers.clone() or
otherwise map svm_config.base.handlers into the SystemConfig.handlers field)
instead of hardcoding handlers: None so BaseConfig.handlers are honored for SVM
configs.

---

Nitpick comments:
In `@packages/cli/src/hypersync_source/types.rs`:
- Around line 160-178: The impl From<&format::Authorization> currently uses
expect() on ruint::aliases::U256::try_from_be_slice and
alloy_primitives::I64::try_from_be_slice which can panic; replace this with a
fallible conversion by implementing TryFrom<&format::Authorization> for
Authorization (returning Result<Authorization, Error>) and use the ? operator
(and add context() or map_err to produce meaningful errors) when calling
ruint::aliases::U256::try_from_be_slice,
alloy_primitives::I64::try_from_be_slice and any other conversions (e.g.,
convert_bigint_unsigned), then update callers to handle the Result or provide a
separate infallible From that calls the TryFrom and maps errors appropriately.

In `@packages/envio/src/sources/HyperSyncSolanaSource.res`:
- Around line 306-311: The code recomputes instructionSelections every time
getItemsOrThrow runs; instead, call buildInstructionSelections(eventConfigs)
once during source construction (in make) and capture the result (e.g., closure
variable or a property on the source) so getItemsOrThrow uses that precomputed
instructionSelections when building the HyperSyncSolanaClient.query
(fromSlot/toSlot/instructions). Update make to compute and store
instructionSelections, remove the repeated call from getItemsOrThrow, and ensure
the stored value is referenced where instructionSelections is currently used.

In `@scenarios/svm_metaplex_demo/src/handlers/TokenMetadataHandlers.ts`:
- Line 40: The code currently uses an empty string sentinel for missing mints
(const mint = accounts.mint ?? "") which obscures whether a mint is genuinely
empty or absent; change both occurrences (the mint assignment at accounts.mint
and the other instance around line 93) to return undefined (e.g., const mint =
accounts.mint ?? undefined) or otherwise leave the value unset when missing, and
update the GraphQL schema/type for the mint field to be nullable so downstream
consumers can distinguish absent vs empty values; ensure any code that consumes
TokenMetadataHandlers' mint variable handles undefined accordingly.
- Line 97: The field createdAtSlot currently set to event.slot is misleading for
accounts that pre-existed indexing; change its semantics to reflect observation
rather than creation by either (A) renaming createdAtSlot to firstObservedSlot
across TokenMetadataHandlers.ts and any related types/interfaces and migrations,
and set firstObservedSlot = event.slot when the account was observed, or (B)
keep createdAtSlot name but store null (or a clearly documented sentinel value)
when the account existed before start_block; update any code that reads/writes
createdAtSlot/firstObservedSlot (constructors, serializers, DB mappings, and
consumers) so type definitions accept null/sentinel and tests/consumers are
adjusted accordingly to avoid implying a false creation time.

In `@scenarios/svm_metaplex_demo/src/indexer.test.ts`:
- Around line 32-42: Replace the unsafe any[] usage and unguarded index access:
define proper types for the entries returned in result.changes (use the exported
types for TokenMetadataAccount and ProgramStats from the envio package) and type
the arrays tokenChanges and statsChanges accordingly; in the loop that reads
(change as any).TokenMetadataAccount / ProgramStats, cast to those concrete
types instead of any, extract sets into tokenChanges and statsChanges with
correct element types, and before computing finalStats use a guard (e.g., if
(statsChanges.length > 0) or optional chaining) to avoid accessing
statsChanges[statsChanges.length - 1] when the array is empty. Ensure
totalInstructionsAcrossBatches logic keeps using the typed
change.eventsProcessed field.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 38c79c00-6c4a-405b-a99e-e395cdb1f775

📥 Commits

Reviewing files that changed from the base of the PR and between d600268 and ab6ba21.

⛔ Files ignored due to path filters (4)

• Cargo.lock is excluded by !**/*.lock
• packages/cli/src/hbs_templating/snapshots/envio__hbs_templating__codegen_templates__test__indexer_code_generates_correct_types_and_values.snap is excluded by !**/*.snap
• packages/cli/src/hbs_templating/snapshots/envio__hbs_templating__codegen_templates__test__indexer_code_multiple_chains.snap is excluded by !**/*.snap
• pnpm-lock.yaml is excluded by !**/pnpm-lock.yaml

📒 Files selected for processing (70)

• STAGE_7B_DECISIONS.md
• packages/cli/Cargo.toml
• packages/cli/CommandLineHelp.md
• packages/cli/src/cli_args/init_config.rs
• packages/cli/src/config_parsing/human_config.rs
• packages/cli/src/config_parsing/public_config.rs
• packages/cli/src/config_parsing/system_config.rs
• packages/cli/src/config_parsing/validation.rs
• packages/cli/src/hbs_templating/codegen_templates.rs
• packages/cli/src/hbs_templating/contract_import_templates.rs
• packages/cli/src/hypersync_source/config.rs
• packages/cli/src/hypersync_source/decode.rs
• packages/cli/src/hypersync_source/mod.rs
• packages/cli/src/hypersync_source/query.rs
• packages/cli/src/hypersync_source/types.rs
• packages/cli/src/hypersync_source_svm/borsh_decoder.rs
• packages/cli/src/hypersync_source_svm/config.rs
• packages/cli/src/hypersync_source_svm/mod.rs
• packages/cli/src/hypersync_source_svm/query.rs
• packages/cli/src/hypersync_source_svm/types.rs
• packages/cli/src/lib.rs
• packages/cli/src/template_dirs.rs
• packages/cli/templates/dynamic/init_templates/shared/package.json.hbs
• packages/cli/templates/static/shared/pnpm-workspace.yaml
• packages/cli/templates/static/svm_metaplex_template/typescript/.env.example
• packages/cli/templates/static/svm_metaplex_template/typescript/.gitignore
• packages/cli/templates/static/svm_metaplex_template/typescript/README.md
• packages/cli/templates/static/svm_metaplex_template/typescript/config.yaml
• packages/cli/templates/static/svm_metaplex_template/typescript/schema.graphql
• packages/cli/templates/static/svm_metaplex_template/typescript/src/handlers/TokenMetadataHandlers.ts
• packages/cli/templates/static/svm_metaplex_template/typescript/tsconfig.json
• packages/cli/test/configs/svm-metaplex-config.yaml
• packages/envio/evm.schema.json
• packages/envio/fuel.schema.json
• packages/envio/index.d.ts
• packages/envio/package.json
• packages/envio/src/ChainFetcher.res
• packages/envio/src/Config.res
• packages/envio/src/Core.res
• packages/envio/src/Envio.res
• packages/envio/src/EventConfigBuilder.res
• packages/envio/src/HandlerLoader.res
• packages/envio/src/Hasura.res
• packages/envio/src/Internal.res
• packages/envio/src/Main.res
• packages/envio/src/SimulateItems.res
• packages/envio/src/SvmTypes.res
• packages/envio/src/sources/EventRouter.res
• packages/envio/src/sources/HyperSyncClient.res
• packages/envio/src/sources/HyperSyncSolanaClient.res
• packages/envio/src/sources/HyperSyncSolanaSource.res
• packages/envio/src/sources/HyperSyncSource.res
• packages/envio/src/sources/Svm.res
• packages/envio/src/tui/Tui.res
• packages/envio/src/tui/components/TuiData.res
• packages/envio/svm.schema.json
• scenarios/svm_metaplex_demo/.envio/.gitignore
• scenarios/svm_metaplex_demo/README.md
• scenarios/svm_metaplex_demo/config.test.yaml
• scenarios/svm_metaplex_demo/config.yaml
• scenarios/svm_metaplex_demo/envio-env.d.ts
• scenarios/svm_metaplex_demo/package.json
• scenarios/svm_metaplex_demo/schema.graphql
• scenarios/svm_metaplex_demo/src/handlers/TokenMetadataHandlers.ts
• scenarios/svm_metaplex_demo/src/indexer.test.ts
• scenarios/svm_metaplex_demo/tsconfig.json
• scenarios/svm_metaplex_demo/vitest.config.ts
• scenarios/test_codegen/test/EventRouter_svm_test.res
• scenarios/test_codegen/test/HyperSyncSolanaClient_test.res
• scenarios/test_codegen/test/RpcSource_test.res

💤 Files with no reviewable changes (1)

• packages/envio/package.json

[coderabbitai]

⚠️ Potential issue | 🟠 Major | ⚡ Quick win

Resolve program.idl relative to config.yaml.

idl is documented as config-relative, but project_root.join(idl_path) makes it repo-root-relative instead. Nested config locations will fail to load their IDL.

Suggested fix

-        let abs = project_paths.project_root.join(idl_path);
+        let abs = path_utils::get_config_path_relative_to_root(
+            project_paths,
+            PathBuf::from(idl_path),
+        )?;
         let body = fs::read_to_string(&abs)
             .with_context(|| format!("reading IDL at '{}'", abs.display()))?;

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@packages/cli/src/config_parsing/system_config.rs` around lines 1500 - 1504,
The code currently resolves idl_path against project_paths.project_root making
config-relative paths fail; change resolution to be relative to the config file
directory instead: when computing abs for reading the IDL, if idl_path is
absolute keep it, otherwise join it with the directory containing the config
(e.g. the parent of project_paths.config or equivalent) rather than
project_paths.project_root; update the use site where abs is computed (the block
that creates abs, reads body, and calls schema_from_anchor_idl_json) to use this
config-relative base so nested configs load their IDL correctly.

[coderabbitai]

⚠️ Potential issue | 🟠 Major | 🏗️ Heavy lift

Keep the parsed SVM program schema around for layout resolution.

resolve_instruction_layout only infers accounts/args when program_schema is Some, but this code only reconstructs a schema for bundled programs and only when no instruction has inline schema. That means Anchor IDLs always fall back to empty layouts, and one inline override on a bundled program disables schema fallback for every other instruction.

Also applies to: 1572-1584

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@packages/cli/src/config_parsing/system_config.rs` around lines 1503 - 1543,
resolve_instruction_layout can't use parsed Anchor IDL schemas because
lookup_program_schema only returns schemas for SvmSchemaSource::Bundled; update
the ABI to keep the parsed program schema and make lookup_program_schema return
it: add a program_schema field (Option<&'static SvmProgramSchema> or owned
SvmProgramSchema as appropriate) to SvmAbi and populate it when creating the ABI
from schema_from_anchor_idl_json and when using bundled_program_schemas (use the
getter result), then change lookup_program_schema to first return
abi.program_schema if present and otherwise fall back to
bundled_program_schemas() like it currently does; reference SvmAbi,
SvmSchemaSource, lookup_program_schema, resolve_instruction_layout,
bundled_program_schemas, and schema_from_anchor_idl_json when applying the
changes.

[coderabbitai]

⚠️ Potential issue | 🟠 Major | ⚡ Quick win

Decode pubkeys instead of only checking alphabet and length.

This accepts any 32-44 char Base58-looking string, including values that decode to something other than 32 bytes. Invalid program_ids and account-filter pubkeys can therefore pass validation and fail much later at runtime.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@packages/cli/src/config_parsing/validation.rs` around lines 216 - 233, The
current is_valid_solana_pubkey only checks Base58-like chars and length,
allowing strings that decode to non-32-byte values; update
is_valid_solana_pubkey to actually decode the input using a Base58 decoder
(e.g., bs58::decode) and return true only if decoding succeeds and the resulting
byte vector is exactly 32 bytes, otherwise return false; keep the existing fast
reject on gross length if desired but remove relying solely on character checks
and handle decode errors gracefully in the function.

[coderabbitai]

⚠️ Potential issue | 🟠 Major | ⚡ Quick win

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
set -euo pipefail

# Verify unchecked casts and silent-negative filtering in this file.
rg -n -C2 'max_num_(blocks|transactions|instructions|logs).*as usize|filter\(\|v\| \*v >= 0\)' packages/cli/src/hypersync_source_svm/query.rs

# Verify whether CI/build config includes any 32-bit targets where wrapping is possible.
fd -HI 'Cargo.toml|config.toml|*.yml|*.yaml' | xargs rg -n -C1 'i686|armv7|x86|32|target'

Repository: enviodev/hyperindex

Length of output: 50732

---

🏁 Script executed:

#!/bin/bash
set -euo pipefail

FILE="packages/cli/src/hypersync_source_svm/query.rs"

echo "=== max_num_* block in context ==="
sed -n '150,230p' "$FILE"

echo "=== occurrences of max_num_* in query.rs ==="
rg -n "max_num_(blocks|transactions|instructions|logs)" "$FILE"

echo "=== type of q and relevant struct definitions (near occurrences) ==="
# Show a bit more around the first occurrence to capture the surrounding struct/impl.
FIRST_LINE=$(rg -n "max_num_blocks" "$FILE" | head -n 1 | cut -d: -f1)
START=$((FIRST_LINE-60))
END=$((FIRST_LINE+40))
sed -n "${START},${END}p" "$FILE"

echo "=== CI/workflow / config scan for 32-bit Rust targets ==="
# GitHub Actions workflows
if [ -d .github/workflows ]; then
  rg -n --hidden --glob '.github/workflows/*.yml' '(i686|i386|armv7|--target\b|target\s*=)' .github/workflows || true
  rg -n --hidden --glob '.github/workflows/*.yaml' '(i686|i386|armv7|--target\b|target\s*=)' .github/workflows || true
else
  echo "(no .github/workflows directory)"
fi

# Cargo/rust config locations (if present)
if [ -d .cargo ]; then
  rg -n --hidden '(i686|i386|armv7|--target\b|target\s*=)' .cargo || true
fi

# Any explicit target triples in repo config files
fd -HI --type f --exclude 'node_modules' 'Cargo.toml' 'Cargo.lock' 'rust-toolchain.toml' 'rust-toolchain' '*.yml' '*.yaml' '*.toml' 2>/dev/null \
  | xargs -r rg -n '(i686|i386|armv7|--target\b|target\s*=|wasm32)' || true

Repository: enviodev/hyperindex

Length of output: 4716

---

🏁 Script executed:

#!/bin/bash
set -euo pipefail

FILE="packages/cli/src/hypersync_source_svm/query.rs"

echo "=== max_num_* block in context ==="
sed -n '150,230p' "$FILE"

echo "=== occurrences of max_num_* in query.rs ==="
rg -n "max_num_(blocks|transactions|instructions|logs)" "$FILE"

echo "=== type context near first max_num_blocks ==="
FIRST_LINE=$(rg -n "max_num_blocks" "$FILE" | head -n 1 | cut -d: -f1)
START=$((FIRST_LINE-80))
END=$((FIRST_LINE+60))
sed -n "${START},${END}p" "$FILE"

echo "=== CI/workflow / config scan for 32-bit Rust targets ==="
if [ -d .github/workflows ]; then
  rg -n --hidden --glob '.github/workflows/*.yml' '(i686|i386|armv7|--target\b|target\s*=)' .github/workflows || true
  rg -n --hidden --glob '.github/workflows/*.yaml' '(i686|i386|armv7|--target\b|target\s*=)' .github/workflows || true
else
  echo "(no .github/workflows directory)"
fi

if [ -d .cargo ]; then
  rg -n --hidden '(i686|i386|armv7|--target\b|target\s*=)' .cargo || true
fi

# Limit search to likely config files to avoid huge outputs
fd -HI --type f \
  -g 'Cargo.toml' -g 'rust-toolchain*' \
  -g '*.yml' -g '*.yaml' -g '*.toml' \
  -g 'Dockerfile*' -g 'Makefile*' -g 'justfile' \
  | xargs -r rg -n '(i686|i386|armv7|--target\b|target\s*=|wasm32)' || true

Repository: enviodev/hyperindex

Length of output: 4694

---

🏁 Script executed:

#!/bin/bash
set -euo pipefail

FILE="packages/cli/src/hypersync_source_svm/query.rs"

echo "=== max_num_* block in context ==="
sed -n '150,230p' "$FILE"

echo "=== occurrences of max_num_* in query.rs ==="
rg -n "max_num_(blocks|transactions|instructions|logs)" "$FILE"

echo "=== type context near first max_num_blocks ==="
FIRST_LINE=$(rg -n "max_num_blocks" "$FILE" | head -n 1 | cut -d: -f1)
START=$((FIRST_LINE-80))
END=$((FIRST_LINE+60))
sed -n "${START},${END}p" "$FILE"

echo "=== CI/workflow / config scan for 32-bit Rust targets ==="
if [ -d .github/workflows ]; then
  rg -n --hidden --glob '.github/workflows/*.yml' '(i686|i386|armv7|--target\b|target\s*=)' .github/workflows || true
  rg -n --hidden --glob '.github/workflows/*.yaml' '(i686|i386|armv7|--target\b|target\s*=)' .github/workflows || true
else
  echo "(no .github/workflows directory)"
fi

if [ -d .cargo ]; then
  rg -n --hidden '(i686|i386|armv7|--target\b|target\s*=)' .cargo || true
fi

# Limit search to likely config files to avoid huge outputs
fd -HI --type f \
  -g 'Cargo.toml' -g 'rust-toolchain*' \
  -g '*.yml' -g '*.yaml' -g '*.toml' \
  -g 'Dockerfile*' -g 'Makefile*' -g 'justfile' \
  | xargs -r rg -n '(i686|i386|armv7|--target\b|target\s*=|wasm32)' || true

Repository: enviodev/hyperindex

Length of output: 4694

---

🏁 Script executed:

#!/bin/bash
set -euo pipefail

FILE="packages/cli/src/hypersync_source_svm/query.rs"

echo "=== imports/top of file ==="
sed -n '1,120p' "$FILE"

echo "=== definition of SolanaQuery (where max_num_* are declared) ==="
rg -n "struct SolanaQuery|pub max_num_blocks|pub max_num_transactions|pub max_num_instructions|pub max_num_logs" "$FILE"
# Print around the struct block that contains the max_num_* fields
FIRST=$(rg -n "pub max_num_blocks" "$FILE" | head -n 1 | cut -d: -f1)
START=$((FIRST-40)); END=$((FIRST+40))
sed -n "${START},${END}p" "$FILE"

echo "=== SolanaQuery -> net::SolanaQuery mapping function signature and any existing anyhow imports/uses ==="
rg -n "impl TryFrom<SolanaQuery> for net::SolanaQuery|type Error|anyhow::ensure!|transpose\\(\\)\\?" "$FILE"
FIRST_TRY=$(rg -n "impl TryFrom<SolanaQuery> for net::SolanaQuery" "$FILE" | head -n 1 | cut -d: -f1)
START2=$((FIRST_TRY-30)); END2=$((FIRST_TRY+80))
sed -n "${START2},${END2}p" "$FILE"

Repository: enviodev/hyperindex

Length of output: 6212

---

🏁 Script executed:

#!/bin/bash
set -euo pipefail
FILE="packages/cli/src/hypersync_source_svm/query.rs"
rg -n "anyhow::\\{[^}]*Context|use anyhow::\\{[^}]*Context|use anyhow::Context|\\bContext\\b" "$FILE" || true
rg -n "\\.context\\(" "$FILE" || true

Repository: enviodev/hyperindex

Length of output: 45

---

🏁 Script executed:

#!/bin/bash
set -euo pipefail

# Find the actual net::SolanaQuery definition in the workspace/deps (likely in hypersync_solana_net_types).
rg -n "struct SolanaQuery" -S . --glob '*.rs' | head -n 50

# Look specifically for max_num_blocks field type usages near the net SolanaQuery.
rg -n "max_num_blocks\s*:" -S . --glob '*.rs' | head -n 50

Repository: enviodev/hyperindex

Length of output: 690

---

Reject invalid max_num_* limits instead of silently dropping/casting

In impl TryFrom<SolanaQuery> for net::SolanaQuery, max_num_blocks, max_num_transactions, max_num_instructions, and max_num_logs currently drop negative values via .filter(|v| *v >= 0) and then cast with v as usize without overflow checking, which can silently produce incorrect limits.

💡 Proposed fix

-use anyhow::Result;
+use anyhow::{Context, Result};
@@
-            max_num_blocks: q.max_num_blocks.filter(|v| *v >= 0).map(|v| v as usize),
+            max_num_blocks: q
+                .max_num_blocks
+                .map(|v| {
+                    anyhow::ensure!(v >= 0, "max_num_blocks must be non-negative");
+                    usize::try_from(v).context("max_num_blocks overflows usize")
+                })
+                .transpose()?,
@@
-            max_num_transactions: q
-                .max_num_transactions
-                .filter(|v| *v >= 0)
-                .map(|v| v as usize),
+            max_num_transactions: q
+                .max_num_transactions
+                .map(|v| {
+                    anyhow::ensure!(v >= 0, "max_num_transactions must be non-negative");
+                    usize::try_from(v).context("max_num_transactions overflows usize")
+                })
+                .transpose()?,
@@
-            max_num_instructions: q
-                .max_num_instructions
-                .filter(|v| *v >= 0)
-                .map(|v| v as usize),
+            max_num_instructions: q
+                .max_num_instructions
+                .map(|v| {
+                    anyhow::ensure!(v >= 0, "max_num_instructions must be non-negative");
+                    usize::try_from(v).context("max_num_instructions overflows usize")
+                })
+                .transpose()?,
@@
-            max_num_logs: q.max_num_logs.filter(|v| *v >= 0).map(|v| v as usize),
+            max_num_logs: q
+                .max_num_logs
+                .map(|v| {
+                    anyhow::ensure!(v >= 0, "max_num_logs must be non-negative");
+                    usize::try_from(v).context("max_num_logs overflows usize")
+                })
+                .transpose()?,

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@packages/cli/src/hypersync_source_svm/query.rs` around lines 203 - 213, The
conversion in impl TryFrom<SolanaQuery> for net::SolanaQuery currently silently
drops negative max_num_* values and uses unchecked casts (v as usize); instead
validate each optional limit (max_num_blocks, max_num_transactions,
max_num_instructions, max_num_logs) and return an Err when a value is negative
or cannot be losslessly converted to usize. Locate the fields in the TryFrom
implementation and replace the .filter(...).map(|v| v as usize) logic with
explicit checks (e.g., match or if let) that reject negative inputs and use a
safe conversion (v.try_into() or similar) with error mapping so invalid or
out-of-range inputs produce a descriptive conversion error rather than being
silently dropped or truncated.

[coderabbitai]

⚠️ Potential issue | 🟡 Minor | ⚡ Quick win

Negative block numbers silently wrap to large values.

from_block and to_block are cast from i64 to u64 without validation. A negative value (e.g., -1) would wrap to u64::MAX, causing unexpected query behavior. Consider validating or using TryFrom with an error.

🛡️ Proposed fix

 Ok(net_types::Query {
-    from_block: query.from_block as u64,
-    to_block: query.to_block.map(|b| b as u64),
+    from_block: query.from_block.try_into().context("from_block must be non-negative")?,
+    to_block: query.to_block.map(|b| b.try_into()).transpose().context("to_block must be non-negative")?,

📝 Committable suggestion

‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.

Suggested change

	Ok(net_types::Query {
	from_block: query.from_block as u64,
	to_block: query.to_block.map(|b| b as u64),
	Ok(net_types::Query {
	from_block: query.from_block.try_into().context("from_block must be non-negative")?,
	to_block: query.to_block.map(|b| b.try_into()).transpose().context("to_block must be non-negative")?,

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@packages/cli/src/hypersync_source/query.rs` around lines 454 - 456, The cast
of query.from_block and query.to_block into u64 (in the net_types::Query
construction) can silently wrap negative i64 values; replace the direct casts
with fallible conversions that validate non-negativity (e.g., use
TryFrom/TryInto or explicit checks) and propagate an error if a negative value
is encountered: validate query.from_block >= 0 before converting to u64 and
convert query.to_block with map(|b| b.try_into().map_err(|_| ...)?), returning
or mapping a descriptive error instead of producing u64::MAX. Ensure the error
type matches the surrounding function's Result so callers get a clear failure
instead of wrapped values.

[coderabbitai]

⚠️ Potential issue | 🟡 Minor | ⚡ Quick win

Tighten the schema to match the CLI's SVM filter validation.

The CLI now rejects positions above 5, empty any_of, and empty groups, but the schema still allows position: 255, any_of: [], and any_of: [[]]. That means editor/schema validation will bless configs the CLI later rejects.

Suggested fix

         "position": {
           "description": "Account position within the instruction (0..=5).",
           "type": "integer",
           "format": "uint8",
           "minimum": 0,
-          "maximum": 255
+          "maximum": 5
         },
@@
         "any_of": {
           "description": "A non-empty list of AND-groups. Each group is itself a non-empty list of `{position, values}` entries that must all match the same instruction. An instruction matches `any_of` when any one group matches.",
           "type": "array",
+          "minItems": 1,
           "items": {
             "type": "array",
+            "minItems": 1,
             "items": {
               "$ref": "`#/`$defs/AccountFilter"
             }
           }
         }

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@packages/envio/svm.schema.json` around lines 333 - 366, The schema currently
allows invalid CLI values: tighten AccountFilter.position by lowering "maximum"
from 255 to 5 (keep integer/format but enforce max 5) and make
AnyOfAccountFilters.any_of a non-empty array with non-empty groups by adding
"minItems": 1 on the any_of array and also add "minItems": 1 to its inner items
arrays (the arrays of AccountFilter groups) so any_of: [] and any_of: [[]] are
disallowed; refer to the AccountFilter definition (properties.position) and the
AnyOfAccountFilters definition (properties.any_of) to apply these changes.

[coderabbitai]

⚠️ Potential issue | 🟡 Minor | ⚡ Quick win

Guard the optional artifact copy command.

Line 19 can fail when CARGO_TARGET_DIR is unset, even though this step is marked optional.

Proposed doc fix

 # 2. (Optional) If you have CARGO_TARGET_DIR set, sync the artifact:
-cp -f "$CARGO_TARGET_DIR/debug/libenvio.so" target/debug/libenvio.so
+if [ -n "${CARGO_TARGET_DIR:-}" ]; then
+  cp -f "$CARGO_TARGET_DIR/debug/libenvio.so" target/debug/libenvio.so
+fi

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@scenarios/svm_metaplex_demo/README.md` around lines 18 - 20, The optional
copy command should be guarded so it only runs when CARGO_TARGET_DIR is set;
wrap the cp -f "$CARGO_TARGET_DIR/debug/libenvio.so" target/debug/libenvio.so
command in a shell conditional (e.g., check [ -n "$CARGO_TARGET_DIR" ] or use
${CARGO_TARGET_DIR:+...}) to skip the copy when CARGO_TARGET_DIR is unset and
optionally print a short message; update the README.md instruction around the
CARGO_TARGET_DIR usage to show this guarded form and keep the step optional.

[coderabbitai]

⚠️ Potential issue | 🔴 Critical | 🏗️ Heavy lift

Remove "DO NOT MERGE" comments before merging to main.

Three tests are marked with "DO NOT MERGE WITH THESE TESTS SKIPPED" comments. Merging code with explicit DO-NOT-MERGE markers is a critical process violation. These comments indicate unfinished work and create confusion about the PR's merge readiness.

According to the comments, the root cause is an infrastructure issue: CI's ENVIO_API_TOKEN lacks access to eth.rpc.hypersync.xyz (HTTP 403). Before merging:

1. Preferred: Resolve the token access issue and restore the tests, or
2. Alternative: If the RPC tests must remain skipped, remove the "DO NOT MERGE" comments and replace them with a clear explanation + tracking issue reference (e.g., "Skipped pending #ISSUE_NUMBER: CI token RPC access").

Also applies to: 143-144, 649-650

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@scenarios/test_codegen/test/RpcSource_test.res` around lines 39 - 44, Remove
the "DO NOT MERGE WITH THESE TESTS SKIPPED" comments in RpcSource_test.res and
either (A) re-enable the tests by addressing the CI ENVIO_API_TOKEN access to
eth.rpc.hypersync.xyz so the it_skip usages (search for it_skip or the specific
skipped test blocks) are no longer needed, or (B) if you must keep the RPC tests
skipped, replace the DO-NOT-MERGE text with a clear skip explanation that
references a tracking issue (e.g., "Skipped pending `#ISSUE_NUMBER`: CI
ENVIO_API_TOKEN lacks access to eth.rpc.hypersync.xyz (403)") and apply the same
change to the other occurrences noted (around the other comment ranges).