This package had neither a security policy nor contributing guidance, and it is the one in the family where a silent failure is the whole risk. The other four messenger repositories carry both. OSTIF asks for a security policy before an audit application, and the OpenSSF Best Practices Badge requires a documented reporting process.
SECURITY.md states the status plainly: not independently audited, with agreement against the Tor Project's own implementation in place of an audit, and what that does and does not establish. It names security@flamenetmessenger.com, the 72-hour acknowledgement, and the case where a finding belongs to the Tor Project rather than to us. The threat model is one paragraph of what this defends against and four of what it does not, including the bridge-address problem that uniform bytes cannot solve.
CONTRIBUTING.md puts the vectors first, because Reference/vectors is the oracle and the reference wins when Swift disagrees. It names Scripts/interop.sh as the check that matters, since everything else compares this code against recordings or against itself and two identically wrong ends agree perfectly.
Every factual claim was checked against the repository:
FN_OBFS4_STATE is read in Sources/obfs4bridge/main.swift.
bridgecheck has a -tls mode.
The 78-test count matches the CI gate's own grep -q "Executed 78 tests, with 5 tests skipped and 0 failures".
goptlib and lyrebird are pinned in Reference/go.mod, and nothing under Sources/ depends on them.
swift-nio and swift-crypto are the only Package.swift dependencies.
One claim did not survive the check. A first draft said swiftformat and swiftlint configurations are in the repository. They are not, and the file now says to match the surrounding code instead.
This package had neither a security policy nor contributing guidance, and it is the one in the family where a silent failure is the whole risk. The other four messenger repositories carry both. OSTIF asks for a security policy before an audit application, and the OpenSSF Best Practices Badge requires a documented reporting process.
`SECURITY.md` states the status plainly: not independently audited, with agreement against the Tor Project's own implementation in place of an audit, and what that does and does not establish. It names `security@flamenetmessenger.com`, the 72-hour acknowledgement, and the case where a finding belongs to the Tor Project rather than to us. The threat model is one paragraph of what this defends against and four of what it does not, including the bridge-address problem that uniform bytes cannot solve.
`CONTRIBUTING.md` puts the vectors first, because `Reference/vectors` is the oracle and the reference wins when Swift disagrees. It names `Scripts/interop.sh` as the check that matters, since everything else compares this code against recordings or against itself and two identically wrong ends agree perfectly.
Every factual claim was checked against the repository:
- `FN_OBFS4_STATE` is read in `Sources/obfs4bridge/main.swift`.
- `bridgecheck` has a `-tls` mode.
- The 78-test count matches the CI gate's own `grep -q "Executed 78 tests, with 5 tests skipped and 0 failures"`.
- goptlib and lyrebird are pinned in `Reference/go.mod`, and nothing under `Sources/` depends on them.
- swift-nio and swift-crypto are the only `Package.swift` dependencies.
One claim did not survive the check. A first draft said swiftformat and swiftlint configurations are in the repository. They are not, and the file now says to match the surrounding code instead.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
This package had neither a security policy nor contributing guidance, and
it is the one in the family where a silent failure is the whole risk. The
other four messenger repositories carry both.
SECURITY.md states the status plainly: not independently audited, with
agreement against the Tor Project's own implementation in place of an
audit, and it says what that does and does not establish. It names the
reporting address, the 72-hour acknowledgement, and the case where a
finding belongs to the Tor Project rather than to us. The threat model is
one paragraph of what this defends against and four of what it does not,
including the bridge-address problem that uniform bytes cannot solve.
CONTRIBUTING.md puts the vectors first, because `Reference/vectors` is
the oracle and the reference is right when Swift disagrees. It names
`Scripts/interop.sh` as the check that matters, since everything else
compares this code against recordings or against itself and two
identically wrong ends agree perfectly.
Every claim in both files was checked against the repository rather than
assumed. One did not survive: a first draft said swiftformat and
swiftlint configurations are here. They are not, and the file now says
to match the surrounding code instead.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Gate, run locally since this repository has no Gitea workflow:
swift test --enable-xctest --disable-swift-testing: 78 tests, 5 skipped, 5 failures. The 5 failures are the interop tests refusing to run without OBFS4_INTEROP_PARAMS, which is their design and the baseline on main.
Scripts/interop.sh: 6 tests, 0 failures, INTEROP OK: both directions, sans-I/O core and NIO handlers alike.
This PR adds two markdown files and no code, so the Swift result is a baseline rather than a verification of it.
Gate, run locally since this repository has no Gitea workflow:
- `swift test --enable-xctest --disable-swift-testing`: 78 tests, 5 skipped, 5 failures. The 5 failures are the interop tests refusing to run without `OBFS4_INTEROP_PARAMS`, which is their design and the baseline on main.
- `Scripts/interop.sh`: 6 tests, 0 failures, `INTEROP OK: both directions, sans-I/O core and NIO handlers alike`.
This PR adds two markdown files and no code, so the Swift result is a baseline rather than a verification of it.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
This package had neither a security policy nor contributing guidance, and it is the one in the family where a silent failure is the whole risk. The other four messenger repositories carry both. OSTIF asks for a security policy before an audit application, and the OpenSSF Best Practices Badge requires a documented reporting process.
SECURITY.mdstates the status plainly: not independently audited, with agreement against the Tor Project's own implementation in place of an audit, and what that does and does not establish. It namessecurity@flamenetmessenger.com, the 72-hour acknowledgement, and the case where a finding belongs to the Tor Project rather than to us. The threat model is one paragraph of what this defends against and four of what it does not, including the bridge-address problem that uniform bytes cannot solve.CONTRIBUTING.mdputs the vectors first, becauseReference/vectorsis the oracle and the reference wins when Swift disagrees. It namesScripts/interop.shas the check that matters, since everything else compares this code against recordings or against itself and two identically wrong ends agree perfectly.Every factual claim was checked against the repository:
FN_OBFS4_STATEis read inSources/obfs4bridge/main.swift.bridgecheckhas a-tlsmode.grep -q "Executed 78 tests, with 5 tests skipped and 0 failures".Reference/go.mod, and nothing underSources/depends on them.Package.swiftdependencies.One claim did not survive the check. A first draft said swiftformat and swiftlint configurations are in the repository. They are not, and the file now says to match the surrounding code instead.
🤖 Generated with Claude Code
Gate, run locally since this repository has no Gitea workflow:
swift test --enable-xctest --disable-swift-testing: 78 tests, 5 skipped, 5 failures. The 5 failures are the interop tests refusing to run withoutOBFS4_INTEROP_PARAMS, which is their design and the baseline on main.Scripts/interop.sh: 6 tests, 0 failures,INTEROP OK: both directions, sans-I/O core and NIO handlers alike.This PR adds two markdown files and no code, so the Swift result is a baseline rather than a verification of it.
🤖 Generated with Claude Code