Vendored AT Protocol interop test vectors #

Known-answer vectors used as an external oracle for the encoding layers of this workspace.

Upstream	https://github.com/bluesky-social/atproto-interop-tests
Pinned revision	056e5741bb330757205d6b16db5266fffcae937b (2026-07-01)
License	Creative Commons Zero 1.0 Universal — see LICENSE-CC0

CC-0 permits direct copying into other projects without restriction or attribution. The vectors are vendored rather than added as a submodule so that the harnesses below need nothing beyond a fresh clone, and so the revision under test is visible in the diff when it is bumped. That is true of these vectors only: the rest of cargo test --workspace also needs the DASL conformance fixtures in crates/atproto-dasl/tests/dasl-testing, which are a submodule, and without git submodule update --init --recursive 13 atproto-dasl tests fail.

mst/gen_keys.py is not vendored; it is a generator for mst/example_keys.txt, which is.

Why these exist #

Every encoding test in this workspace prior to these vectors was a round trip — encode, decode, compare. A round trip passes against a wrong encoding, because it only proves the code agrees with itself. These files are the only thing in the repository that proves the code agrees with the protocol.

Which harnesses consume which files #

Vectors	Harness
mst/key_heights.json, mst/common_prefix.json, mst/example_keys.txt, firehose/commit-proof-fixtures.json	crates/atproto-repo/tests/interop_mst.rs
data-model/data-model-fixtures.json, data-model/data-model-{valid,invalid}.json (all but six cases of the invalid file, which it lists in NOT_AN_ENCODING_CONCERN)	crates/atproto-dasl/tests/interop_data_model.rs
data-model/data-model-invalid.json (those six: record structure, $type and blob shape)	crates/atproto-lexicon/tests/interop_data_model_structure.rs
syntax/ (all 24 files)	crates/atproto-lexicon/tests/interop_syntax.rs
syntax/datetime_*.txt (3 of those files, again)	crates/atproto-record/tests/interop_datetime.rs
crypto/ (all 3 files)	crates/atproto-identity/tests/interop_crypto.rs
lexicon/ (all 9 files)	crates/atproto-lexicon/tests/interop_lexicon.rs

Every vendored file is consumed, by seven harnesses.

The corpus is at the workspace root, outside every crate's package, so a harness is workspace-only. Six are left out of their crate's package by its exclude, and atproto-dasl's interop_data_model skips itself in a packaged crate, which it recognizes by the Cargo.toml.orig that cargo package writes. A new harness needs one or the other, because crates/readme-doctests/tests/package_contents.rs refuses a published test that reads outside its package.

The table above is checked by an external conformance harness, which is how the three claims this section previously got wrong were found: data-model-{valid,invalid}.json were described as unconsumed after the harness that reads them landed, and mst/example_keys.txt appeared in neither list at all. Keep them accurate in the same change as the harness — a map that is right about most of the corpus is worse than no map, because a reader cannot tell which parts were maintained.

Known failures #

Each of the seven harnesses carries a KNOWN_FAILURES table naming each vector this workspace does not yet satisfy, together with the gap-analysis finding that explains why. A listed vector is required to fail: if it starts passing, the harness fails and instructs you to delete the entry. That way the tables cannot silently rot, and the moment an encoding fix lands the gate tells you which finding it closed.

Every harness enforces it, too, which this said before it was true: interop_crypto.rs declared a table nothing read, and interop_data_model_structure.rs had none. An empty table is not an exemption — it is the statement that nothing is pinned, and it only means that if adding the first entry would do something. All seven now check both directions, so a rule stated here holds everywhere it is stated about. So does an eighth table outside this map, crates/atproto-pds/tests/integration/interop_firehose.rs, which follows the same rule for the firehose's frames but reads nothing in this directory.

Do not add an entry to relax a genuine regression. An entry is a statement that a known, filed defect is still open.

Bumping the pin #

git clone --depth 1 https://github.com/bluesky-social/atproto-interop-tests /tmp/interop
rsync -a --delete --exclude .git --exclude .gitignore --exclude mst/gen_keys.py \
      /tmp/interop/ tests/interop/

Then restore this README, update the pinned revision above, and re-run cargo test --workspace. New vectors that fail are new findings, not licence to extend KNOWN_FAILURES.