Every captured commit receives deterministic security triage and a separate communication-quality score. Security candidates and broader second-pass signals receive full-patch Ollama analysis.
Message quality measures whether a commit identifies its scope, purpose, rationale, testing, and supporting references. It does not change the security-severity score.
This commit reorganizes existing hardware design files for the original Jade device into a new subfolder and adds the open-source hardware design files for a new product, the Jade Plus. It does not change any software, firmware, or device …
This commit fixes a boot failure on the M5CoreS3 hardware. A recent update to the underlying Espressif SDK (ESP-IDF v5.5) changed the rules for deleting an I2C bus: it now refuses to delete a bus that still has devices attached. The power …
This commit fixes a bug where the touchscreen driver failed to clean up internal handles when shutting down. On newer ESP-IDF firmware, this cleanup failure caused the device to panic and freeze during boot and when using the camera or QR …
Device freeze/DoS-like symptom on affected hardwareMemory leak fixedNo input validation, authentication, or cryptographic weakness visible
This commit adds a new command-line command called 'selfcheck' to the Jade CLI tool. It simply calls an existing device function and prints the result. There is no indication of a security fix or vulnerability being patched.
This commit changes how Blockstream Jade handles the 'genesis blockhash' when signing Liquid transactions. Previously, the device only accepted the network's built-in genesis hash. Now it lets users provide a custom genesis hash for test n…
Adds input validation for caller-supplied genesis blockhashPrevents mainnet genesis hash from being used on Liquid test networksRestricts genesis_hash parameter to Liquid networks only
This commit adds compiler annotations (WARN_UNUSED_RESULT) to functions that return success/failure codes, so the compiler will warn if a caller ignores the result. It also fixes a few places where return values were being ignored, replaci…
Functions returning error/success codes annotated with WARN_UNUSED_RESULTCallers now check return values of rpc_get_bip32_path() and rpc_get_boolean()Use of default-value helpers (rpc_get_boolean_or) to ensure predictable behavior on missing fields
This commit simply renames an internal helper function from `rpc_get_uint64_t` to `rpc_get_uint64` and updates every place that calls it. There are no changes to what the function does, how it checks data, or how secure it is. It is a rout…
This commit is a simple rename of two internal helper functions from `rpc_get_boolean`/`rpc_get_boolean_or` to `rpc_get_bool`/`rpc_get_bool_or`, updating all call sites accordingly. There is no change to what the code does, no bug fix, and…
This commit is a straightforward code cleanup that replaces verbose 'read this optional value, or leave the default' patterns with shorter helper functions that do the same thing. There is no change in behavior and no security issue visibl…
This commit tightens how Blockstream Jade handles the 'precision' value in asset contracts for Liquid assets. Previously, a very large precision value could be accepted and later used to read past the end of an internal lookup table (POW_1…
Out-of-bounds read mitigation: bounds-checking of user-controlled precision before indexing fixed-size POW_10 arrayInteger truncation safety: explicit cast of validated precision to uint8_t after range checkCompile-time invariant enforcement: JADE_STATIC_ASSERT ties array size to ASSET_PRECISION_MAX
This commit simply updates the SHA256 digest of the 'blockstream/jade_builder' Docker image used for building and testing Jade in four configuration files. There is no code change, no disclosed security fix, and no evidence in the commit o…
This commit is a large feature/refactor patch for Blockstream Jade's libjade (a desktop emulator of the Jade hardware wallet). It adds RPC endpoints that let a host application read the emulated device's screen, push fake button presses, r…
New RPC endpoints allow external read/write of emulated NVS storageNew RPC endpoints allow external screen capture and synthetic input injectionNew RPC endpoints allow external camera frame injection
This commit fixes a user-interface navigation quirk on the M5StickC Plus 2 hardware for Blockstream Jade. When entering a PIN, the up/down button directions were not inverted to match the rest of the device's menus, making number entry con…
This commit is a preventive build-hardening change. It adds compiler annotations (WARN_UNUSED_RESULT) to many functions that return success/failure codes, so the compiler will warn if a caller ignores the result. The goal is to catch futur…
Compiler annotation added to enforce checking of error-returning functionsNo specific vulnerability fixed in the diffNo functional code changes or call-site fixes included
This commit adds a safety check to Blockstream Jade's 'sign identity' feature. Before the change, the device would accept any identity string a connected app sent. Now it rejects identities that contain non-printable or unusual characters.…
Input validation added to a signing/identity processNew length-bounded string helper reduces reliance on null-terminationRejection of non-printable characters in user-presented identity string
This commit tightens error handling when decoding the 'issuer' field of an OTP (one-time password) setup URI. Previously, a malformed issuer string could fail to decode silently; now the device shows an error and cancels the operation inst…
Input validation added for OTP URI issuer fieldSilent failure on malformed URL-decoding replaced with explicit error returnDebug assertion added for label/issuer URL-decoding in UI display path
This commit fixes a small but meaningful bug in how Jade, a hardware wallet, reads key paths from Bitcoin PSBT data. The code was passing the byte size of a buffer when it should have passed the number of 32-bit entries. Depending on how t…
Buffer-size semantic mismatch in security-critical PSBT parsingUse of sizeof() on uint32_t array where element count is expectedCode path involved in 2of3 multisig recovery key identification
This commit fixes a logic gap in how Blockstream Jade checks cryptographic commitments for transaction outputs during signing. Previously, the code treated a missing commitment the same as an invalid one, which could cause the device to re…
Inconsistent commitment validation between signing code pathsPossible transaction signing failure or acceptance of unvalidated output commitmentsLogic alignment with existing sign_tx commitment handling
This commit changes the declared return type of a transaction-signing helper function from a transaction-type code (TxType_t) to a simple success/fail boolean (bool). The function body already appears to return true/false, so the change fi…
Return-type mismatch between function declaration/definition and headerFunction is in transaction-signing utility code (sign_utils.c/h)Function handles CBOR transaction parameters and writes output transaction metadata
This commit tightens bounds checks in URL encoding/decoding functions used in Blockstream Jade, a hardware wallet. The changes fix potential off-by-one/out-of-bounds memory reads and writes when handling percent-encoded characters. Because…
Bounds-check rewrite in URL decoder to prevent out-of-bounds reads before percent-decodingCast to unsigned char moved into isxdigit() to avoid undefined behavior with signed char inputsDestination bounds checks in URL encoder changed from pointer-subtraction form to remaining-space form, mitigating possible underflow/wrap
Expand any commit for its author, full message, clarity score, changed files, triage signals, analysis, and source link.
AI review queuedsign_tx: move the common ui confirmation decls to their own headerby Jon Griffiths · fbc09a18 · Sep 4, 2025 · 5 filesMessage 50 · ThinInformational 15Details
Commit message · Jon Griffiths
sign_tx: move the common ui confirmation decls to their own header
50/100 · ThinMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component! No meaningful explanatory body
Why it was queued
signing or wallet pathsecond-pass: security-sensitive path
AI analysis · Informational 15/100
This commit is a straightforward code cleanup: it moves several on-screen confirmation function declarations from individual source files into a shared header file. There is no change to what the code actually does, no bug fix, and no security-related behavior change.
Security candidatesign_tx: add sign_utils.h to declare the types and functions used by signingby Jon Griffiths · afb052a6 · Sep 4, 2025 · 8 filesMessage 50 · ThinInformational 15Details
Commit message · Jon Griffiths
sign_tx: add sign_utils.h to declare the types and functions used by signing
50/100 · ThinMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component! No meaningful explanatory body
Why it was queued
signing boundarysigning or wallet path
AI analysis · Informational 15/100
This commit is a straightforward code cleanup: it moves type and function declarations that were scattered across several files into a single new header file called sign_utils.h. There is no change to what the code actually does, only to how the source files are organized. It does not fix or introduce any security issue.
AI review queuedsign_tx: move btc fee confirmation impl to sign_utilsby Jon Griffiths · f0896818 · Sep 4, 2025 · 2 filesMessage 50 · ThinInformational 15Details
Commit message · Jon Griffiths
sign_tx: move btc fee confirmation impl to sign_utils
50/100 · ThinMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component! No meaningful explanatory body
Why it was queued
signing or wallet pathsecond-pass: security-sensitive path
AI analysis · Informational 15/100
This commit simply moves an existing Bitcoin transaction fee confirmation function from one source file to another. The code logic is copied verbatim with no functional changes, so it does not fix, introduce, or change any security behavior.
Security candidatesign_tx: use explicit_fee for consistency with psbt signingby Jon Griffiths · 20cde04a · Sep 4, 2025 · 1 fileMessage 50 · ThinInformational 15Details
Commit message · Jon Griffiths
sign_tx: use explicit_fee for consistency with psbt signing
50/100 · ThinMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component! No meaningful explanatory body
Why it was queued
signing boundarysigning or wallet path
AI analysis · Informational 15/100
This commit simply renames a local variable from `liquid_fees` to `explicit_fee` inside the transaction-signing code. The change is purely cosmetic and does not alter any logic, calculations, or security behavior. It is described by the author as making the code consistent with PSBT signing code elsewhere.
This means we don't have to find the fee ourselves, and ensures that the checks we perform (e.g. duplicate fees) are consistent whether we are signing a tx or psbt.
68/100 · AdequateMessage clarity
✓ Descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context
Why it was queued
signing boundarysigning or wallet path
AI analysis · Low 44/100
This commit refactors how the Blockstream Jade hardware wallet checks transaction fees when signing a PSBT. Previously, the fee was calculated and validated inside the PSBT-specific code. Now it uses a shared validation routine that is also used for regular transaction signing. The goal is consistency: the same duplicate-fee and fee-validation checks apply regardless of whether the user is signing a raw transaction or a PSBT. The commit removes several local fee checks (such as rejecting blinded fee outputs or unexpected fee asset IDs) and relies on the shared function instead. This is a code-quality and consistency improvement, not a clear-cut security fix, but it touches security-critical fee logic.
Security candidatesign_tx: rename signing cbor getters to match the other params() callsby Jon Griffiths · 25b1dd6a · Sep 4, 2025 · 3 filesMessage 50 · ThinInformational 15Details
Commit message · Jon Griffiths
sign_tx: rename signing cbor getters to match the other params() calls
50/100 · ThinMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component! No meaningful explanatory body
Why it was queued
signing boundarysigning or wallet path
AI analysis · Informational 15/100
This commit is a simple code cleanup: it renames several internal helper functions from names starting with 'rpc_get_' to names starting with 'params_' so they match the naming style used elsewhere in the project. It also makes two helpers 'static' (private to their file) and fixes a typo in a comment. There is no change to behavior, no bug fix, and no security impact.
Security candidatesign_tx: split out signing tx validation, add more checksby Jon Griffiths · 1705ca93 · Sep 4, 2025 · 3 filesMessage 65 · AdequateModerate 57Details
Commit message · Jon Griffiths
sign_tx: split out signing tx validation, add more checks
Also move the explicit fee fetching/validation into the split out function.
65/100 · AdequateMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component✓ Provides an explanatory body
Why it was queued
signing boundarysigning or wallet path
AI analysis · Moderate 57/100
This commit reorganizes how Blockstream Jade validates transaction data before signing. It moves several Liquid-network checks earlier in the process and adds new rules: a transaction output with no recipient script (typically a network fee) must now be unblinded, non-zero, use the network's official policy asset, and there can only be one such fee output. The change also prevents a transaction from being silently treated as a different network type (Bitcoin vs. Liquid). These are defensive hardening measures rather than a fix for a confirmed active attack.
✓ Descriptive subject! No meaningful explanatory body! Opaque security-relevant change
Why it was queued
signing boundarysigning or wallet path
AI analysis · Low 40/100
This commit changes the Blockstream Jade hardware wallet firmware so it can now sign Liquid/Elements PSET transactions instead of rejecting them. It also adds a check that the PSBT type matches the selected network (Bitcoin vs. Liquid). A new FIXME comment asks whether Bitcoin signing should only accept 'non-witness utxo' data, and descriptor-based wallet signing is disabled for Elements. The change is a feature addition with some defensive tightening, not a clear vulnerability fix, but it touches security-critical signing code.
Security candidatepset: allow sign_psbt to be called for liquidby Jamie C. Driver · 3ab9dcd2 · Sep 4, 2025 · 4 filesMessage 70 · AdequateLow 35Details
Commit message · Jamie C. Driver
pset: allow sign_psbt to be called for liquid
This initial commit just enables support with tests coming later.
70/100 · AdequateMessage clarity
✓ Descriptive subject✓ Names a concrete action or component✓ Provides an explanatory body✓ Mentions testing or verification
Why it was queued
signing boundarysigning or wallet path
AI analysis · Low 35/100
This commit enables the Jade hardware wallet to sign Liquid network transactions using the existing sign_psbt command. Previously, this command was rejected for Liquid. The change mainly removes that rejection, skips Bitcoin-only fee math on Liquid, and updates test fixtures. It is a feature-enablement patch rather than a clear security fix, and the commit message says full tests will follow later.
AI review queuedpset: display liquid outputs for user confirmationby Jamie C. Driver · e713f138 · Sep 4, 2025 · 1 fileMessage 65 · AdequateLow 28Details
Commit message · Jamie C. Driver
pset: display liquid outputs for user confirmation
Note this assumes simple payment for now - TODO: look for more complex txs
65/100 · AdequateMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component✓ Provides an explanatory body
Why it was queued
signing or wallet pathsecond-pass: security-sensitive path
AI analysis · Low 28/100
This commit adds user confirmation screens for Liquid (Elements) transactions in the Blockstream Jade hardware wallet. Previously, Liquid PSBT signing likely did not show the user the outputs and fee before signing, or used Bitcoin-only confirmation logic that did not properly handle Liquid assets. The change makes the device ask the user to confirm what is being spent and what fee is being paid before producing a signature. This is a security-hardening improvement rather than an obvious vulnerability fix, but the absence of such confirmation could have allowed malicious software to trick a user into signing an unwanted Liquid transaction.
Lower-prioritytests: update a missed liquid test case descriptionby Jon Griffiths · 83e1626d · Sep 4, 2025 · 1 fileMessage 60 · AdequateTriage 0Details
Commit message · Jon Griffiths
tests: update a missed liquid test case description
60/100 · AdequateMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component✓ Mentions testing or verification! No meaningful explanatory body
AI review queuedpset: examine outputs looking for unblinded liquid databy Jamie C. Driver · 7db69311 · Sep 4, 2025 · 1 fileMessage 50 · ThinModerate 57Details
Commit message · Jamie C. Driver
pset: examine outputs looking for unblinded liquid data
50/100 · ThinMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component! No meaningful explanatory body
Why it was queued
signing or wallet pathsecond-pass: security-sensitive path
AI analysis · Moderate 57/100
This commit changes how the Blockstream Jade hardware wallet examines transaction outputs before signing a Liquid (Elements-based) PSBT. It now inspects outputs for confidential/unblinded asset and value data, detects scriptless fee outputs, validates that fee outputs use the correct policy asset, and rejects blinded fee outputs. Previously, the code only tried to identify 'change' outputs on Bitcoin-like transactions. The change appears to be a security hardening/fix for Liquid transactions, but the commit message does not explicitly call it a security fix.
AI review queuedpset: parse pset bytes using temporary task with larger stackby Jamie C. Driver · 5b4e61da · Sep 4, 2025 · 1 fileMessage 45 · ThinModerate 57Details
Commit message · Jamie C. Driver
pset: parse pset bytes using temporary task with larger stack
Unblinding and verifying explicit proofs requires a ~54Kb stack. See also: e1b8c18bf7e38137e4b95b683833ccb20b7ce761
45/100 · ThinMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component✓ Provides an explanatory body! Contains work-in-progress language
Why it was queued
signing or wallet pathsecond-pass: security-sensitive path
AI analysis · Moderate 57/100
This commit changes how Blockstream Jade hardware wallets parse PSET (a variant of Bitcoin's PSBT format used on the Liquid sidechain). Previously, parsing PSET data could crash the device because verifying confidential transaction proofs needs about 54 KB of temporary memory—more than the normal task stack provides. The fix runs PSET parsing in a temporary task with a larger 54 KB stack, while ordinary PSBT parsing stays unchanged. The commit also adds explicit recognition of the PSET magic bytes, whereas before Jade only checked for PSBT magic bytes and noted that PSET was not yet supported.