Skip to main content
For SDK maintainers and indexer operators who want to know which corpus entries exercise a given CAP or protocol feature, where their bytes come from, and how often the shape occurs on pubnet. Use it to pick the entries to run against your code when you add support for a CAP, or to judge whether a failing entry is a shape your production traffic contains.

How to use this page

  1. Find the row in the Rows table: one row per CAP, per protocol upgrade without a CAP, and per cross-cutting group (ledger, classic, xdr-strict, limits, strkey, xdr-json). Each label links to the row’s section (#cap-46, #classic).
  2. The section lists the row’s entries. An entry tagged with several CAPs is listed under each of them, so a row can hold entries whose ID is in another group.
  3. Run the entries: xdrcheck entry --cap 40 lists a CAP’s entries; xdrcheck entry ID shows one entry (tags, variants, expected values, declared divergences); xdrcheck conform run --entries 'cap-0040/*' runs an ID group through the SDK columns, and xdrcheck replay --caps 40 replays a CAP through your adapter (CLI).
How to read an entry’s line: Entries without a CAP number: 143. Their pattern predates CAPs, shipped in a protocol upgrade without a CAP, or is an encoding rule. Such an entry has caps: [] and a caps_note that says why, carries its protocol in protocol.introduced, and is listed under the row of its group (protocol-01 1, protocol-02 1, protocol-09 1, protocol-15 2, protocol-29 1, ledger 11, classic 97, xdr-strict 11, limits 2, strkey 15, xdr-json 1). Where a union arm is held by only one or two entries, the section has a collapsed table with how often the scan of pubnet history saw that arm, its first and last ledger, and example transactions. A count per entry is on Mainnet scan.

Rows

Captured: bytes from a network ledger. Core: emitted by stellar-core, locally or in its goldens. Synthetic: built by a recipe. Rare arms: union arms with a prevalence table in the row’s section.

CAP-1

Bump Sequence. Protocol 10. Row cap-0001. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-3

Asset-backed offers. Protocol 10. Row cap-0003. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-5

Throttling and transaction pricing improvements. Protocol 11. Row cap-0005. All rows

CAP-6

Add ManageBuyOffer Operation. Protocol 11. Row cap-0006. All rows No entry: an ordinary union arm with nothing reader-specific; exercised by the bulk tier (18 golden operations)

CAP-15

Fee-Bump Transactions. Protocol 13. Row cap-0015. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-18

Fine-Grained Control of Authorization. Protocol 13. Row cap-0018. All rows

CAP-20

Bucket Initial Entries. Protocol 11. Row cap-0020. All rows No entry: no testnet and pubnet twins: a bucket-file shape, not ledger meta: buckets are not on the meta stream; a both-passphrase bucket record is possible from a local chain but not planned for the curated tier

CAP-21

Preconditions: Generalized transaction preconditions. Protocol 19. Row cap-0021. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-23

Two-Part Payments with ClaimableBalanceEntry. Protocol 14. Row cap-0023. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-24

Make PathPayment Symmetrical. Protocol 12. Row cap-0024. All rows

CAP-26

Disable Inflation Mechanism. Protocol 12. Row cap-0026. All rows No entry: no twins yet; planned with testnet and pubnet variants: 1 (first cap-0026/inflation-op-not-supported)
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-27

First-class multiplexed accounts. Protocol 13. Row cap-0027. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-28

Clear pre-auth transaction signer on failed transactions. Protocol 13. Row cap-0028. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-30

Remove NO_ISSUER Operation Results. Protocol 13. Row cap-0030. All rows No entry: no testnet and pubnet twins: the payload arm is produced only before Protocol 13 (stellar-core checks issuers in path payments only below 13), and the local generator brings chains up to Protocol 13 at the lowest, the first that accepts ENVELOPE_TYPE_TX envelopes
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-33

Sponsored Reserve. Protocol 14. Row cap-0033. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-34

Preserve Transaction-Set/Close-Time Affinity During Nomination. Protocol 14. Row cap-0034. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-35

Asset Clawback. Protocol 17. Row cap-0035. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-38

Automated Market Makers. Protocol 18. Row cap-0038. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-40

Signed-Payload: Ed25519 Signed Payload Signer for Atomic Transaction Signature Disclosure. Protocol 19. Row cap-0040. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-42

Multi-Part Transaction Sets. Protocol 20. Row cap-0042. All rows

CAP-46

Soroban smart contract system overview. Protocol 20. Row cap-0046. All rows 1 more entry of this row, at a nesting-depth limit, is listed in corpus/COVERAGE.md.
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-54

Soroban refined VM instantiation cost model. Protocol 21. Row cap-0054. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-58

Constructors for Soroban contracts. Protocol 22. Row cap-0058. All rows

CAP-62

Soroban Live State Prioritization. Protocol 23. Row cap-0062. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-63

Parallelism-friendly Transaction Scheduling. Protocol 23. Row cap-0063. All rows

CAP-66

Soroban In-memory Read Resource. Protocol 23. Row cap-0066. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-67

Unified Asset Events. Protocol 23. Row cap-0067. All rows

CAP-70

Configurable SCP Timing Parameters. Protocol 23. Row cap-0070. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-71

Authentication delegation and address-bound Soroban credentials. Protocol 27. Row cap-0071. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-73

Allow SAC to create G-account balances. Protocol 26. Row cap-0073. All rows

CAP-76

P23 State Archival bug remediation. Protocol 24. Row cap-0076. All rows

CAP-77

Freeze Ledger Entries via Network Configuration. Protocol 26. Row cap-0077. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-83

Allow validators to vote to drop the transaction set from the current ledger. Protocol 28. Row cap-0083. All rows No entry: no testnet and pubnet twins: needs validators that drop a proposed transaction set; a single standalone node with manual close does not do that, so the entry is a synthetic header
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-84

Muxed Contract Addresses. Protocol 30. Row cap-0084. All rows

CAP-85

Externally managed contract executables. Protocol 28. Row cap-0085. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-87

Host functions for ML-DSA signature verification. Protocol 30. Row cap-0087. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

CAP-88

Millisecond-Resolution Close Times. Protocol 30. Row cap-0088. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

protocol-01

Earliest ledgers. Protocol 1. Row protocol-01. All rows No entry: no testnet and pubnet twins: a property of the public network’s own early history; only pubnet has these ledgers

protocol-02

Protocol 2 upgrade ledger. Protocol 2. Row protocol-02. All rows No entry: no testnet and pubnet twins: a property of the public network’s own upgrade history; only pubnet has this ledger

protocol-09

Protocol 9 upgrade ledger. Protocol 9. Row protocol-09. All rows No entry: no testnet and pubnet twins: a property of the public network’s own upgrade history; only pubnet has this ledger
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

protocol-13

Protocol 13 upgrade ledger. Protocol 13. Row protocol-13. All rows No entry: no testnet and pubnet twins: a property of the public network’s own upgrade history; only pubnet has this ledger

protocol-15

Protocol 15 upgrade ledger. Protocol 15. Row protocol-15. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

protocol-18

Protocol 18 upgrade ledger. Protocol 18. Row protocol-18. All rows No entry: no testnet and pubnet twins: a property of the public network’s own upgrade history; only pubnet has this ledger
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

protocol-20

Protocol 20 upgrade ledger. Protocol 20. Row protocol-20. All rows No entry: no testnet and pubnet twins: a property of the public network’s own upgrade history; only pubnet has these ledgers
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

protocol-23

Protocol 23 upgrade ledger. Protocol 23. Row protocol-23. All rows No entry: no testnet and pubnet twins: a property of the public network’s own upgrade history; only pubnet has these ledgers

protocol-24

Protocol 24 upgrade ledger. Protocol 24. Row protocol-24. All rows No entry: no testnet and pubnet twins: the pubnet upgrade ledger itself; the local-chain demonstration is under cap-0076

protocol-29

Protocol 29 (no CAPs). Protocol 29. Row protocol-29. All rows

ledger

Ledger-level shapes. Row ledger. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

classic

Classic shapes without a CAP. Row classic. All rows
Arms that only one or two corpus entries hold, and how often the scan of pubnet ledgers 2 to 65,505,999 saw each (how these are chosen).

xdr-strict

XDR strictness. Row xdr-strict. All rows No entry: no testnet and pubnet twins: passphrase-independent encoding vectors; there is nothing to emit on two networks

limits

Reader limits. Row limits. All rows No entry: no testnet and pubnet twins: bare XDR values with no ledger around them, so passphrase-independent; ledgers that carry deep values are in the cap-0046 row 2 more entries of this row, at a nesting-depth limit, are listed in corpus/COVERAGE.md.

strkey

SEP-23 strkeys. Row strkey. All rows No entry: no testnet and pubnet twins: strkey text vectors are passphrase-independent

xdr-json

XDR-JSON (SEP-51). Row xdr-json. All rows No entry: no testnet and pubnet twins: a passphrase-independent decoding vector

How prevalence is joined

The prevalence tables count union arms. A count per entry, from the scan’s token table, is on Mainnet scan; it is the count of the entry’s rarest token, not of the arm. For the arms, this page uses one join that is exact:
  1. xdrcheck-corpus coverage --arms --report FILE walks every committed input, the entries first, and lists per union arm the first three inputs that hold it.
  2. An arm is kept when that list is complete for entries (fewer than three labels, or a label that is not an entry), so the entries named are all the entries that hold it.
  3. The arm Type::ARM is the scan’s type-level token T|arm|Type::ARM; xdrcheck-corpus scan locate --summary gives its occurrences and first and last ledger, and --limit 3 its first three transactions.
This gives 147 arms, 120 of them seen on pubnet. An arm the scan’s token census does not hold is listed as such only when another arm of its union is in the census. An arm that three or more entries hold, or an entry whose pattern is not an arm (a value, a length, an ordering), has no prevalence row.

What the counts do not say

  • The scan reads the public data lake. Its older ledgers were regenerated by a recent stellar-core, so their transaction meta is v4: an arm that only older meta versions carry (such as TransactionMeta::v2) is not in the census even where the network once produced it.
  • Counts cover pubnet ledgers 2 to 65,505,999 only; Protocol 30 shapes are after that range.
  • Counts are occurrences of the arm in the scanned ledgers; examples are the first transactions in ledger order (ledger:tx, tx in apply order). An arm’s count says nothing about any SDK or indexer result.
The data is corpus/index.json, corpus/coverage.json (the same rows as corpus/COVERAGE.md) and apps/docs/data/cap-prevalence.json, refreshed by apps/docs/scripts/cap_prevalence.py from a locate index. The scan’s integrity check and census: Mainnet scan. 3 entries at a nesting-depth limit are listed in corpus/COVERAGE.md rather than here.
  • The corpus: what an entry is and how its bytes are made
  • xdrcheck find --cap N lists the captured ledgers of a CAP: find