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 only updates documentation. It adds new version entries to the manual page for the lightning-downgrade tool and mentions the tool in the upgrade guide. No code, database logic, or security behavior is changed.
This is a packaging/build fix, not a security vulnerability. The `lightning-downgrade` utility was being compiled but accidentally left out of the list of programs installed by `make install`. As a result, official release archives contain…
This commit fixes a database upgrade bug in Core Lightning. Wallets created with version v26.06.10 were missing two new tables (our_outputs and our_txs) because they skipped the migrations that create them. When users later upgraded, a fol…
Database migration failure could prevent node startup after upgradeFixes missing table creation before data backfillAdds regression test for upgrade path from v26.06.10
This commit is a routine version bump from v26.06.10 to v26.12rc1 for Core Lightning. It only updates version strings in packaging files, a changelog, and a lockfile. There are no code changes that introduce, fix, or modify any security-se…
This commit fixes a database bug in Core Lightning where failure details for a payment attempt were being overwritten by later retry attempts that shared the same payment identifier fragment. The fix makes the failure record match on an ad…
Data integrity bug causing incorrect error reporting across payment retry groupsNo memory safety, authentication, authorization, or cryptographic issues presentNo external input is trusted or validated differently
This commit fixes a bug in Core Lightning's Rust plugin library where a plugin could crash with a panic if it tried to log a message after the internal log-writing task had already stopped, such as during shutdown. The fix makes the plugin…
panic on send failure replaced with stderr fallbacklog message no longer silently dropped on writer task exitunit test added for dropped-receiver scenario
This commit only adds documentation to the manual page for an existing command-line tool (hsmtool). It explains how to recover funds from old Lightning payment commitments (HTLCs) after losing channel state. There are no code changes, no n…
This commit only changes a test file to make a flaky test more reliable. It does not modify any production code, so it has no security impact on real users.
This commit only updates documentation to explain that the `listtransactions` command reports a block height and transaction index of 0 when a transaction is not currently in a block (for example, because it hasn't been mined yet or was re…
This is a small bug fix in a Python helper library used by Core Lightning. The Millisatoshi.to_approx_str() method previously crashed with a math error when converting a zero amount to a human-readable string. The patch makes zero return "…
No security-relevant signals presentFix is a routine correctness/robustness improvement
This commit is a trivial formatting cleanup (removing an extra blank line and adding one) in a test file. It does not change any code behavior, fix any bug, or alter any security-relevant logic.
This update fixes a crash in two wallet commands, addpsbtoutput and addpsbtinput. These commands let users add extra transaction outputs or inputs to a PSBT (a Bitcoin transaction planning file). If a user supplied a PSBT whose version 2 '…
Daemon crash (assertion failure) on crafted/strict PSBT inputMissing validation of PSBT v2 modifiable flags before mutationUser-controlled input directly passed to libwally without precondition check
This commit only adds a new test case that demonstrates two Core Lightning commands, addpsbtoutput and addpsbtinput, crash with an internal assertion when given a v2 PSBT that does not allow adding inputs or outputs. The test is marked as …
Daemon abort due to failed assertion in PSBT helper functionsRPC commands addpsbtoutput and addpsbtinput lack validation of PSBT modifiability flags before calling libwallyPotential denial-of-service vector: any caller able to submit a crafted v2 PSBT can crash lightningd
This commit fixes a minor bug in Core Lightning's Python client library where calling a pretty-printing method on a zero-amount payment would crash with a ValueError. It is a routine correctness fix, not a security vulnerability. There is …
This commit adds missing safety checks in Core Lightning when a peer sends signatures for opening or splicing a dual-funded Lightning channel. Before the patch, the software accepted peer-provided transaction witnesses without verifying th…
Missing input validation on peer-provided witness dataBOLT-2 compliance fix for tx_signaturesRejection of empty witness stacks
This fix prevents a user-issued 'abort channels' command from incorrectly failing a Lightning channel when a splice transaction has already been signed. Previously, the code would pass the latest in-flight splice to a routine that aborts i…
Denial-of-service/availability impact: user-initiated abort could inadvertently fail a funded channelProtocol compliance fix: avoids sending tx_abort after tx_signatures, which BOLT 2 forbidsState-handling bug in splice negotiation lifecycle
This update fixes a bug where a specific type of Lightning channel close could crash the node. It only happens when both parties cooperatively close a dual-funded channel that has not yet been confirmed and has gone through a funding bump …
NULL pointer dereference leading to daemon crash (denial of service)Crash occurs in cooperative-close path for unconfirmed dual-funded/RBF channelsFix uses canonical channel funding key instead of splice-only field
This commit fixes a security bug in Core Lightning's gRPC plugin (cln-grpc). Previously, a malformed request from a gRPC client could crash the plugin: bad public keys, hashes, secrets, channel IDs, or enum values triggered Rust 'unwrap()'…
Denial-of-service via malformed gRPC request causing tonic worker task panicDenial-of-service via std::process::exit(1) in ProofField conversion on missing oneofSilent integer truncation of u16/u8 fields via 'as' casts
This commit removes the protobuf package from the Nix build environment because it is no longer needed. A previous change made the cln-grpc component use its own bundled protoc compiler instead of relying on the system one. The Python bind…
This commit adds a new automated check to the project's build process that compares gRPC protocol definitions against the last release to catch accidental breaking changes early. It does not change any runtime code, fix a bug, or introduce…
Expand any commit for its author, full message, clarity score, changed files, triage signals, analysis, and source link.
Security candidatedoc: refresh the release verification examples from v26.06.9by Níckolas Goline · 1112b47b · Oct 22, 2026 · 1 fileMessage 83 · StrongInformational 15Details
Commit message · Níckolas Goline
doc: refresh the release verification examples from v26.06.9
The gpg example showed 2020 signatures from keys that no longer sign releases, and the sha256sum example listed v24.11 files. Replace both with the output for v26.06.9, covering all three release manifests (amd64 and source, arm64, armv7) and every file they list.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
83/100 · StrongMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context✓ Mentions testing or verification
This commit only updates the user documentation that shows how to verify Core Lightning release files. It replaces outdated example filenames, GPG signatures, and SHA256 checksums with current ones for version v26.06.9. No program code was changed, and there is no security vulnerability in the commit itself.
The signing subkey in the repo copy expired on 2026-06-30, so the v26.06.9 signature could not be verified from it. Re-export the key with the subkey extended to 2027-06-30.
Changelog-None Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
73/100 · AdequateMessage clarity
✓ Descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context✓ Names security-relevant behavior explicitly
Why it was queued
signing boundarydocumentation-only discount
AI analysis · Informational 15/100
This commit simply refreshes an expired PGP signing key stored in the repository so that future release signatures can be verified. It is a routine key update with no code or security vulnerability fix.
Security candidateconnectd: never block the io loop on DNS resolutionby cdecker · 71572cae · Oct 21, 2026 · 7 filesMessage 73 · AdequateHigh 71Details
Commit message · cdecker
connectd: never block the io loop on DNS resolution
Resolving a DNS address announced in a node_announcement used to call getaddrinfo() directly, on connectd's single io thread. A slow or unresponsive nameserver held up every peer on the machine for the duration of the lookup: ~10s per query with glibc defaults, enough to trip the 60s ping-liveness check and disconnect all peers.
getaddrinfo() now runs in a forked child which writes the results back over a pipe; connectd reads the pipe from its io loop and resumes the connect attempt from a loop callback, so the io thread never waits on resolution. Keeping libc for the actual lookup preserves /etc/hosts, search domains, systemd-resolved and AI_ADDRCONFIG ordering exactly, and keeps DNS packet parsing out of connectd itself.
A 30-second cap covers each lookup end to end (a wedged child is killed), and at most ten resolvers run at once: a gossip-driven reconnect storm queues the excess FIFO rather than forking without bound. Before resolving, the child closes every fd it inherited: a wedged child must not keep peer connections alive, nor pin EOF on another lookup's pipe with its copy of that write end.
The connect RPC resolved a host name the same way, with a blocking getaddrinfo() in lightningd's main loop, and ignored --disable-dns. It now hands the unresolved name to connectd as a DNS address, which goes through the same asynchronous resolver and honours --disable-dns (unless a proxy resolves names for us).
Reported-by: Vincenzo Palazzo Changelog-Fixed: connectd: a stalling DNS name in a peer's announced address no longer freezes all peer traffic; resolution is now done asynchronously. Changelog-Changed: `connect` with a host name now resolves it asynchronously in connectd, and fails with `--disable-dns`.
73/100 · AdequateMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context
Why it was queued
access control
AI analysis · High 71/100
This commit fixes a denial-of-service weakness in Core Lightning's connection daemon. Previously, when the node needed to look up a peer's internet address from a domain name, it did that lookup directly in its main networking thread. A slow or malicious DNS server could freeze all peer traffic for about 10 seconds per lookup, potentially disconnecting every peer. The fix moves DNS lookups into separate child processes and reads the results back without blocking the main loop. It also makes the `connect` RPC command use the same non-blocking path and respect the `--disable-dns` setting.
AI review queuedci: replace broken katyo/publish-crates with a native cargo publish stepby daywalker90 · 2646184f · Oct 20, 2026 · 1 fileMessage 85 · StrongInformational 20Details
Commit message · daywalker90
ci: replace broken katyo/publish-crates with a native cargo publish step
The katyo/publish-crates action no longer publishes anything: the Release Rust workflow ran on the push that bumped lsps-types to 0.7.0 and did not publish it.
The new step publishes each crate only when its Cargo.toml version is newer than the highest version on crates.io, in dependency order (cln-rpc, cln-grpc, cln-plugin, lsps-types).
Cap the job at 30 minutes and give the crates.io index request a 30 second timeout so a hung network call cannot stall the release.
85/100 · StrongMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component✓ Uses a recognizable type or scope✓ Provides detailed explanatory context
This commit updates the project's automated release workflow for publishing Rust crates to crates.io. It replaces a third-party GitHub Action that had stopped working with a custom shell script that checks whether each crate's local version is newer than what is already published before running the official cargo publish command. The change also adds a 30-minute job timeout and a 30-second network timeout for checking crates.io. There is no direct security vulnerability in the diff, but any change to a release pipeline that handles a secret publishing token deserves a routine security review.
Security candidatetests: mock LSPS2 service charges the opening fee from its paramsby Peter Neuroth · d315f9f9 · Oct 20, 2026 · 2 filesMessage 95 · StrongInformational 15Details
Commit message · Peter Neuroth
tests: mock LSPS2 service charges the opening fee from its params
Until now the mock deducted a fixed `htlc.fee_msat` (default 0) and did none of the fee checks bLIP-52 asks of an LSP. Compute the fee instead:
- `lsps2.buy` with a `payment_size_msat` fails with 202 if the fee is not below the payment size and with 203 if computing it overflows. - In no-MPP+var-invoice mode the fee comes from the first HTLC, which is failed with unknown_next_peer if it is outside the payment size limits, if the fee overflows or if nothing would be left to forward. - The honest deduction is the computed fee; `htlc.fee_msat` still overrides it.
Failed HTLCs now record why they were failed.
95/100 · StrongMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context✓ Explains rationale or failure mode✓ Mentions testing or verification
Why it was queued
memory safety
AI analysis · Informational 15/100
This commit only changes test code. It updates a mock Lightning Service Provider (LSP) used in Core Lightning's test suite so that its simulated 'opening fee' follows the real bLIP-52 formula instead of being a hardcoded test value. No production wallet, node, or protocol code is touched, so it cannot directly affect real users or funds.
Security candidatetests: add configurable misbehavior to the mock LSPS2 serviceby Peter Neuroth · b323f375 · Oct 19, 2026 · 2 filesMessage 100 · StrongInformational 15Details
Commit message · Peter Neuroth
tests: add configurable misbehavior to the mock LSPS2 service
Each phase of the mock gets a table of named behaviors, each with its own parameters, so that every attack in the security matrices is one config entry:
- transport: drop, wrong_id, duplicate (optionally with a changed second result), replace (send a raw payload), add_fields and error, optionally limited to some methods. - buy: the jit_channel_scid is a fixture too, next to lsp_cltv_expiry_delta and client_trusts_lsp. - open: never, or twice. - htlc: deduct (per-part deduction and extra_fee, or no TLV), forward_subset (fail the other parts) and deduct_again (charge the fee again after the session).
lsps2-mock-send sends a message outside of any request and lsps2-mock-open opens a zero-conf channel outside of any session. Unknown behaviors and parameters, and missing required ones, are rejected when the config is set.
100/100 · StrongMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context✓ Explains rationale or failure mode✓ Mentions testing or verification✓ Names security-relevant behavior explicitly
Why it was queued
explicit security language
AI analysis · Informational 15/100
This commit only adds new testing capabilities to a mock LSPS2 service used in Core Lightning's test suite. It lets developers simulate misbehaving LSPs (e.g., dropping messages, sending wrong IDs, opening channels twice, deducting fees incorrectly) so they can verify that the real client handles those cases safely. The change does not alter production code, wallets, or node behavior, and it does not fix or introduce any real vulnerability.
Security candidatetests: fix races in test_channel_lease_unilat_closesby daywalker90 · d3e6bebe · Oct 14, 2026 · 1 fileMessage 95 · StrongInformational 15Details
Commit message · daywalker90
tests: fix races in test_channel_lease_unilat_closes
l3's commitment used a blockheight 2 blocks behind the chain, because l2 was stopped before l3 had saved a commitment with the new height. A second run flaked with:
E pyln.client.lightning.RpcError: RPC call failed: method: fundchannel ... error: {'code': 301, 'message': 'Could not afford 506268sat using all 0 available UTXOs: 506696sat short'}
l2 broadcast the l1 funding tx, but did not record its change output until after the l2->l3 fundchannel with minconf=0 had already failed.
Wait for l3 to log the commitsig carrying the new blockheight before stopping l2, and wait for l2 to see the funding change before the minconf=0 fundchannel.
Changelog-None
95/100 · StrongMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context✓ Explains rationale or failure mode✓ Mentions testing or verification
Why it was queued
signing boundary
AI analysis · Informational 15/100
This commit fixes timing problems in a single automated test. It adds waits so the test nodes are fully synchronized before the test proceeds. There is no change to production code, no security bug, and no real-world attack possibility.
Security candidatedoc: fix references to manpages that don't existby mikeua · 33fda2d0 · Oct 13, 2026 · 11 filesMessage 68 · AdequateInformational 15Details
Commit message · mikeua
doc: fix references to manpages that don't exist
Several schema descriptions and see_also lists name manpages that are never generated, so the links in the rendered docs are dead:
contrib/msggen/msggen/schema.json is updated to match.
Changelog-None
68/100 · AdequateMessage clarity
✓ Descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context
Why it was queued
authentication path
AI analysis · Informational 15/100
This commit only fixes broken documentation links. It corrects the names of referenced manual pages in JSON schema files so that generated documentation points to pages that actually exist. There are no code, configuration, or behavior changes, and no security implications.
AI review queuedlsps-types: echo valid_until verbatim and send the LSPS0 datetime formatby Peter Neuroth · 2d187383 · Oct 11, 2026 · 3 filesMessage 85 · StrongLow 35Details
Commit message · Peter Neuroth
lsps-types: echo valid_until verbatim and send the LSPS0 datetime format
bLIP-52, "2. Request JIT Channel":
Clients MUST copy it verbatim from an entry of `opening_fee_params_menu` from a result of a `lsps2.get_info` call.
bLIP-50, LSPS0.datetime:
Particular points of time in the modern era (a "datetime") MUST be encoded as a JSON string containing the [ISO 8601][] format `"YYYY-MM-DDThh:mm:ss.uuuZ"`. These are always in the UTC timezone.
`DateTime` was serialized by chrono, which writes as many fractional digits as needed: a spec-conforming "2023-02-23T08:47:30.000Z" from an LSP came back from the client as "2023-02-23T08:47:30Z", and "...30.5112Z" as "...30.511200Z". That only worked because the in-tree service checks its promise against the parsed value; an LSP that checks the string it sent would reject the client. The service itself sent whole seconds without the milliseconds.
`DateTime` now keeps the string it was deserialized from and serializes it back unchanged. A datetime built from a chrono value is truncated to milliseconds and serialized as `YYYY-MM-DDThh:mm:ss.uuuZ`. Equality, ordering and hashing still only compare the point in time. Parsing is unchanged. `DateTime` is no longer `Copy`.
The service builds the `valid_until` it sends from the policy plugin's value this way and computes the promise over the truncated value, so the promise still verifies on what the client echoes.
Behavior change: the client echoes `valid_until` exactly as received; the service sends `valid_until` with millisecond precision in the LSPS0 format.
Changelog-Fixed: lsps-plugin: the LSPS2 client now copies `opening_fee_params` verbatim (including `valid_until`), and the service sends `valid_until` in the LSPS0 datetime format.
85/100 · StrongMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context✓ Explains rationale or failure mode
Why it was queued
second-pass: broader security terminology
AI analysis · Low 35/100
This commit fixes a protocol-compatibility bug in Core Lightning's LSPS (Lightning Service Provider Specification) support. When a client received an opening-fee offer from a service provider, it was accidentally changing the exact timestamp text before echoing it back. Some strict providers would then reject the client's reply because the timestamp no longer matched. The fix makes the client preserve the original timestamp text exactly, and makes the service send timestamps in the precise format the spec requires. There is no direct theft-of-funds or remote-code-execution vulnerability here, but the bug could cause legitimate JIT-channel requests to fail.
Security candidatelsps-plugin: use lsps-types for the protocol typesby Peter Neuroth · 8ecf4b62 · Oct 11, 2026 · 12 filesMessage 78 · AdequateInformational 24Details
Commit message · Peter Neuroth
lsps-plugin: use lsps-types for the protocol types
Replace the plugin's copies of the JSON-RPC, LSPS0 and LSPS2 protocol types and of the opening fee computation with the ones from lsps-types, so client and service share one implementation. The plugin's proto modules re-export lsps-types for now; the imports are switched in the next commit.
Service-only code stays in proto/lsps2.rs: the policy plugin API, the datastore entry, JIT scid generation, the BOLT4 failure codes and the HMAC promise. The HMAC check moves out of `OpeningFeeParams::validate` into `verify_promise()`, which `lsps2.buy` calls before `OpeningFeeParams::validate()`. The order of the checks and the error codes are unchanged: bad promise or expired params -> 201, fee overflow -> 203, fee not below the payment size -> 202.
Other adaptations, no behavior change: - `into_request()` with a random id becomes an extension trait in the plugin, so lsps-types does not need `rand` and `hex`. - `ShortChannelId` is converted via u64 where it meets cln-rpc (the `htlc_accepted` onion, the client's datastore record). - `DateTime` is converted with `as_chrono()` where the plugin computes with chrono values.
The one behavior change from switching is the promise character check that lsps-types added in "lsps-types: restrict promise to printable ASCII without JSON escapes".
Changelog-None
78/100 · AdequateMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context✓ Names security-relevant behavior explicitly
Why it was queued
memory safetydefensive validation
AI analysis · Informational 24/100
This commit is a code cleanup in Core Lightning's LSPS (Lightning Service Provider Specification) plugin. It removes duplicated protocol type definitions and replaces them with a shared library called lsps-types. The only functional change mentioned is a stricter check on the 'promise' string, limiting it to printable ASCII characters without JSON escape characters. This is a hardening improvement, not an obvious vulnerability fix.
Security candidatelsps-types: test the opening fee computationby Peter Neuroth · 41df2752 · Oct 11, 2026 · 3 filesMessage 83 · StrongInformational 15Details
Commit message · Peter Neuroth
lsps-types: test the opening fee computation
Unit tests for the rounding boundaries (a product of k*10^6 gives a fee of k, k*10^6+1 gives k+1), the min_fee_msat floor, and overflow of the multiplication and of the `+ 999999` addition near u64::MAX. Property tests compare the function against an independent u128 reference and check that it is the ceiling of the proportional part.
bLIP-52 has no test vectors for the fee computation.
Adds proptest as a dev-dependency.
Changelog-None
83/100 · StrongMessage clarity
✓ Descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context✓ Mentions testing or verification✓ Names security-relevant behavior explicitly
Why it was queued
memory safetyfuzzing or regression evidence
AI analysis · Informational 15/100
This commit only adds unit tests and property-based tests for an existing fee-calculation function. It does not change the function's behavior, fix a bug, or alter any production code paths. There is no security issue introduced or resolved by this change.
Security candidatelsps-types: add LSPS2 messages and opening fee paramsby Peter Neuroth · fee9b0e7 · Oct 11, 2026 · 5 filesMessage 78 · AdequateInformational 15Details
Commit message · Peter Neuroth
lsps-types: add LSPS2 messages and opening fee params
Move the LSPS2 messages, `OpeningFeeParams`, `Promise` and `compute_opening_fee` from plugins/lsps-plugin/src/proto/lsps2.rs into lsps-types. Promise and fee computation are unchanged here; they are reworked in their own commits.
`OpeningFeeParams::validate` in the plugin mixed the LSP's HMAC promise check with the checks every LSP must do. Only the latter move: the promise format is the LSP's own business and stays with the service. `validate(now, payment_size)` keeps the order and error mapping of the plugin (expired -> 201, fee overflow -> 203, fee >= payment size -> 202) and takes `now` from the caller. The unused `receivable` argument is dropped. New are `check_limits()` for min <= max payment size, `is_expired_at()`, `check_payment_size()` and `opening_fee()`.
Also add the `unrecognized_or_stale_token` (200) error code from bLIP-52, which was missing.
Service-only types (policy plugin API, datastore entry, JIT scid generation, BOLT4 failure codes) stay in the plugin.
Changelog-None
78/100 · AdequateMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context✓ Names security-relevant behavior explicitly
Why it was queued
memory safetydefensive validation
AI analysis · Informational 15/100
This commit is a routine code reorganization for the Core Lightning project. It moves data structures and helper functions related to the LSPS2 (Just-In-Time channel) protocol from one internal plugin file into a shared library module. It also adds validation helpers and one missing error code. There is no indication this fixes or introduces a security vulnerability; it is a structural refactor with tests.
AI review queuedlsps-types: add empty crateby Peter Neuroth · e5ca96e3 · Oct 11, 2026 · 9 filesMessage 80 · StrongInformational 15Details
Commit message · Peter Neuroth
lsps-types: add empty crate
Add a new top-level crate `lsps-types` for the LSPS0 and LSPS2 protocol types and the LSPS2 opening fee computation, so that LSPS clients and services share one implementation and cannot disagree on amounts.
The crate is wired up like cln-rpc: a workspace member and workspace dependency versioned in lockstep with cln-rpc, cln-grpc and cln-plugin, a Makefile included when building with Rust, and a crates.io publish step. The LSPS plugin binaries list its sources as prerequisites so they rebuild once the plugin depends on it.
Changelog-None
80/100 · StrongMessage clarity
✓ Descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context✓ Explains rationale or failure mode
Why it was queued
second-pass: broader security terminology
AI analysis · Informational 15/100
This commit adds a new, empty Rust code crate called lsps-types to the Core Lightning project. It contains no executable logic, only build and packaging configuration. There is no security issue in this change.
armv7 cannot build natively like arm64: cln-grpc's build script takes protoc from protoc-bin-vendored, which ships no armv7 binary, so the build panics there.
Add tools/repro-build.armv7.sh, which cross-compiles instead. It runs in the amd64 cl-repro-<dist> image, adds armhf from ports.ubuntu.com (same release pocket, updates and security still disabled), and pins the arm-linux-gnueabihf toolchain and armhf libraries the same way the other scripts pin theirs. The Rust plugins build for armv7-unknown-linux-gnueabihf.
The build runs a few armv7 programs (configure tests, tools/headerversions), so the host needs qemu-arm registered with binfmt_misc.
bin-Ubuntu-<dist>-armv7 runs the amd64 image with REPRO_ARCH=armv7, and the tarball lands in release-armv7/.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
83/100 · StrongMessage clarity
✓ Descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context✓ Mentions testing or verification✓ Names security-relevant behavior explicitly
Why it was queued
access control
AI analysis · Informational 14/100
This commit adds a new build script and updates Dockerfiles so that Core Lightning release tarballs can be produced for 32-bit ARM (armv7) devices. It is purely a build-system/infrastructure change for cross-compilation and does not alter any wallet, networking, or cryptographic code. There is no indication it fixes or introduces a security vulnerability.
AI review queuedtools: add sign-release-arm.shby Níckolas Goline · 224130eb · Oct 6, 2026 · 1 fileMessage 68 · AdequateInformational 15Details
Commit message · Níckolas Goline
tools: add sign-release-arm.sh
build-release.sh's sign target cannot be used for the ARM tarballs: it hardcodes `cd release/` rather than honouring RELEASEDIR, and globs the zip, so running it for an ARM build would rewrite the amd64 manifest.
Checksum one architecture's tarballs, from release-arm64/ or release-armv7/, into SHA256SUMS-<version>-<arch> and sign it. arm64 and armv7 keep separate manifests: arm64 builds natively and armv7 is cross-compiled, so they are reproduced and co-signed separately.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
68/100 · AdequateMessage clarity
✓ Descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context
Why it was queued
signing or wallet pathsecond-pass: security-sensitive path
AI analysis · Informational 15/100
This commit adds a new helper script used by Core Lightning release engineers to create and sign checksum files for ARM release tarballs. It does not change any wallet, network, or node code, and there is no indication it fixes or introduces a security vulnerability. It is purely a release tooling addition.
Security candidateCI: never replace files already on the releaseby Níckolas Goline · f918c23f · Oct 6, 2026 · 3 filesMessage 97 · StrongModerate 60Details
Commit message · Níckolas Goline
CI: never replace files already on the release
The release job uploaded release/* with action-gh-release, which replaces any asset of the same name. Re-running it after the release captains had added their signatures to SHA256SUMS-<version>.asc replaced that file with one carrying only the CI signature, so users verifying the release found the maintainers' signatures gone.
Upload with tools/publish-release-assets.sh instead:
- files the release does not have yet are uploaded; - files it already has must be identical, otherwise the job stops before touching anything (the builds disagree, and nothing should be signed); - an .asc it already has gets our signature appended, after normalising the armor so gpg reads every block; one that already carries our signature is left alone, so re-runs are harmless.
The draft release is created with gh when it does not exist yet, with the same title and pre-release flag as before.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
97/100 · StrongMessage clarity
✓ Descriptive subject✓ Names a concrete action or component✓ Uses a recognizable type or scope✓ Provides detailed explanatory context✓ Explains rationale or failure mode✓ Names security-relevant behavior explicitly
Why it was queued
signing boundary
AI analysis · Moderate 60/100
This commit fixes a release-pipeline bug where re-running the automated release job could accidentally wipe out human maintainers' cryptographic signatures from the release's checksum file. Before the fix, the CI tool replaced any existing file with the same name, so if release captains had already added their GPG signatures to SHA256SUMS-<version>.asc, a re-run would overwrite it with a version signed only by the CI bot. Users who later checked the signatures would see only one signature and could not verify that multiple maintainers had approved the release. The fix introduces a new script that only adds missing files, refuses to change files that differ, and appends the CI signature to existing signature files instead of replacing them.
AI review queuedchanneld: bound HTLC additions by our own splice-out amountby jaonoctus · 0b41f819 · Oct 5, 2026 · 1 fileMessage 73 · AdequateModerate 66Details
Commit message · jaonoctus
channeld: bound HTLC additions by our own splice-out amount
While a splice is inflight, every HTLC has to fit on the commitment for each funding candidate, so get_room_above_reserve() charges each side's lowest splice change (lowest_splice_amnt) against its balance. But update_view_from_inflights() set our side's bound from the inflight's new funding total, which is positive, so it stayed at zero and our splice-out was never charged.
We then admitted HTLCs that fit only the pre-splice commitment. Building the post-splice commitment took our balance negative, channel_txs() returned NULL, and channeld dereferenced it and crashed, then did the same on every reconnect:
during the splice, payments were forwarded out through the channel cumulatively they exceeded our post-splice balance (invalid on the post-splice commitment) ... channeld crashed building the post-splice commitment: our balance went negative, the commitment build returned nothing, and channeld dereferenced it every reconnect rebuilds the same commitment, so it loops
Use the inflight's signed local change for our bound, as the remote side already does. Also fix the REMOTE-view assignments, which compared against one side's bound and wrote the other's.
Reported-by: Ken Sedgwick <ken@bonsai.com> Changelog-Fixed: channeld: no longer crash-loops when an HTLC that only the pre-splice commitment can carry is added during a splice-out. Co-Authored-By: Ken Sedgwick <ken@bonsai.com>
(cherry picked from commit 733c42e823bb204e6d91cca4f0450158842404e3)
73/100 · AdequateMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context
Why it was queued
second-pass: broader security terminology
AI analysis · Moderate 66/100
This fix repairs a bug in Core Lightning's channel daemon that could crash and permanently loop-crash a node during a specific kind of Bitcoin channel upgrade called a 'splice-out.' While the splice was in progress, the program failed to subtract the node's outgoing splice amount when checking whether new forwarded payments would fit in the updated channel. It therefore allowed payments that were valid only for the old channel state. When it later tried to build the post-splice commitment transaction, the node's balance went negative, the builder returned nothing, and the daemon dereferenced that null pointer and crashed. On every reconnect it rebuilt the same bad commitment, so the node would crash again in a loop. The patch corrects the accounting so the splice-out amount is properly charged against the balance, and also fixes a related indexing mistake that mixed up local and remote views.
AI review queuedchanneld: fail the channel instead of crashing if we can't build a commitmentby jaonoctus · 77e1f180 · Oct 5, 2026 · 1 fileMessage 85 · StrongModerate 60Details
Commit message · jaonoctus
channeld: fail the channel instead of crashing if we can't build a commitment
channel_txs() returns NULL when it can't build a commitment, for example when a splice amount would take one side's balance below zero:
channel_txs fails to amount_msat_add_sat_s64 ... LOCAL (side_pay + splice_amnt)
Neither send_commit_part() nor handle_peer_commit_sig() checked for that, so channeld dereferenced NULL and crashed. Since every reconnect rebuilds the same commitment, it then crashed again on each one, leaving the channel stuck in a crash loop that only a unilateral close would end.
There's no way to continue on a commitment we can't build, so fail the channel and say why. The next commit stops the case we know about reaching here at all.
Changelog-None
Co-Authored-By: Ken Sedgwick <ken@bonsai.com>
(cherry picked from commit 2594edab350be11c68459e546921e6c0d1d07739)
85/100 · StrongMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context✓ Explains rationale or failure mode
Why it was queued
second-pass: broader security terminology
AI analysis · Moderate 60/100
This update fixes a bug where a part of the Core Lightning node (channeld) would crash and enter an endless crash loop if it couldn't build a commitment transaction, for example due to a splice that would push one side's balance below zero. The fix makes the node fail the channel gracefully instead of crashing. A malicious or buggy peer could potentially trigger this crash repeatedly on reconnection, causing denial of service for that channel.
Security candidatelightningd: use a splice's HTLC sigs once we move onto its commitmentby jaonoctus · a34027c2 · Oct 5, 2026 · 6 filesMessage 85 · StrongModerate 68Details
Commit message · jaonoctus
lightningd: use a splice's HTLC sigs once we move onto its commitment
update_channel_from_inflight() moves channel->last_tx onto a splice's commitment, both when the splice locks and when we adopt a splice which confirmed before splice_locked because its output got spent. But channel->last_htlc_sigs stayed the peer's signatures for the pre-splice commitment until the next commitment update, or a restart, replaced them.
If that commitment is the one we close with first, onchaind gets signatures which can't spend its HTLC outputs, so our HTLC-timeout transactions are invalid and we can't get those HTLCs back. That's the case in a report of a lost balance after a close on an unlocked splice:
the close broadcasts our inflight commitment (it spends the splice funding output, because the splice had confirmed but never locked in) lightningd still treats the pre-splice output as the channel funding, so onchaind is started with the pre-splice funding, balance, and commitment
The sigs go with the commitment, so switch them in the same place: wallet_htlcsigs_confirm_inflight() now returns the sigs it makes current, and update_channel_from_inflight() calls it for a splice instead of handle_peer_splice_locked() doing so on the side.
Reported-by: Ken Sedgwick <ken@bonsai.com> Changelog-Fixed: splicing: closing with a splice's commitment before any later commitment update can now time out or claim the HTLCs on it. Co-Authored-By: Ken Sedgwick <ken@bonsai.com>
(cherry picked from commit 275086a9b8eb36abe0dbb7499d7e5425a77e32ed)
85/100 · StrongMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context✓ Explains rationale or failure mode
Why it was queued
signing boundarysigning or wallet path
AI analysis · Moderate 68/100
This fix repairs a bug in Core Lightning's splicing feature. When a channel was moved onto a new 'splice' commitment transaction, the old HTLC (multi-hop payment) signatures were still being used. If the channel later closed using that new splice commitment, the stored signatures could not actually spend the HTLC outputs, meaning the node could lose money it was owed. The patch makes sure the correct HTLC signatures are kept and used whenever the channel switches to a splice commitment.
Security candidatechanneld: tell lightningd the HTLC sigs for a splice's first commitmentby jaonoctus · 61ab88e5 · Oct 5, 2026 · 5 filesMessage 78 · AdequateModerate 64Details
Commit message · jaonoctus
channeld: tell lightningd the HTLC sigs for a splice's first commitment
Every commitment the peer signs comes with its signatures for our HTLC transactions, and lightningd stores them so onchaind can spend the HTLC outputs if we close with that commitment. For a splice, got_commitsig stores them for each inflight on every later commitment update, but the inflight's first commitment, signed during the splice negotiation itself, goes to lightningd in channeld_update_inflight, which only carried the commitment and its signature. So any HTLC already pending when the splice was negotiated had no signature stored for the splice's commitment.
That matters once the channel moves onto that commitment: after splice_locked, wallet_htlcsigs_confirm_inflight() promotes the inflight's stored sigs, which then lack those HTLCs, and the same goes for a close on a splice which confirmed before splice_locked.
Add them to channeld_update_inflight, sent whenever it carries a new commitment, and have lightningd replace that inflight's stored sigs with them. channeld_update_inflight could be resent with the commitment lightningd already had when the user signs the splice; that now sends no commitment, so it can't replace those sigs with nothing.
Changelog-None
Co-Authored-By: Ken Sedgwick <ken@bonsai.com>
(cherry picked from commit ebf69a8e6ba29560584060304bfe4a1e2b75dd52)
78/100 · AdequateMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context✓ Names security-relevant behavior explicitly
Why it was queued
signing boundarysigning or wallet path
AI analysis · Moderate 64/100
This patch fixes a bug in Core Lightning's channel splicing feature. When a Bitcoin Lightning channel is being spliced (a fancy way of resizing the channel while it stays open), the software was forgetting to save the peer's signatures for any pending HTLC payments on the very first version of the new commitment transaction. If the channel later closed using that commitment, the node might not be able to claim its funds from those pending payments on the blockchain, potentially leading to loss of money. The fix makes sure those signatures are passed along and stored in the wallet database.
AI review queuedlightningd: conceal secret config values in listconfigsby Andezion · ef48dcd2 · Oct 5, 2026 · 5 filesMessage 73 · AdequateModerate 65Details
Commit message · Andezion
lightningd: conceal secret config values in listconfigs
Changelog-Changed: JSON-RPC: `listconfigs` shows `...` instead of the value of `wallet`, `recover`, `tor-service-password`, `bitcoin-rpcpassword` and the old `bookkeeper-db`, for every caller; unset ones are omitted. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
(cherry picked from commit 6ed212e4856bce2b3392bac1d2fe2df90f3edfde)
73/100 · AdequateMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context
Why it was queued
second-pass: broader security terminology
AI analysis · Moderate 65/100
This commit changes Core Lightning's `listconfigs` RPC command so that several secret or sensitive configuration values are hidden and shown as '...' instead of their real content. Previously, anyone who could call `listconfigs` could see values like the Bitcoin RPC password, wallet database connection string, recovery seed, Tor service password, and old bookkeeper database details. The commit also fixes a bug where concealed options were still being shown when they were not actually set. This is a security hardening change that reduces the risk of leaking credentials to unauthorized callers.
Security candidatecontrib: add jaonoctus's public keyby jaonoctus · 9ba88c5d · Oct 5, 2026 · 1 fileMessage 73 · AdequateInformational 15Details
Commit message · jaonoctus
contrib: add jaonoctus's public key
Commit 41278b4ef ("contrib: add jaonoctus's signing key") added jaonoctus's fingerprint to the "Signatures For Releases" table in SECURITY.md and the security-policy doc, but did not add the armored public key under contrib/keys/ as was done for the other signers.
Add it, so release signatures can be verified against the key shipped in the tree. The primary key fingerprint is 7B69 6A61 6F73 1337 520B 8A19 D8F3 1505 B581 D617, matching the table entry.
✓ Descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context✓ Names security-relevant behavior explicitly
Why it was queued
signing boundarydocumentation-only discount
AI analysis · Informational 15/100
This commit simply adds a public PGP key file for a release signer named jaonoctus. It is a housekeeping change to make the key available in the repository so people can verify release signatures. There is no code change, no vulnerability, and no security risk.
common: ignore proof-only fields when checking omitted markers
find_included_marker_predecessor treated every non-signature field as an invoice TLV. next_marker() of the signature range is 1000000000, so a proof whose only omitted marker was 1000000000 passed with no type-239 field. Skip 1001..999999999, same as next_field_check.
78/100 · AdequateMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context✓ Names security-relevant behavior explicitly
Why it was queued
signing boundary
AI analysis · Moderate 60/100
This commit fixes a validation bug in Core Lightning's handling of BOLT12 payer proofs. A payer proof is a cryptographic receipt that helps verify a payment. The bug allowed a malformed proof—one missing a required type-239 field—to be accepted as valid if it only contained a special 'omitted marker' value of 1,000,000,000. This happened because the code mistakenly treated certain proof-only fields as if they were regular invoice fields when checking whether a marker was legitimate. The fix makes the validator ignore those proof-only fields during that check, matching what another part of the code already did. A new test confirms the invalid proof is now rejected.
AI review queuedconnectd: keep the subd fd until connectd has received itby dovgopoly · 5e19c704 · Sep 29, 2026 · 10 filesMessage 73 · AdequateLow 30Details
Commit message · dovgopoly
connectd: keep the subd fd until connectd has received it
On macOS, a socket passed over SCM_RIGHTS whose sender closes its copy before the receiver's recvmsg completes can arrive unable to read. lightningd closed its end of a new subd's connection right after queueing it for connectd, so openingd or channeld could start with a dead connection: openingd logged "Peer connection lost", and fundchannel failed or hung waiting for accept_channel.
connectd now replies to peer_connect_subd once it has the fd, and lightningd keeps its copy until the reply arrives.
Changelog-Fixed: macOS: fundchannel no longer fails or hangs with "Peer connection lost" when opening channels under load.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
73/100 · AdequateMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context
Why it was queued
signing or wallet pathsecond-pass: security-sensitive path
AI analysis · Low 30/100
This commit fixes a macOS-specific bug where opening a Lightning channel could fail or hang with 'Peer connection lost'. The root cause was a race condition in how file descriptors (sockets) were passed between internal processes: the sender closed its copy too early, before the receiver had fully received it. The fix makes the sender wait for an acknowledgment before closing its copy. It is a reliability/availability fix, not an exploitable security vulnerability.
lightningd: reject newlines in persistent setconfig values
Changelog-Fixed: `setconfig` no longer permits config-line injection through persistent option values.
(cherry picked from commit 54178251aafc0ef793ab9a4dd9c44320bd40812d)
73/100 · AdequateMessage clarity
✓ Specific, descriptive subject✓ Names a concrete action or component✓ Provides detailed explanatory context
Why it was queued
second-pass: broader security terminology
AI analysis · Moderate 62/100
This commit fixes a bug in Core Lightning's `setconfig` RPC command. Previously, a user or attacker who could call `setconfig` could store a configuration value containing newline characters. Because persistent settings are written into a config file line-by-line, those newlines could split one setting into multiple lines, effectively injecting extra configuration directives. The patch now rejects any persistent value containing `\n` or `\r`.