Skip to content

Commit 12ae885

Browse files
committed
Release RelayX v0.1.12 evidence taxonomy
1 parent 11f24ae commit 12ae885

28 files changed

Lines changed: 338 additions & 53 deletions

README.md

Lines changed: 6 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -73,8 +73,8 @@ telemetry.
7373
- Lab response differential analysis for stable policy-state pairs, including
7474
discriminator keys, context-only differences, and promotion support.
7575
- Evidence completeness reporting for finding/path records, including protocol
76-
judgement fields, confidence distribution, missing contract keys, and
77-
remaining uncertainty.
76+
judgement fields, source taxonomy, confidence distribution, missing contract
77+
keys, and remaining uncertainty.
7878
- Guarded validation and execution records with dry-run, armed, and confirmed
7979
modes, operator context, timebox/noise/scope checks, and JSONL audit logs.
8080
- Source validation planning for WebClient/WebDAV, RPC coercion surfaces,
@@ -222,7 +222,7 @@ relayx lab-stability Assess repeat-capture lab stability and drift
222222
relayx lab-diff Compare stable lab policy-state response differences
223223
relayx lab-index Summarize lab signature corpuses
224224
relayx lab-profile Generate a calibration profile draft from corpuses
225-
relayx evidence-report Audit evidence completeness and judgement fields
225+
relayx evidence-report Audit evidence completeness, source taxonomy, and judgement fields
226226
relayx validate Run guarded active validation for one path
227227
relayx profiles List bundled RelayX profiles
228228
relayx export Export graph, JSONL, CSV, report, or diagram artifacts
@@ -313,7 +313,9 @@ an artifact does not satisfy the selected contract.
313313
`relayx evidence-report -r result.json` audits an existing result without
314314
network activity. It highlights candidate or relayable records without
315315
evidence, protocol judgement records missing policy inference or remaining
316-
uncertainty, and evidence entries that still carry unknown confidence.
316+
uncertainty, evidence entries that still carry unknown confidence, and source
317+
taxonomy counts such as wire observation, policy inference, lab calibration,
318+
source model, route model, control mapping, and operator context.
317319

318320
## Calibration
319321

docs/CLI.md

Lines changed: 6 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -105,9 +105,12 @@ relayx schema validate -k evidence-report evidence-report.json
105105
`evidence-report` audits an existing result offline. It reports whether
106106
candidate or relayable records have evidence, whether protocol judgement
107107
records expose response classification, policy inference, and remaining
108-
uncertainty, and whether any evidence still has unknown confidence. It is meant
109-
for lab promotion review, enterprise handoff, and fixture quality checks; it
110-
does not scan, validate, relay, or mutate the result file.
108+
uncertainty, whether any evidence still has unknown confidence, and how each
109+
evidence item maps into the source taxonomy. The taxonomy separates wire
110+
observation, policy inference, lab calibration, source model, route model,
111+
control mapping, operator context, error, and unsupported-boundary evidence.
112+
It is meant for lab promotion review, enterprise handoff, and fixture quality
113+
checks; it does not scan, validate, relay, or mutate the result file.
111114

112115
## Schema Contracts
113116

docs/ENTERPRISE_OUTPUTS.md

Lines changed: 3 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -86,8 +86,9 @@ for CI and ingestion pipeline checks.
8686
`relayx quality-gate` is the local CI and release gate. It validates package
8787
metadata, schema catalog coverage, JSON fixtures, schema fixture directories,
8888
enterprise output matrix coverage, lab matrix coverage, lab stability checks,
89-
lab differential checks, evidence-report checks, documentation coverage, and
90-
GitHub Actions workflow presence. Failed gates return exit code `2`.
89+
lab differential checks, evidence-report checks, evidence source taxonomy
90+
coverage, documentation coverage, and GitHub Actions workflow presence. Failed
91+
gates return exit code `2`.
9192

9293
```bash
9394
relayx quality-gate -C . -f json -o relayx-quality-gate.json

docs/LAB_VALIDATION.md

Lines changed: 8 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -21,7 +21,8 @@ by reports, ranking, and regression tests.
2121

2222
Use `relayx evidence-report -r result.json` before promoting lab evidence. The
2323
report checks whether finding/path records carry evidence, protocol judgement
24-
fields, confidence, and remaining uncertainty in a consistent structure.
24+
fields, source taxonomy, confidence, and remaining uncertainty in a consistent
25+
structure.
2526

2627
| Oracle | Required evidence keys |
2728
| --- | --- |
@@ -137,6 +138,12 @@ offline audits of an existing result and do not prove protocol correctness; they
137138
make missing judgement fields and unknown-confidence evidence visible before a
138139
lab profile or enterprise handoff relies on the result.
139140

141+
The evidence-report source taxonomy distinguishes observed wire evidence,
142+
policy inference, lab calibration evidence, modeled source or route context,
143+
control mappings, operator context, errors, and unsupported boundaries. Lab
144+
promotion should be based on the appropriate source categories rather than a
145+
single undifferentiated evidence count.
146+
140147
For baseline comparisons, promotion requires both:
141148

142149
1. A candidate signature that matches a promotable calibrated lab state.

docs/README.fr.md

Lines changed: 6 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -85,8 +85,8 @@ echec.
8585
de policy states et separer les vrais response discriminators des champs de
8686
contexte.
8787
- Evidence completeness reporting pour auditer les records finding/path, les
88-
champs de protocol judgement, la distribution de confidence, les contract
89-
keys manquantes et l'incertitude restante.
88+
champs de protocol judgement, la source taxonomy, la distribution de
89+
confidence, les contract keys manquantes et l'incertitude restante.
9090
- Exports entreprise pour graph analysis, ingestion SIEM, revue CSV, rapports
9191
HTML/Markdown, scan diff et simulation d'impact de remediation.
9292
- Generation d'enterprise bundle avec manifest, hashes d'artefacts, schema
@@ -220,7 +220,7 @@ relayx lab-stability Evalue stabilite et drift des captures lab repetees
220220
relayx lab-diff Compare les response differentials entre policy states
221221
relayx lab-index Resume les corpuses de signatures lab
222222
relayx lab-profile Genere un draft de calibration profile depuis corpus
223-
relayx evidence-report Audite evidence completeness et champs de judgement
223+
relayx evidence-report Audite evidence completeness, source taxonomy et judgement
224224
relayx validate Lance une validation active controlee pour un chemin
225225
relayx profiles Liste les profils RelayX integres
226226
relayx export Exporte graph, JSONL, CSV, rapport ou diagramme
@@ -313,7 +313,9 @@ lorsque l'artefact ne respecte pas le contrat choisi.
313313
`relayx evidence-report -r result.json` audite un resultat existant hors ligne,
314314
sans trafic reseau. Il signale les records candidate/relayable sans evidence,
315315
les protocol judgement records sans policy inference ou remaining uncertainty,
316-
et les evidence entries qui gardent une confidence `unknown`.
316+
les evidence entries qui gardent une confidence `unknown`, et les categories de
317+
source comme wire observation, policy inference, lab calibration, source model,
318+
route model, control mapping et operator context.
317319

318320
## Calibration laboratoire
319321

docs/README.zh-CN.md

Lines changed: 6 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -73,7 +73,8 @@ validation 必须显式开启,并可能产生失败登录类遥测。
7373
- lab response differential analysis,用于比较稳定 lab policy state
7474
signature,区分真正的 response discriminator 和仅作上下文参考的字段。
7575
- evidence completeness reporting,用于审计 finding/path 证据记录、
76-
protocol judgement 字段、confidence 分布、缺失 contract key 和剩余不确定性。
76+
protocol judgement 字段、source taxonomy、confidence 分布、缺失 contract key
77+
和剩余不确定性。
7778
- graph、SIEM、CSV、HTML/Markdown report、scan diff、remediation impact
7879
simulation 等企业输出。
7980
- enterprise bundle 生成,包含 manifest、artifact hash、schema status、
@@ -204,7 +205,7 @@ relayx lab-stability 评估重复 lab capture 的稳定性和漂移
204205
relayx lab-diff 比较稳定 lab policy-state response differential
205206
relayx lab-index 汇总 lab signature corpus
206207
relayx lab-profile 从 corpus 生成 calibration profile 草案
207-
relayx evidence-report 审计 evidence completeness 和 judgement 字段
208+
relayx evidence-report 审计 evidence completeness、source taxonomy 和 judgement 字段
208209
relayx validate 对单条 path 运行受控 active validation
209210
relayx profiles 列出内置 RelayX profile
210211
relayx export 导出 graph、JSONL、CSV、report 或 diagram 产物
@@ -292,7 +293,9 @@ artifact 不符合所选 contract 时返回 exit code `2`。
292293
`relayx evidence-report -r result.json` 会离线审计已有 result,不产生网络流量。
293294
它会标出缺少 evidence 的 candidate/relayable record、缺少 policy inference 或
294295
remaining uncertainty 的 protocol judgement record,以及仍为 unknown confidence
295-
的 evidence entry。
296+
的 evidence entry,并统计 wire observation、policy inference、lab calibration、
297+
source model、route model、control mapping 和 operator context 等 evidence source
298+
类别。
296299

297300
## 实验室校准
298301

docs/ROADMAP.md

Lines changed: 6 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -40,6 +40,9 @@ uncertain, and what an operator is allowed to do next.
4040
- Evidence completeness reporting for existing results, including finding/path
4141
evidence counts, confidence distribution, protocol judgement fields, missing
4242
contract keys, and remaining uncertainty.
43+
- Evidence source taxonomy for separating wire observations, policy
44+
inferences, lab-calibration evidence, source and route models, control
45+
mappings, operator context, errors, and unsupported boundaries.
4346
- Guarded validation and controlled execution state machines with dry-run,
4447
armed, and confirmed modes.
4548
- Execution module inventory, module planning, and Adapter SDK dispatch,
@@ -83,9 +86,9 @@ Near-term work:
8386
- Expand real HTTP/IIS EPA, AD CS Web Enrollment EPA, LDAP signing, LDAPS CBT,
8487
and MSSQL encryption/EPA response-difference captures beyond the bundled
8588
synthetic fixture corpus.
86-
- Expand evidence-report source taxonomy so protocol judgements consistently
87-
distinguish observed wire evidence, inferred policy state, lab-calibrated
88-
promotion evidence, and operator-supplied context.
89+
- Extend the source taxonomy with real lab-corpus provenance, endpoint build
90+
metadata, and operator-reviewed promotion decisions once those artifacts are
91+
available from authorized labs.
8992
- Keep synthetic authentication rejection states subdivided without treating
9093
invalid-credential rejection as proof of relayability.
9194

docs/SCHEMA.md

Lines changed: 11 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -46,8 +46,8 @@ Machine-readable validation reports use `--format json`.
4646
- `lab-differential`: lab response differential report comparing stable
4747
policy-state signatures.
4848
- `evidence-report`: offline result evidence completeness report for
49-
finding/path records, protocol judgement fields, confidence, and remaining
50-
uncertainty.
49+
finding/path records, source taxonomy, protocol judgement fields, confidence,
50+
and remaining uncertainty.
5151
- `execution-record`: guarded validation or controlled execution audit record.
5252
- `module-manifest`: JSON manifest for execution adapter capability planning.
5353
- `opsec-policy`: OPSEC policy used by validation, execution, and source
@@ -84,6 +84,15 @@ judgement records are expected to expose response classification, policy
8484
inference, and remaining uncertainty; missing fields are reported as warnings
8585
so teams can decide whether lab promotion or handoff is ready.
8686

87+
Evidence reports also include a source taxonomy. Each evidence item is mapped
88+
to a category such as `wire_observation`, `policy_inference`,
89+
`lab_calibration`, `source_model`, `route_model`, `control_mapping`,
90+
`operator_context`, `error`, or `unsupported`, plus a judgement role such as
91+
`observation`, `response_semantics`, `policy_inference`, or
92+
`uncertainty_boundary`. This lets teams separate what RelayX observed on the
93+
wire from what it inferred, modeled, calibrated, or received as operator
94+
context.
95+
8796
## Lab Confidence Contract
8897

8998
Calibration reports, baseline comparisons, and lab verification reports include

docs/TUTORIAL.md

Lines changed: 3 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -128,8 +128,9 @@ relayx -q schema validate -k evidence-report /tmp/relayx-evidence-report.json
128128
```
129129

130130
The evidence report is offline. It highlights records that lack evidence,
131-
protocol judgement fields, remaining uncertainty, or confidence assignments
132-
before the result is used for lab promotion or enterprise handoff.
131+
protocol judgement fields, remaining uncertainty, confidence assignments, or
132+
clear source taxonomy before the result is used for lab promotion or
133+
enterprise handoff.
133134

134135
## 3. Review Route And Pivot Awareness
135136

examples/tutorial/baseline-result.json

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -78,7 +78,7 @@
7878
"started_at": "2026-06-06T08:00:00+00:00",
7979
"target_count": 2,
8080
"tool": "RelayX",
81-
"version": "0.1.11"
81+
"version": "0.1.12"
8282
},
8383
"paths": [
8484
{

0 commit comments

Comments
 (0)