replay-test

name	replay-test
description	Run replay fidelity tests — verify that stored artifacts can reconstruct the original HTTP exchange, DOM state, and resource graph. Use after modifying artifact, storage, or replay crates.
version	0.1.0
user-invocable	true
allowed-tools	Bash(cargo *) Read Grep Glob
argument-hint	[test-filter]

Replay Fidelity Testing

Verify Law 5: stored artifacts must be sufficient to reconstruct what happened.

Test Matrix

HTTP Exchange Replay

• Request method, URL, headers preserved exactly
• Response status, headers, body preserved exactly
• Content-Encoding handled (gzip, br, deflate round-trip)
• Chunked transfer encoding reconstructed
• Trailer headers captured and replayed

DOM Snapshot Replay (browser mode)

• DOM structure matches post-JS-execution state
• Inline styles computed and captured
• Shadow DOM boundaries preserved
• iframe content captured with parent context

Resource Graph Replay

• All sub-resources referenced in graph exist in storage
• Load order matches original browser execution
• Initiator chains (CSS -> font, JS -> XHR) preserved
• Failed resource loads recorded as errors, not omitted

Temporal Consistency

• Replayed artifacts carry original CaptureInstant
• Logical clock ordering matches capture order
• Crawl context ID links back to the originating crawl

Procedure

1. Run cargo test with filter $ARGUMENTS (or replay if not specified).
2. For each test, verify the round-trip: capture -> serialize -> store -> retrieve -> deserialize -> compare.
3. Report any fidelity gaps.

Output

Replay Fidelity Report
======================
Tests run: {n}
Passed: {n}
Failed: {n}

Failures:
- test_replay_gzip_response: body hash mismatch (expected abc..., got def...)
- test_replay_resource_graph: missing sub-resource for font.woff2

Verdict: {PASS / FAIL}