Directional compatibility evidence

OpenInference 0.1.33 → OTLP 1.11

OpenInference kinds and attributes remain ordinary OTLP span attributes, but parseability does not repair broken correlation or prove a complete capture.

OpenInference semantic conventionsopeninference-semantic-conventions 0.1.33 · Python package tag d10212b2049d097e101f7ec0fa5a7dd814a7d9c9

OTLP 1.11.0 HTTP/protobuf1.11.0 · OTLP 1.11.0

Source transport: OpenTelemetry spans over OTLP
Destination transport: HTTP with binary Protocol Buffers
Source maturity: Versioned semantic-conventions package
Evaluation protocol: tracevity-compatibility-v0.2

Read the evidence layers separately

Documentation describes intended semantics. Tracevity interpretation binds those meanings directionally. Tracevity-tested evidence reports only what the named local, synthetic fixture protocol produced. None of those layers proves resulting external state.

Additive correction trail · current proof

One defective v0.2 receipt is preserved as superseded, not presented as current.

OTLP masks W3C trace flags to 0xff and sampled is bit 0. The exact raw value 256 records known remote-context status, trace_flags=0, sampled=false, and remote=false. The full sampling policy remains external.

Current proof status: CORRECTED_FORWARD_PROOF
Correction ID: v03-otlp-span-flags-mask
Current assertion: openinference-capture_provenance
Corrected run · fixture · check: eval-v03-openinference-span-flags-correction · openinference-capture-provenance-v03-corrected · sampling_state_recorded
Historical receipt disposition: SUPERSEDED_DEFECTIVE
Superseded v0.2 run · fixture · check: eval-openinference-0_1_33 · openinference-capture-completeness · sampled_bit_present

Defect: The frozen v0.2 generic receipt treated OTLP span.flags bit 8 (0x100) as the sampled bit. The frozen release, raw capture, and defective receipt remain unchanged for auditability; only the additive corrected proof is current.

Open the official OTLP flags definition ↗

Four answers

What this mapping preserves—and what it cannot carry alone

What survives?

Dropped-record indication

What is degraded or lost?

Product identity
Agent instance ID
Tool call ID
Initiating principal
Sampling and capture provenance
Truncation indication

What still requires external evidence?

No assertion in this mapping names an external-evidence dependency.

How do we know?

4 bounded primary-source records, 2 synthetic fixtures, and 1 linked reproducible evaluation run support the separate evidence layers below.

Semantic bindings

Every result stays attached to a location and evidence layer.

“External evidence required” means trace-format preservation cannot establish the fact. Destination state, identity, authorization, or another independently controlled source remains necessary.

Directional semantic assertions for OpenInference 0.1.33 → OTLP 1.11
Capability	Concrete location	Result	Tracevity interpretation	Evidence layers	Limitations
Product identityIdentity	resource.telemetry.sdk.* plus instrumentation scope	Partial	OTLP identifies the emitting SDK/scope, but that is not a stable product deployment identity.	DOCUMENTEDThe pinned primary source documents the named representation or transformation. TRACEVITY INTERPRETEDOTLP identifies the emitting SDK/scope, but that is not a stable product deployment identity. TRACEVITY TESTEDTracevity exercised this bounded behavior locally with the linked synthetic fixture and evaluation run. Exact assertion proof v0.3 assertion-specific predicate `PREDICATE_MATCHED` means this bounded assertion-specific predicate plan matched the admitted synthetic inputs. It does not establish capability completeness, whole-system conformance, or certification. Proof status HISTORICAL_ASSERTION_PROOF Evaluator 0.3.0 Plan v0.3.0 · ALL_CLAUSES_TRUE · 2 clause(s) Observed outcome `PREDICATE_MATCHED` Predicate plan SHA-256 `945bc5dfe453c742c311fe7ddbf7d757ab78b9f85a08ef5835fd0e46fc302333` Input set SHA-256 `699bf92a7dd6a386998d7dbe09376c53b975d47c2fb8b9d29c58f2e83110d509` Predicate result SHA-256 `7438c867c60da545f00328dfb9bad72104004471866275b3b07dc71044098d93` Observed clauses openinference_kind_presentATTRIBUTE_EXISTS · matched · 1 observed match(es) · 1 value digest(s) Show value digests (1) `aa8afcb616454013a5e75920d60ecc502e5232b53a6f69c9e47bb927d8c7c8c5` emitter_resource_scope_presentRESOURCE_SCOPE_EXISTS · matched · 1 observed match(es) · 0 value digest(s) Admitted predicate inputs OBSERVED · openinference-capture-completeness Normalized `df7e0fc1b1a1f7b7116e52ea0b91ff89e5003e4bcae2ce9d8571d1a29246e038` Historical result `ba5a6a1db3a025c9deb5916b9a88dfaf36e1baad5e882c88beaf7d610bb534ec` Historical v0.2 passed-check receipt This separate release receipt records the admitted run, fixture, checks, and artifact digests. The v0.3 predicate execution does not rewrite or broaden it. Run eval-openinference-0_1_33 Fixture openinference-capture-completeness Tested result PASS Passed checks otlp_decoded, openinference_kind Observed locator $.resource.attributes and $.scope Extractor RESULT_CHECK Historical normalized SHA-256 `df7e0fc1b1a1f7b7116e52ea0b91ff89e5003e4bcae2ce9d8571d1a29246e038` Historical check-result SHA-256 `ba5a6a1db3a025c9deb5916b9a88dfaf36e1baad5e882c88beaf7d610bb534ec`	Resource identity is emitter-configured. Binding conditions and proof ceiling Presence CONDITIONAL Correlation Use OTLP trace_id and parent_span_id; use a semantic call identifier when the capability depends on a request/result join. Relationship A field is evidence only for the span or relationship on which it is recorded; sibling presence does not imply a join. Sampling Sampling provenance must be known before absence is treated as a negative. Dropped records Dropped-record counters or exporter drop risk reduce completeness. Truncation Content limits must be known; absent content may be filtered or truncated. Verification ceiling The exercised local representation can establish recorded trace structure only; authorization, exporter completeness, backend acceptance, and destination state require separate evidence.
Agent instance IDIdentity	span.attributes['agent.name'] + openinference.span.kind=AGENT	Partial	Agent semantics survive as attributes, but a stable instance identifier is not guaranteed.	DOCUMENTEDThe pinned primary source documents the named representation or transformation. TRACEVITY INTERPRETEDAgent semantics survive as attributes, but a stable instance identifier is not guaranteed. TRACEVITY TESTEDTracevity exercised this bounded behavior locally with the linked synthetic fixture and evaluation run. Exact assertion proof v0.3 assertion-specific predicate `PREDICATE_MATCHED` means this bounded assertion-specific predicate plan matched the admitted synthetic inputs. It does not establish capability completeness, whole-system conformance, or certification. Proof status HISTORICAL_ASSERTION_PROOF Evaluator 0.3.0 Plan v0.3.0 · ALL_CLAUSES_TRUE · 2 clause(s) Observed outcome `PREDICATE_MATCHED` Predicate plan SHA-256 `2d685e6b733cd143be70c8a6d9a2e9334455a477abde901b2a929a89d3ff60d0` Input set SHA-256 `699bf92a7dd6a386998d7dbe09376c53b975d47c2fb8b9d29c58f2e83110d509` Predicate result SHA-256 `ba44010b89740518e2ff0342a085a073a29ee8d8fb728f7ebbb37e1ab0383bc6` Observed clauses agent_name_presentATTRIBUTE_EXISTS · matched · 1 observed match(es) · 1 value digest(s) Show value digests (1) `5d9e28d8a733d37969dffa976933ca13553cdaf7935c351343f5a864fc7ac0fc` stable_agent_id_absentATTRIBUTE_KEYS_ABSENT · matched · 0 observed match(es) · 0 value digest(s) Admitted predicate inputs OBSERVED · openinference-capture-completeness Normalized `df7e0fc1b1a1f7b7116e52ea0b91ff89e5003e4bcae2ce9d8571d1a29246e038` Historical result `ba5a6a1db3a025c9deb5916b9a88dfaf36e1baad5e882c88beaf7d610bb534ec` Historical v0.2 passed-check receipt This separate release receipt records the admitted run, fixture, checks, and artifact digests. The v0.3 predicate execution does not rewrite or broaden it. Run eval-openinference-0_1_33 Fixture openinference-capture-completeness Tested result PASS Passed checks openinference_kind Observed locator $.spans[*].attributes['agent.name'] Extractor JSONPATH_EXISTENCE Historical normalized SHA-256 `df7e0fc1b1a1f7b7116e52ea0b91ff89e5003e4bcae2ce9d8571d1a29246e038` Historical check-result SHA-256 `ba5a6a1db3a025c9deb5916b9a88dfaf36e1baad5e882c88beaf7d610bb534ec`	A name is not necessarily unique or authenticated. Binding conditions and proof ceiling Presence CONDITIONAL Correlation Use OTLP trace_id and parent_span_id; use a semantic call identifier when the capability depends on a request/result join. Relationship A field is evidence only for the span or relationship on which it is recorded; sibling presence does not imply a join. Sampling Sampling provenance must be known before absence is treated as a negative. Dropped records Dropped-record counters or exporter drop risk reduce completeness. Truncation Content limits must be known; absent content may be filtered or truncated. Verification ceiling The exercised local representation can establish recorded trace structure only; authorization, exporter completeness, backend acceptance, and destination state require separate evidence.
Tool call IDTools	flattened message tool-call id and tool span call id	Partial	Records remain parseable when identifiers disagree, but the tool join fails.	DOCUMENTEDThe pinned primary source documents the named representation or transformation. TRACEVITY INTERPRETEDRecords remain parseable when identifiers disagree, but the tool join fails. TRACEVITY TESTEDTracevity exercised this bounded behavior locally with the linked synthetic fixture and evaluation run. Exact assertion proof v0.3 assertion-specific predicate `PREDICATE_MATCHED` means this bounded assertion-specific predicate plan matched the admitted synthetic inputs. It does not establish capability completeness, whole-system conformance, or certification. Proof status HISTORICAL_ASSERTION_PROOF Evaluator 0.3.0 Plan v0.3.0 · ALL_CLAUSES_TRUE · 1 clause(s) Observed outcome `PREDICATE_MATCHED` Predicate plan SHA-256 `d20a4fe5db6ecf2096941e2237ebd02316f44acb4351f7fe23fc4f02796dae7f` Input set SHA-256 `3dc8d165935f1bc19bceaefb7480160fd501edfd54c48576649dd8800522c870` Predicate result SHA-256 `30079a5a90bad7cfbe7b31900400e7fcdf509aa97724c93340361dd116b8257a` Observed clauses tool_call_ids_disagreeATTRIBUTE_VALUES_NOT_EQUAL · matched · 2 observed match(es) · 2 value digest(s) Show value digests (2) `11b081cc9f036bf7236dbf6f59aeef89519533607251cc8573f1e6904ea89663` `e06073c4e9290f2dd8f35d9715c2ca3da09cc84742e154219cfd1bea30a7e1e5` Admitted predicate inputs OBSERVED · openinference-correlation-break Normalized `618b280c7d8084beaaf167618994e3f359812a51c6c171ed253ee94411eae152` Historical result `d232dff5d0f86b7ec80bf571150a1fd4a93d0ff9b1835a67fd60dd6a710973ea` Historical v0.2 passed-check receipt This separate release receipt records the admitted run, fixture, checks, and artifact digests. The v0.3 predicate execution does not rewrite or broaden it. Run eval-openinference-0_1_33 Fixture openinference-correlation-break Tested result PASS Passed checks individual_records_parse, join_fails Observed locator $.spans[*].attributes tool-call id fields ('llm.output_messages...tool_call.id' and 'tool.call.id') Extractor JSONPATH_INEQUALITY Historical normalized SHA-256 `618b280c7d8084beaaf167618994e3f359812a51c6c171ed253ee94411eae152` Historical check-result SHA-256 `d232dff5d0f86b7ec80bf571150a1fd4a93d0ff9b1835a67fd60dd6a710973ea`	OTLP decoding alone cannot establish semantic correlation. Binding conditions and proof ceiling Presence CONDITIONAL Correlation Use OTLP trace_id and parent_span_id; use a semantic call identifier when the capability depends on a request/result join. Relationship A field is evidence only for the span or relationship on which it is recorded; sibling presence does not imply a join. Sampling Sampling provenance must be known before absence is treated as a negative. Dropped records Dropped-record counters or exporter drop risk reduce completeness. Truncation Content limits must be known; absent content may be filtered or truncated. Verification ceiling The exercised local representation can establish recorded trace structure only; authorization, exporter completeness, backend acceptance, and destination state require separate evidence.
Initiating principalPrincipal	optional user.id	Ambiguous	A contextual user identifier is not proof of an authenticated initiating principal or delegation chain.	DOCUMENTEDThe pinned primary source documents the named representation or transformation. TRACEVITY INTERPRETEDA contextual user identifier is not proof of an authenticated initiating principal or delegation chain. TRACEVITY TESTEDTracevity exercised this bounded behavior locally with the linked synthetic fixture and evaluation run. Exact assertion proof v0.3 assertion-specific predicate `PREDICATE_MATCHED` means this bounded assertion-specific predicate plan matched the admitted synthetic inputs. It does not establish capability completeness, whole-system conformance, or certification. Proof status HISTORICAL_ASSERTION_PROOF Evaluator 0.3.0 Plan v0.3.0 · ALL_CLAUSES_TRUE · 2 clause(s) Observed outcome `PREDICATE_MATCHED` Predicate plan SHA-256 `e5c64ba7d24d539f4cfde7f0abb9dac87b9d0c9bf9c96de18b37a845225b79b1` Input set SHA-256 `699bf92a7dd6a386998d7dbe09376c53b975d47c2fb8b9d29c58f2e83110d509` Predicate result SHA-256 `ec8d909ba4f0bee7e1de6682113615a87575c498175e09888071a2a937925e78` Observed clauses openinference_kind_presentATTRIBUTE_EXISTS · matched · 1 observed match(es) · 1 value digest(s) Show value digests (1) `aa8afcb616454013a5e75920d60ecc502e5232b53a6f69c9e47bb927d8c7c8c5` principal_keys_absentATTRIBUTE_KEYS_ABSENT · matched · 0 observed match(es) · 0 value digest(s) Admitted predicate inputs OBSERVED · openinference-capture-completeness Normalized `df7e0fc1b1a1f7b7116e52ea0b91ff89e5003e4bcae2ce9d8571d1a29246e038` Historical result `ba5a6a1db3a025c9deb5916b9a88dfaf36e1baad5e882c88beaf7d610bb534ec` Historical v0.2 passed-check receipt This separate release receipt records the admitted run, fixture, checks, and artifact digests. The v0.3 predicate execution does not rewrite or broaden it. Run eval-openinference-0_1_33 Fixture openinference-capture-completeness Tested result PASS Passed checks otlp_decoded, openinference_kind Observed locator $.spans[*].attributes['user.id'] Extractor JSONPATH_ABSENCE Historical normalized SHA-256 `df7e0fc1b1a1f7b7116e52ea0b91ff89e5003e4bcae2ce9d8571d1a29246e038` Historical check-result SHA-256 `ba5a6a1db3a025c9deb5916b9a88dfaf36e1baad5e882c88beaf7d610bb534ec`	Identity assurance requires external authentication evidence. Binding conditions and proof ceiling Presence OPTIONAL Correlation Use OTLP trace_id and parent_span_id; use a semantic call identifier when the capability depends on a request/result join. Relationship A field is evidence only for the span or relationship on which it is recorded; sibling presence does not imply a join. Sampling Sampling provenance must be known before absence is treated as a negative. Dropped records Dropped-record counters or exporter drop risk reduce completeness. Truncation Content limits must be known; absent content may be filtered or truncated. Verification ceiling The exercised local representation can establish recorded trace structure only; authorization, exporter completeness, backend acceptance, and destination state require separate evidence.
Dropped-record indicationIntegrity	OTLP span.dropped_attributes_count	Preserved	The exercised OTLP payload reports dropped attributes.	DOCUMENTEDThe pinned primary source documents the named representation or transformation. TRACEVITY INTERPRETEDThe exercised OTLP payload reports dropped attributes. TRACEVITY TESTEDTracevity exercised this bounded behavior locally with the linked synthetic fixture and evaluation run. Exact assertion proof v0.3 assertion-specific predicate `PREDICATE_MATCHED` means this bounded assertion-specific predicate plan matched the admitted synthetic inputs. It does not establish capability completeness, whole-system conformance, or certification. Proof status HISTORICAL_ASSERTION_PROOF Evaluator 0.3.0 Plan v0.3.0 · ALL_CLAUSES_TRUE · 1 clause(s) Observed outcome `PREDICATE_MATCHED` Predicate plan SHA-256 `8f6394cc7259e9dea71e8d9f1843365a93d3f288a1824fb4d0fea352c621d3b0` Input set SHA-256 `699bf92a7dd6a386998d7dbe09376c53b975d47c2fb8b9d29c58f2e83110d509` Predicate result SHA-256 `704438750589193e7a571b7aaf023fd9cd93a08b9a9b4a524c13606abb66cecc` Observed clauses dropped_attributes_positiveSPAN_NUMBER_GREATER_THAN · matched · 1 observed match(es) · 1 value digest(s) Show value digests (1) `e7f6c011776e8db7cd330b54174fd76f7d0216b612387a5ffcfb81e6f0919683` Admitted predicate inputs OBSERVED · openinference-capture-completeness Normalized `df7e0fc1b1a1f7b7116e52ea0b91ff89e5003e4bcae2ce9d8571d1a29246e038` Historical result `ba5a6a1db3a025c9deb5916b9a88dfaf36e1baad5e882c88beaf7d610bb534ec` Historical v0.2 passed-check receipt This separate release receipt records the admitted run, fixture, checks, and artifact digests. The v0.3 predicate execution does not rewrite or broaden it. Run eval-openinference-0_1_33 Fixture openinference-capture-completeness Tested result PASS Passed checks drop_indicated Observed locator OTLP span dropped_attributes_count Extractor RESULT_CHECK Historical normalized SHA-256 `df7e0fc1b1a1f7b7116e52ea0b91ff89e5003e4bcae2ce9d8571d1a29246e038` Historical check-result SHA-256 `ba5a6a1db3a025c9deb5916b9a88dfaf36e1baad5e882c88beaf7d610bb534ec`	A zero counter would not prove that exporters or upstream systems dropped nothing. Binding conditions and proof ceiling Presence CONDITIONAL Correlation Use OTLP trace_id and parent_span_id; use a semantic call identifier when the capability depends on a request/result join. Relationship A field is evidence only for the span or relationship on which it is recorded; sibling presence does not imply a join. Sampling Sampling provenance must be known before absence is treated as a negative. Dropped records Dropped-record counters or exporter drop risk reduce completeness. Truncation Content limits must be known; absent content may be filtered or truncated. Verification ceiling The exercised local representation can establish recorded trace structure only; authorization, exporter completeness, backend acceptance, and destination state require separate evidence.
Sampling and capture provenanceIntegrity	OTLP trace flags plus fixture manifest	Partial	Current corrected interpretation OTLP masks W3C trace flags to 0xff and sampled is bit 0. The exact raw value 256 records known remote-context status, trace_flags=0, sampled=false, and remote=false. The full sampling policy remains external.	DOCUMENTEDThe pinned primary source documents the named representation or transformation. TRACEVITY INTERPRETEDOTLP masks W3C trace flags to 0xff and sampled is bit 0. The exact raw value 256 records known remote-context status, trace_flags=0, sampled=false, and remote=false. The full sampling policy remains external. TRACEVITY TESTEDThe additive v0.3 correction exercised the unchanged historical raw capture with the corrected OTLP flags decoder and exact predicate. Additive v0.3 corrected proof v0.3 corrected assertion-specific predicate `PREDICATE_MATCHED` means this bounded assertion-specific predicate plan matched the admitted synthetic inputs. It does not establish capability completeness, whole-system conformance, or certification. Proof status CORRECTED_FORWARD_PROOF Correction ID v03-otlp-span-flags-mask Evaluator 0.3.0 Plan v0.3.0 · ALL_CLAUSES_TRUE · 4 clause(s) Observed outcome `PREDICATE_MATCHED` Predicate plan SHA-256 `a70dceb3e56abd59724fb434dc09d565fd3946d4e0acf3f154648d8f7241e4a1` Input set SHA-256 `ec946d2f625a3d6e8c4aeee18c77fe25bf93820edd7a2ac8be8ef1969af9eb9a` Predicate result SHA-256 `fd564bef1418f892d06c91306f50c3aba51751edd66693ecdc4899d64d489329` Observed clauses trace_flags_unsampledSPAN_FIELD_EQUALS · matched · 1 observed match(es) · 0 value digest(s) sampled_falseSPAN_FIELD_EQUALS · matched · 1 observed match(es) · 0 value digest(s) remote_context_knownSPAN_FIELD_EQUALS · matched · 1 observed match(es) · 0 value digest(s) remote_context_falseSPAN_FIELD_EQUALS · matched · 1 observed match(es) · 0 value digest(s) Corrected predicate inputs OBSERVED · openinference-capture-provenance-v03-corrected Normalized `272b6bda69e71297a9e6d12e19ef5c41f822fa1946ab879e09f2692fc32a40e7` Corrected result `e1eeea258a6534edb7b3aa1bc8b67c309caedd2a3a99cca156a0b4d79d6226b1` Additive v0.3 corrected check receipt This current forward proof uses a new run, fixture, check, and artifact digests. It does not rewrite the frozen v0.2 release. Run eval-v03-openinference-span-flags-correction Fixture openinference-capture-provenance-v03-corrected Tested result PASS Passed checks sampling_state_recorded Observed locator OTLP span.flags low-byte trace flags and separately decoded remote-context metadata Extractor RESULT_CHECK Corrected normalized SHA-256 `272b6bda69e71297a9e6d12e19ef5c41f822fa1946ab879e09f2692fc32a40e7` Corrected check-result SHA-256 `e1eeea258a6534edb7b3aa1bc8b67c309caedd2a3a99cca156a0b4d79d6226b1` Superseded historical v0.2 receipteval-openinference-0_1_33 · openinference-capture-completeness · sampled_bit_present is SUPERSEDED_DEFECTIVE. It is retained only as correction history and is not current proof.	This corrected local fixture establishes an unsampled span and known non-remote context only for the exact raw capture. It does not establish exporter sampling policy, capture completeness, or behavior of another trace. Binding conditions and proof ceiling Presence CONDITIONAL Correlation Use OTLP trace_id and parent_span_id; use a semantic call identifier when the capability depends on a request/result join. Relationship A field is evidence only for the span or relationship on which it is recorded; sibling presence does not imply a join. Sampling Sampling provenance must be known before absence is treated as a negative. Dropped records Dropped-record counters or exporter drop risk reduce completeness. Truncation Content limits must be known; absent content may be filtered or truncated. Verification ceiling The exercised local representation can establish recorded trace structure only; authorization, exporter completeness, backend acceptance, and destination state require separate evidence.
Truncation indicationIntegrity	No standard generic truncation marker in the exercised attributes	Ambiguous	Declared truncation risk prevents absence from becoming a definitive negative.	DOCUMENTEDThe pinned primary source documents the named representation or transformation. TRACEVITY INTERPRETEDDeclared truncation risk prevents absence from becoming a definitive negative. TRACEVITY TESTEDTracevity exercised this bounded behavior locally with the linked synthetic fixture and evaluation run. Exact assertion proof v0.3 assertion-specific predicate `PREDICATE_MATCHED` means this bounded assertion-specific predicate plan matched the admitted synthetic inputs. It does not establish capability completeness, whole-system conformance, or certification. Proof status HISTORICAL_ASSERTION_PROOF Evaluator 0.3.0 Plan v0.3.0 · ALL_CLAUSES_TRUE · 2 clause(s) Observed outcome `PREDICATE_MATCHED` Predicate plan SHA-256 `1c523f5e73ec1206795f91a2607b963dff57caa84e957f7ee7c6f896dd31ab63` Input set SHA-256 `699bf92a7dd6a386998d7dbe09376c53b975d47c2fb8b9d29c58f2e83110d509` Predicate result SHA-256 `326df56398d1f91fb529008672f9ba07d3532a8e1cbb00394fcdc7a9a7b25271` Observed clauses fixture_truncation_risk_presentATTRIBUTE_EXISTS · matched · 1 observed match(es) · 1 value digest(s) Show value digests (1) `b8dd268b25e7b4ceaab4539b59f7613d131e95a0fead493198437cd713411448` standard_truncation_marker_absentATTRIBUTE_KEYS_ABSENT · matched · 0 observed match(es) · 0 value digest(s) Admitted predicate inputs OBSERVED · openinference-capture-completeness Normalized `df7e0fc1b1a1f7b7116e52ea0b91ff89e5003e4bcae2ce9d8571d1a29246e038` Historical result `ba5a6a1db3a025c9deb5916b9a88dfaf36e1baad5e882c88beaf7d610bb534ec` Historical v0.2 passed-check receipt This separate release receipt records the admitted run, fixture, checks, and artifact digests. The v0.3 predicate execution does not rewrite or broaden it. Run eval-openinference-0_1_33 Fixture openinference-capture-completeness Tested result PASS Passed checks truncation_not_proven Observed locator $.spans[*].attributes['tracevity.fixture.truncation_risk'] Extractor RESULT_CHECK Historical normalized SHA-256 `df7e0fc1b1a1f7b7116e52ea0b91ff89e5003e4bcae2ce9d8571d1a29246e038` Historical check-result SHA-256 `ba5a6a1db3a025c9deb5916b9a88dfaf36e1baad5e882c88beaf7d610bb534ec`	The risk marker is Tracevity fixture metadata, not an OpenInference guarantee. Binding conditions and proof ceiling Presence ABSENT Correlation Use OTLP trace_id and parent_span_id; use a semantic call identifier when the capability depends on a request/result join. Relationship A field is evidence only for the span or relationship on which it is recorded; sibling presence does not imply a join. Sampling Sampling provenance must be known before absence is treated as a negative. Dropped records Dropped-record counters or exporter drop risk reduce completeness. Truncation Content limits must be known; absent content may be filtered or truncated. Verification ceiling The exercised local representation can establish recorded trace structure only; authorization, exporter completeness, backend acceptance, and destination state require separate evidence.

Requirement results

Good enough remains a use-case question.

Each result is scoped to one assertion, one named reconstruction use case, and this exact directional mapping. It is not an aggregate verdict about either system.

Partial
Incident reconstruction
Evidence item: Product identity · Requirements v0.2
OTLP identifies the emitting SDK/scope, but that is not a stable product deployment identity.
Partial
Multi-agent reconstruction
Evidence item: Agent instance ID · Requirements v0.2
Agent semantics survive as attributes, but a stable instance identifier is not guaranteed.
Not established
Tool-action attribution
Evidence item: Tool call ID · Requirements v0.2
Records remain parseable when identifiers disagree, but the tool join fails.
Not established
Human-delegation reconstruction
Evidence item: Initiating principal · Requirements v0.2
A contextual user identifier is not proof of an authenticated initiating principal or delegation chain.
Partial
Incident reconstruction
Evidence item: Dropped-record indication · Requirements v0.2
The exercised OTLP payload reports dropped attributes.
Partial
Incident reconstruction
Evidence item: Sampling and capture provenance · Requirements v0.2
OTLP masks W3C trace flags to 0xff and sampled is bit 0. The exact raw value 256 records known remote-context status, trace_flags=0, sampled=false, and remote=false. The full sampling policy remains external.
Partial
Incident reconstruction
Evidence item: Truncation indication · Requirements v0.2
Declared truncation risk prevents absence from becoming a definitive negative.

Fixture artifacts

Synthetic inputs and normalized outputs with stable digests

These public artifacts contain bounded synthetic data generated through a provider-free local protocol. They are supplied for reproducibility and do not accept customer traces.

openinference-0_1_33

Valid OpenInference spans parse, but their tool-call identifiers do not join.

correlation_break

Raw SHA-256: 1749e8a5e6a3eacb6eeb42b9c1bae9b0e243e4002b66b34e3e2c6ede3d977bccNormalized SHA-256: 618b280c7d8084beaaf167618994e3f359812a51c6c171ed253ee94411eae152Result SHA-256: d232dff5d0f86b7ec80bf571150a1fd4a93d0ff9b1835a67fd60dd6a710973ea

Generated August 29, 2026 · synthetic public fixture

Raw capture Normalized semantics Evaluation result

Frozen v0.2 fixture · superseded receipt

Historical artifact preserved for audit; its sampled-bit receipt is superseded.

capture_completeness_problem

Raw SHA-256: 52750868511e95f7f1d85e1ecea8e4423781d0ee5495e8a8dfad146c07dbf5faNormalized SHA-256: df7e0fc1b1a1f7b7116e52ea0b91ff89e5003e4bcae2ce9d8571d1a29246e038Result SHA-256: ba5a6a1db3a025c9deb5916b9a88dfaf36e1baad5e882c88beaf7d610bb534ec

Generated August 29, 2026 · synthetic public fixture

Raw capture Normalized semantics Evaluation result

Frozen v0.2 evaluation · affected receipt superseded

eval-openinference-0_1_33

Evaluator 0.2.0 · protocol tracevity-compatibility-v0.2 · August 29, 2026

Result PASSED · reproducibility REPRODUCIBLE · provider network DISABLED · source versions openInferenceSemanticConventions: 0.1.33 tag d10212b2049d097e101f7ec0fa5a7dd814a7d9c9; otlpSpecification: 1.11.0; openTelemetrySdk: 1.44.0

This run remains historical. Its sampled_bit_present receipt is SUPERSEDED_DEFECTIVE; current proof uses eval-v03-openinference-span-flags-correction.

Recorded fixture digests

Raw openinference-correlation-break: 1749e8a5e6a3eacb6eeb42b9c1bae9b0e243e4002b66b34e3e2c6ede3d977bcc
Raw openinference-capture-completeness: 52750868511e95f7f1d85e1ecea8e4423781d0ee5495e8a8dfad146c07dbf5fa
Normalized openinference-correlation-break: 618b280c7d8084beaaf167618994e3f359812a51c6c171ed253ee94411eae152
Normalized openinference-capture-completeness: df7e0fc1b1a1f7b7116e52ea0b91ff89e5003e4bcae2ce9d8571d1a29246e038

Compatibility evidence

Bounded observations and primary sources

OpenTelemetry Protocol 1.11.0OpenTelemetry · observed August 29, 2026
Successful protobuf decoding establishes a transport representation, not semantic equivalence, completeness, or backend acceptance. Protocol details and trace service
Open source ↗
OpenInference specificationArize AI · observed August 29, 2026
OTLP compatibility is structural; OpenInference meaning resides in attributes and relationships. Relationship to OpenTelemetry
Open source ↗
OpenInference semantic conventionsArize AI · observed August 29, 2026
Individual semantic records can parse while cross-record reconstruction still fails when correlation identifiers do not join. Span kinds and semantic fields
Open source ↗
openinference-semantic-conventions 0.1.33Arize AI · observed August 29, 2026
The exercised constants package is pinned to 0.1.33 rather than treated as timeless OpenInference. Release metadata
Open source ↗

Compare the mapping result with the evidence your use case requires.

Trace Reconstruction Requirements v0.2 adds capture-completeness requirements while the unchanged v0.1 baseline remains available. Apply this binding to a supported local artifact with the Inspector Alpha.

Download binding schema ↓ · Download assertion proofs ↓

Open Requirements Open Inspector guide

One defective v0.2 receipt is preserved as superseded, not presented as current.

What this mapping preserves—and what it cannot carry alone

What survives?

What is degraded or lost?

What still requires external evidence?

How do we know?

Every result stays attached to a location and evidence layer.

Good enough remains a use-case question.

Incident reconstruction

Multi-agent reconstruction

Tool-action attribution

Human-delegation reconstruction

Incident reconstruction

Incident reconstruction

Incident reconstruction

Synthetic inputs and normalized outputs with stable digests

Valid OpenInference spans parse, but their tool-call identifiers do not join.

Historical artifact preserved for audit; its sampled-bit receipt is superseded.

eval-openinference-0_1_33

Bounded observations and primary sources

Compare the mapping result with the evidence your use case requires.