todo · 129/129 files · 1,841 functions · live from the analyzer’s own telemetry

index built 2026-09-13 22:45 UTC · supplied export passed graph validation; producer freshness is caller-supplied

129 files analysed of 2,302 in the tree — 3,251 paths excluded (2,015 skipped directory, 154 unsupported extension, 4 untracked) · 52 production files in the repository. Every percentage below is against the analysed set, not the repository.

EVIDENCE INCOMPLETE · 31 complete · 58 partial · 2 N/A

Inventory complete. Evidence coverage is not.

All 91 layer questions are represented, but only 31 have complete positive evidence; 58 remain partial. An explicit unknown is an honest answer row, not proof that the code is understood or correct.

layer, as written in the plan

triangulated

density

covered

of its own base

Ast

AST++ function inventory

100.0% answer · 100.0% evidence

—

0.1facts each

1,841of 1,841

Scope assignments

AST++ module/class namespace assignment inventory

100.0% answer · 100.0% evidence

—

1.0facts each

123of 123

Call graph

call_graph caller/callee endpoints

100.0% answer · 98.3% evidence

78–98%5 counts

1.9facts each

1,810of 1,84131 none found1,069 open questions

Type graph

type_evidence owners

100.0% answer · 100.0% evidence

100–100%5 counts

5.5facts each

1,841of 1,8411,104 open questions

Type surface

AST++/type-graph function type-inspection owners

100.0% answer · 100.0% evidence

—

1.0facts each

1,841of 1,841166 open questions

Type closure

AST++ function types with all source-visible producers accounted for

one or more source-visible type producers remain unresolved; closure status remains unknown

100.0% answer · 92.7% evidence

—

1,707of 1,841134 unknown

Data flow

source-linked data_flow owners

100.0% answer · 100.0% evidence

100–100%5 counts

11.8facts each

1,841of 1,84112 open questions

Data flow surface

AST++/data-flow function inspection owners

100.0% answer · 100.0% evidence

—

1.0facts each

1,841of 1,8411,504 open questions

Non turing resources

resource_effects owners

no catalogued non-Turing resource effect was observed; status remains unknown

100.0% answer · 50.5% evidence

50–68%5 counts

19.3facts each

929of 1,841753 unknown159 none found

Resource surface

AST++ function direct-resource surface inventory owners

no direct resource effect was observed; unresolved callee effects remain unknown

100.0% answer · 48.1% evidence

—

2.1facts each

886of 1,841796 unknown159 none found796 open questions

Resource lifecycle

source-linked lifecycle owners: managed-context, explicit-cleanup, API-completion, or lazy-boundary facts; subtype owner counts are non-additive because one owner may have several lifecycle kinds

no source-linked lifecycle fact was observed; status remains unknown

made of 506 api completion, 196 managed context, 24 lazy api boundary, 1 explicit cleanup; an owner counted under more than one kind appears in more than one number, so these do not sum to the bar

every owner is one of 541 test or 83 production

100.0% answer · 33.9% evidence

—

2.0facts each

624of 1,8411,058 unknown159 none found

Resource lifecycle surface

AST++ function direct-lifecycle surface inventory owners

no direct lifecycle construct was observed; unresolved callee lifecycle remains unknown

100.0% answer · 47.2% evidence

—

2.1facts each

869of 1,841813 unknown159 none found813 open questions

Entry pointsclassification

resolved entry_symbols plus source-linked entry-point inspection

100.0% answer · 61.1% evidence

—

1.0facts each

1,124of 1,841717 none found

Entry point surface

AST++ function entry-point inspection owners

100.0% answer · 61.1% evidence

—

1.6facts each

1,124of 1,841717 none found717 open questions

Reachability

resolved reachable and known-unreachable symbols

100.0% answer · 80.6% evidence

81–91%5 counts

1.0facts each

1,484of 1,841357 none found60 bounded candidate

Reachability surface

AST++ function reachability inspection owners

100.0% answer · 100.0% evidence

—

1.0facts each

1,841of 1,84160 bounded candidate

Exception flow

exception_edges owners

no source-linked exception edge contract was produced; status remains unknown

100.0% answer · 31.0% evidence

31–51%5 counts

2.5facts each

570of 1,8411,132 unknown139 none found

Exception surface

AST++ function exception-surface inventory owners

no source-linked exception-surface inspection was produced; status remains unknown

100.0% answer · 7.0% evidence

—

14.4facts each

128of 1,8411,580 unknown133 none found1,580 open questions

Exception topology

AST++ try/handler topology owners

100.0% answer · 7.1% evidence

—

1.1facts each

131of 1,8411,710 none found

Exception flow surface

AST++ function exception-flow inspection owners

no source-linked exception-flow inspection was produced; status remains unknown

100.0% answer · 56.1% evidence

—

1.8facts each

1,033of 1,841669 unknown139 none found669 open questions

Colouringclassification

function_colors rows

100.0% answer · 100.0% evidence

—

1.0facts each

1,841of 1,841

Side effect surface

function_colors side_effect entries over function symbols

100.0% answer · 68.7% evidence

—

22.5facts each

1,264of 1,841577 none found

Dead code review

unreachable_known_symbols review queue

357 of 357 unreachable known function symbols — 19.4% of the 1,841-function corpus

100.0% answer · 100.0% evidence

—

0.9facts each

357of 35729 bounded candidate

Dead code surface

AST++ function conservative dead-code review owners

100.0% answer · 100.0% evidence

—

1.0facts each

1,841of 1,841311 open questions60 bounded candidate

Test provenance surface

AST++ function static/runtime test provenance owners

100.0% answer · 99.7% evidence

—

1.0facts each

1,836of 1,8415 none found4 open questions

Test fixture realism surface

AST++ test-function test-double/production-boundary realism owners

145 of 145 eligible test functions with substitutions — 7.9% of the 1,841-function corpus

100.0% answer · 100.0% evidence

—

7.5facts each

145of 145145 open questions

Security

api_status security_risk rows

no catalogued security-risk API use was observed; status remains unknown

100.0% answer · 3.3% evidence

—

1.2facts each

60of 1,8411,642 unknown139 none found8 open questions

Deprecations

api_status lifecycle_status rows

no catalogued API lifecycle status was observed; status remains unknown

100.0% answer · 0.0% evidence

—

0of 1,8411,252 unknown589 none found

Deprecation markers

AST++ lexical deprecation markers in function comments/docstrings

100.0% answer · 0.2% evidence

—

1.3facts each

3of 1,8411,838 none found

Api status surface

AST++/API hyperlexicon function status-inspection owners

no catalogued API security or lifecycle status was observed; status remains unknown

100.0% answer · 3.3% evidence

—

30.7facts each

60of 1,8411,642 unknown139 none found1,305 open questions

Deprecated or removed

api_status lifecycle_status in deprecated/removed

no deprecated or removed lifecycle status was observed; status remains unknown

100.0% answer · 0.0% evidence

—

0of 1,8411,252 unknown589 none found

Contracts

function contracts over source-derived contract-mechanism candidates

no source-linked function contract was indexed; status remains unknown

288 of 406 eligible function symbols with contract mechanisms — 15.6% of the 1,841-function corpus

100.0% answer · 70.9% evidence

15–71%5 counts

1.1facts each

288of 406118 unknown

Contract surface

AST++ function-contract inspection owners

no source-linked function-contract inspection was produced; contract status remains unknown

100.0% answer · 64.0% evidence

—

1.6facts each

1,179of 1,841559 unknown103 none found1,441 open questions

Purity surface

AST++ function purity/state-effect inspection owners

no source-linked purity/state-effect inspection was produced; purity status remains unknown

100.0% answer · 29.9% evidence

—

3.3facts each

550of 1,8411,223 unknown68 none found1,223 open questions

State effect surface

AST++ function direct/inferred state-effect inspection owners

no source-linked state-effect inspection was produced; unresolved delegated effects remain unknown

100.0% answer · 56.3% evidence

—

1.8facts each

1,037of 1,841689 unknown115 none found689 open questions

Reference surface

AST++ function reference/access inspection owners

100.0% answer · 99.7% evidence

—

1.0facts each

1,836of 1,8415 none found5 open questions

Patterns

explicit pattern bindings or source-declared roles over class symbols

no explicit pattern binding targets this class; adherence remains unknown

100.0% answer · 0.3% evidence

—

1of 331330 unknown

Pattern validation

explicit pattern targets with deterministic template-validator outcomes

100.0% answer · 100.0% evidence

—

1of 1

Pattern template catalog

formal pattern-template validator definitions in the QA catalog

100.0% answer · 100.0% evidence

—

1.0facts each

7of 7

Pattern surface

AST++ class pattern-template inspection owners

100.0% answer · 0.3% evidence

—

331.0facts each

1of 331330 none found331 open questions

Purity state effects

function_purity joined by file/name/line

no source-linked purity fact joined to this function; status remains unknown

100.0% answer · 29.9% evidence

30–33%5 counts

1.0facts each

550of 1,8411,291 unknown

State written where

flow_effects self-field mutation owners over direct AST self-field mutation candidates

56 of 56 eligible function symbols with self field mutations — 3.0% of the 1,841-function corpus

100.0% answer · 100.0% evidence

62–100%5 counts

132.4facts each

56of 56

State container mutations

flow_effects self-container mutation owners over direct AST self-container mutation candidates

25 of 25 eligible function symbols with self container mutations — 1.4% of the 1,841-function corpus

100.0% answer · 100.0% evidence

—

37.8facts each

25of 25

State argument mutations

flow_effects parameter-container mutations over direct AST parameter-container mutation candidates

5 of 5 eligible function symbols with parameter container mutations — 0.3% of the 1,841-function corpus

100.0% answer · 100.0% evidence

—

139.6facts each

5of 5

State global mutations

flow_effects explicit global writes over direct AST global-write candidates

no eligible function symbols with explicit global writes exist in this corpus — nothing to find, rather than nothing found

N/A

—

0of 0

State nonlocal mutations

flow_effects explicit nonlocal writes over direct AST nonlocal-write candidates

no eligible function symbols with explicit nonlocal writes exist in this corpus — nothing to find, rather than nothing found

N/A

—

0of 0

Foreign written where

flow_effects foreign-write owners over AST++ foreign-write writers

171 of 171 eligible function symbols — 9.3% of the 1,841-function corpus

100.0% answer · 100.0% evidence

—

8.8facts each

171of 171

Ownership

ownership over source-definable self/foreign writes and self-field control reads

290 of 290 eligible function symbols with source definable ownership — 15.8% of the 1,841-function corpus

100.0% answer · 100.0% evidence

—

6.3facts each

290of 290

Ownership surface

AST++ function ownership/control inspection owners

no source-linked ownership/control inspection was produced; classification status remains unknown

100.0% answer · 36.1% evidence

—

2.8facts each

665of 1,8411,073 unknown103 none found1,409 open questions

Requirement surface

AST++ function comment/docstring requirement inspection owners

100.0% answer · 74.9% evidence

—

1.3facts each

1,378of 1,841463 none found7 open questions

Documentation surface

AST++ function objective comment-shape inspection owners

100.0% answer · 7.3% evidence

—

13.7facts each

134of 1,8411,707 none found

Name surface

AST++ function symbol-name shape inspection owners

100.0% answer · 100.0% evidence

—

1.0facts each

1,841of 1,841

Call graph surface

AST++ function resolved/unresolved call inspection owners

100.0% answer · 100.0% evidence

—

1.0facts each

1,841of 1,8411,338 open questions

Reference access

AST++ source-linked name/reference access owners

100.0% answer · 99.7% evidence

—

16.7facts each

1,836of 1,8415 none found

Concurrency

source-proven async and catalogued concurrency observations

no source-proven async or catalogued concurrency observation was produced; status remains unknown

100.0% answer · 2.4% evidence

—

1.2facts each

44of 1,8411,637 unknown160 none found

Concurrency surface

AST++ function async/concurrency inspection owners

no source-linked async/concurrency observation was produced; status remains unknown

100.0% answer · 28.2% evidence

—

3.5facts each

520of 1,8411,161 unknown160 none found1,161 open questions

Environment platform

source-linked environment/platform effects

no source-linked environment/platform effect was produced; status remains unknown

100.0% answer · 35.6% evidence

—

5.0facts each

656of 1,8411,025 unknown160 none found

Environment surface

AST++ function environment/platform inspection owners

no source-linked environment/platform inspection was produced; status remains unknown

100.0% answer · 35.6% evidence

—

2.8facts each

656of 1,8411,025 unknown160 none found1,025 open questions

Configuration surface

AST++ function configuration/feature-state inspection owners

no source-linked configuration/feature-state inspection was produced; mode status remains unknown

100.0% answer · 1.0% evidence

—

96.9facts each

19of 1,8411,662 unknown160 none found1,662 open questions

Deployment surface

deployment artifacts and non-test entry topology units

no source-linked deployment artifact or entry-topology inspection was produced; runtime topology remains unknown

36 of 39 deployment units — 2.0% of the 1,841-function corpus

100.0% answer · 92.3% evidence

—

1.1facts each

36of 393 unknown3 open questions

Privacy surface

AST++ function data-classification/privacy inspection owners

no source-linked data-classification/privacy inspection was produced; handling obligations remain unknown

100.0% answer · 94.9% evidence

—

1.1facts each

1,747of 1,84136 unknown58 none found1,783 open questions

State recovery surface

AST++ function state-machine/failure-recovery inspection owners

no source-linked state-machine/failure-recovery inspection was produced; transition and recovery behavior remain unknown

100.0% answer · 2.3% evidence

—

43.8facts each

42of 1,8411,684 unknown115 none found1,684 open questions

Schema entity

source-proven schema/entity definitions over class symbols

no source-proven schema/entity definition was produced; status remains unknown

100.0% answer · 1.5% evidence

—

1.0facts each

5of 331326 unknown

Schema surface

AST++ class schema/entity inspection owners

100.0% answer · 1.5% evidence

—

66.2facts each

5of 331326 none found326 open questions

Class body surface

AST++ class-namespace assignment and catalogued-effect inspection owners

100.0% answer · 8.5% evidence

—

3.4facts each

28of 331303 none found

Dynamic binding

source-linked reflection/dynamic-binding observations

no source-linked reflection or dynamic-binding observation was produced; status remains unknown

100.0% answer · 1.6% evidence

—

1.1facts each

29of 1,8411,679 unknown133 none found

Dynamic binding surface

AST++ function dynamic-binding inspection owners

no source-linked dynamic-binding inspection was produced; status remains unknown

100.0% answer · 1.6% evidence

—

63.5facts each

29of 1,8411,679 unknown133 none found1,679 open questions

Runtime evidence

executed coverage lines joined to AST++ function spans

no source-linked execution observation was produced; unobserved code is not safe or dead

100.0% answer · 99.7% evidence

—

1.0facts each

1,836of 1,8415 unknown

Runtime evidence surface

AST++ function runtime-evidence inspection owners

no source-linked runtime-evidence inspection was produced; artifact availability remains explicit

100.0% answer · 99.7% evidence

—

1.0facts each

1,836of 1,8415 unknown

Symbol scope

AST++ lexical function scope owners

100.0% answer · 100.0% evidence

—

7.8facts each

1,841of 1,841

Control flow

AST++ explicit CFG function rows

100.0% answer · 100.0% evidence

—

19.4facts each

1,841of 1,841

Taint trust

source/sanitizer/sink/guard trust graph rows

no source-linked trust graph fact was produced; status remains unknown

100.0% answer · 94.9% evidence

—

13.3facts each

1,747of 1,84194 unknown

Taint trust surface

AST++ function taint/trust inspection owners

no source-linked taint/trust inspection was produced; status remains unknown

100.0% answer · 94.9% evidence

—

1.1facts each

1,747of 1,84194 unknown36 open questions

Authorization surface

AST++ function authorization/policy inspection owners

no source-linked authorization/policy inspection was produced; policy status remains unknown

100.0% answer · 0.0% evidence

—

0of 1,8411,726 unknown115 none found1,726 open questions

Transaction surface

AST++ function transaction/consistency inspection owners

no source-linked transaction/consistency inspection was produced; consistency status remains unknown

100.0% answer · 0.4% evidence

—

230.1facts each

8of 1,8411,718 unknown115 none found1,718 open questions

Observability surface

AST++ function observability/KPI inspection owners

no source-linked observability/KPI inspection was produced; measurement status remains unknown

100.0% answer · 21.2% evidence

—

4.7facts each

390of 1,8411,335 unknown116 none found1,335 open questions

Performance surface

AST++ function performance/resource-budget inspection owners

no source-linked performance/resource-budget inspection was produced; budget status remains unknown

100.0% answer · 0.9% evidence

—

115.1facts each

16of 1,8411,665 unknown160 none found1,665 open questions

Boundary protocol

source-proven ingress, serialization, and schema rows

no source-proven ingress, protocol, or schema fact was produced; status remains unknown

100.0% answer · 64.0% evidence

—

1.3facts each

1,178of 1,841663 unknown99 open questions

Boundary surface

AST++ function boundary/protocol inspection owners

no source-linked boundary/protocol inspection was produced; status remains unknown

100.0% answer · 64.0% evidence

—

1.6facts each

1,178of 1,841539 unknown124 none found539 open questions

Grammar dispatch

source-linked grammar action-to-target-type bindings

100.0% answer · 100.0% evidence

—

1.0facts each

124of 124

Grammar target resolution

grammar bindings whose target names resolve to one indexed source class

one or more grammar target names do not resolve to exactly one indexed source class; grammar interpretation remains unknown

100.0% answer · 99.2% evidence

—

1.0facts each

123of 1241 unknown

Repository contextclassification

repository_context rows with filesystem provenance

100.0% answer · 100.0% evidence

—

1.0facts each

3,389of 3,389

Artifact rolesclassification

one role answer per inventory file; unknown is explicit

100.0% answer · 100.0% evidence

—

1.0facts each

2,302of 2,302

Names lexiconclassification

filesystem name tokens

100.0% answer · 100.0% evidence

—

1.0facts each

3,389of 3,389

Documentation

one documentation row per analyzed production file; external source/module references, normalized basename matches, or a source module docstring are evidence, unmatched is explicit

100.0% answer · 100.0% evidence

—

29.8facts each

52of 52

Dependency graph

import edges classified by local/stdlib/third-party evidence

one or more third-party import edges have no matched parsed dependency declaration; status remains unknown

100.0% answer · 84.2% evidence

—

1.2facts each

592of 703111 unknown112 open questions

Test topology

one topology row per executable test file; support artifacts remain inventory facts

100.0% answer · 100.0% evidence

—

1.1facts each

73of 731,267 open questions

Change historyclassification

one git-history row per analyzed file; unavailable history is explicit

100.0% answer · 100.0% evidence

—

1.0facts each

129of 129

Build runtimeclassification

recognized build, CI, container, and runtime configuration artifacts

100.0% answer · 100.0% evidence

—

2.5facts each

2of 21 open questions

Requirementsclassification

source comment/docstring story tags linked to non-test code symbols

one or more story requirements have no non-test code link; status remains unknown

100.0% answer · 59.5% evidence

—

10.4facts each

25of 4217 unknown17 open questions

Test resultsclassification

JUnit testcase identities; one artifact-status probe when no rows exist

100.0% answer · 100.0% evidence

—

1.0facts each

1,054of 1,05436 open questions

strong partial under 5% 100% by definition — a row exists, not an answer a queue, not knowledge

layers with complete
positive evidence

layers with partial
positive evidence

layers with no applicable
denominator

33,108

open questions it has
written down as unresolved

Test outcomes: 1,049 passed · 5 failed — 5 failed or errored. The 100% test-results bar means the JUnit identities were parsed; it is not a pass-rate or a test-quality score.

Requirement traceability: 25 of 42 requirements have explicit production-code links; 17 are tests-only. The export contains 2,309 conservative test-target candidates (726 for tests-only requirements) as review context. Candidates are not code links, so 17 requirements remain unresolved.

Carried questions: 336 open questions (42 acceptance authority · 42 business outcome · 42 implementation binding · 42 product surface · 42 requirement definition · 42 telemetry definition · 42 test examination · 42 validation sufficiency); 4,292 observed evidence bindings (917 assertion observes production · 1 production exposes behavior · 121 requirement examined by · 944 test has assertion · 2,309 test reaches production). Bindings show where to inspect; they do not claim that production code implements the requirement. Acceptance: 0 attributed acceptance records; no requirement is accepted by this export. Acceptance remains a separate human or LLM decision.

same layers, over time

trend

now

change

Ast

100.0%

▲ +0.1counts, rebased

Scope assignments

23.5%

▲ +1.0counts, rebased

Call graph

98.3%

▲ +16.7counts, rebased

Type graph

100.0%

▲ +67.7counts, rebased

Type surface

100.0%

▲ +2.4counts, rebased

Type closure

100.0%

▲ +0.6counts, rebased

Data flow

100.0%

▲ +98.3counts, rebased

Data flow surface

100.0%

▲ +4.1counts, rebased

Non turing resources

100.0%

▲ +24.5counts, rebased

Resource surface

100.0%

▲ +79.8counts, rebased

Resource lifecycle

100.0%

▲ +69.1counts, rebased

Resource lifecycle surface

100.0%

▲ +77.8counts, rebased

Entry points

100.0%

▲ +97.0counts, rebased

Entry point surface

100.0%

▲ +1.8counts, rebased

Reachability

100.0%

▲ +94.7counts, rebased

Reachability surface

100.0%

▲ +5.6counts, rebased

Exception flow

100.0%

▲ +85.4counts, rebased

Exception surface

100.0%

▲ +0.8counts, rebased

Exception topology

100.0%

▲ +36.6counts, rebased

Exception flow surface

100.0%

▲ +87.7counts, rebased

Colouring

100.0%

▲ +0.1counts, rebased

Side effect surface

68.7%

▼ -31.3counts, rebased

Dead code review

21.4%

▼ -65.5counts, rebased

Dead code surface

100.0%

▲ +5.6counts, rebased

Test provenance surface

100.0%

▲ +100.0counts, rebased

Test fixture realism surface

100.0%

—counts, rebased

Security

100.0%

▲ +48.3counts, rebased

Deprecations

0.0%

—counts, rebased

Deprecation markers

33.3%

▼ -66.7counts, rebased

Api status surface

100.0%

▲ +48.3counts, rebased

Deprecated or removed

0.0%

—counts, rebased

Contracts

95.7%

▲ +1.0counts, rebased

Contract surface

99.8%

▲ +75.7counts, rebased

Purity surface

100.0%

▲ +1.6counts, rebased

State effect surface

100.0%

▲ +66.5counts, rebased

Reference surface

100.0%

▲ +2.2counts, rebased

Patterns

100.0%

▲ +100.0counts, rebased

Pattern validation

100.0%

—counts, rebased

Pattern template catalog

100.0%

—

Pattern surface

100.0%

▲ +100.0counts, rebased

Purity state effects

100.0%

▲ +1.6counts, rebased

State written where

16.1%

▼ -83.9counts, rebased

State container mutations

100.0%

▲ +8.0counts, rebased

State argument mutations

100.0%

—

State global mutations

0.0%

—

State nonlocal mutations

0.0%

—

Foreign written where

100.0%

▲ +4.7counts, rebased

Ownership

78.4%

▲ +31.4counts, rebased

Ownership surface

100.0%

▲ +70.7counts, rebased

Requirement surface

100.0%

▲ +3.7counts, rebased

Documentation surface

99.3%

▲ +3.7counts, rebased

Name surface

100.0%

▲ +1.7counts, rebased

Call graph surface

100.0%

▲ +8.9counts, rebased

Reference access

100.0%

▲ +2.2counts, rebased

Concurrency

100.0%

▲ +9.1counts, rebased

Concurrency surface

100.0%

▲ +92.3counts, rebased

Environment platform

100.0%

▲ +40.2counts, rebased

Environment surface

100.0%

▲ +40.2counts, rebased

Configuration surface

100.0%

▲ +21.1counts, rebased

Deployment surface

92.3%

—

Privacy surface

100.0%

▲ +100.0counts, rebased

State recovery surface

100.0%

▲ +11.9counts, rebased

Schema entity

20.8%

▲ +4.2counts, rebased

Schema surface

20.8%

▲ +4.2counts, rebased

Class body surface

93.3%

▲ +10.0counts, rebased

Dynamic binding

100.0%

▲ +6.9counts, rebased

Dynamic binding surface

100.0%

▲ +6.9counts, rebased

Runtime evidence

100.0%

▲ +100.0counts, rebased

Runtime evidence surface

100.0%

▲ +100.0counts, rebased

Symbol scope

100.0%

▲ +1.8counts, rebased

Control flow

100.0%

▲ +1.8counts, rebased

Taint trust

100.0%

▲ +76.6counts, rebased

Taint trust surface

100.0%

▲ +76.4counts, rebased

Authorization surface

0.0%

—counts, rebased

Transaction surface

29.6%

—counts, rebased

Observability surface

100.0%

▲ +99.0counts, rebased

Performance surface

100.0%

—counts, rebased

Boundary protocol

100.0%

▲ +2.3counts, rebased

Boundary surface

100.0%

▲ +1.8counts, rebased

Grammar dispatch

100.0%

—

Grammar target resolution

99.2%

▲ +2.4

Repository context

86.5%

▼ -6.1counts, rebased

Artifact roles

82.7%

▼ -9.6counts, rebased

Names lexicon

86.5%

▼ -6.1counts, rebased

Documentation

100.0%

▲ +3.8counts, rebased

Dependency graph

86.5%

▼ -13.5counts, rebased

Test topology

93.6%

▼ -6.4counts, rebased

Change history

100.0%

▲ +2.3counts, rebased

Build runtime

100.0%

—

Requirements

64.1%

▼ -35.9counts, rebased

Test results

100.0%

▲ +100.0counts, rebased

A falling line is not necessarily a regression. When the analyzer finds it has been over-claiming, the honest fix lowers the number. That has happened: a soundness audit found bare method() calls were being resolved as self.method(), and removing those unproven edges took state-written-where from 19.2% to 4.4% and reachability from 72.7% to 66.4%. Three quarters of what one layer claimed to know had been inferred through calls the source never proved. Read a drop by asking which commits landed in that window before assuming it is damage — each reading below records the analyzer commit it was taken at.

Why some lines are drawn from counts. qa rebuilt this measurement scheme repeatedly while the history accumulated — the layer set went from 17 to 68, and 31 of 68 layers changed denominator, several more than once. A percentage over a moving base is not comparable to itself, so those rows are marked counts, rebased and plot the covered count against its own peak. That curve moves when work happens and stays flat when only the ruler changes.

Same layers, same order — a line instead of a bar. One reading per rebuild of the analyzer’s index, so a re-run against unchanged code cannot invent a data point. Each chart is scaled 0–100% so the rows are comparable to each other.

316 readings so far, first at 2026-08-28 15:38 UTC. History starts when this page did — there is no back-fill, because earlier numbers were computed a different way and splicing them in would draw a trend that never happened.

how to read it

Three of these are not what they look like

Two layers sit at 100% by definition. “The code is read” and “every function tagged” count the inventory against itself — every function is in the list of functions. They mean a row exists, not that anything is understood.

The review queue is worth less than it reads. Checking its entries against what actually runs at runtime showed roughly three quarters were live code. High coverage of a question nobody can afford to answer.

The last-place layers are the commercial ones — untrusted input reaching danger, dangerous libraries, obsolete libraries, and what the code is meant to be. Those catch the expensive mistakes, and the climb has barely begun.

Five layers are classifications, and the analyzer now says so itself. A classification answers “what is this” with one row per thing — a file has one role, a symbol one colour — so its 100% is a different achievement from an analysis at 100%. Those rows are labelled and greyed. The density column shows facts per covered unit for everything else: control flow carries 19.4, resources 25.2, taint 8.1. A low density on an analysis layer is the thing to look at — reachability and purity both sit at 1.0.

The range is five independent counts, not one. qa answers per function, but a function is a poor unit of size — a three-line __repr__ weighs the same as a two-hundred-line handler. So the same coverage is recounted straight from the source four more ways: lines of function body, statements, AST nodes, and branch points. The column shows the full range; hover it for each figure. A narrow range means the answer is robust to how you count. A wide one means the layer's coverage is skewed — errors reach 4% of functions but up to 12% of the code, so what it misses is small functions; resources span 38–55% for the same reason.

These bars measure completeness, not usefulness — and that is a known gap, not an oversight. A layer at 94% has an answer for 94% of the base. It says nothing about whether those answers are right. Nothing on this page is a precision figure. The plan now names a ground-truth and evaluation layer — human labels, validator decisions, false-positive reasons, precision and recall — as the highest product priority, in its own words “because it keeps the green bars honest”. Until it exists, the only precision numbers anyone has here are hand-adjudicated: 41 of 43 on the largest severe block, 2 of 2 on architecture findings, 4 of 7 on SQL, checked by reading the flagged code.

Every gap here is a stated unknown, not a blind spot. The bars show evidence coverage — how much of each base has a source-linked answer. Measured the other way, answer coverage is 100% on every layer: every function has a determinate result, and where that result is "unknown" the analyzer records why. The reason is printed under each row. A tool that reports nothing found and a tool that reports why it could not tell are not the same tool, and this is the second.

The rows do not share a denominator. Most are a fraction of the 1,841 functions, but repository topology and vocabulary are over all paths, artifact roles over files, documentation over production files only, dependencies over import edges, and test topology over tests. Each row states its own base, because a single one would be a lie about most of them.

Two numbers read worse than reality. “What actually runs” reflects an unfinished call map rather than dead code — the test suite executes far more than it reports. And “where the program starts” is low by nature: a program has few entry points.

Build time

Whole export run

—

last 164s

Semantic index start → document build

84s

last 84s

The whole-run duration is not in the artifact. qa’s telemetry declares a duration_ms field and hardcodes it to None, so a reading published from their export carries no run time at all. Row 1 is populated only when this page ran the export itself, or when qa reported a figure by hand, and 2 of 4 recorded exports have one. I have asked them to populate the field; every export would then be a measurement instead of an anecdote.

Row 2 is a phase, not the run. It is the interval between built_at, stamped when qa constructs the semantic index, and generated_at, stamped when the export document starts being assembled. It covers index construction and every pass after it. It excludes discovery, parsing and AST work before it, and document assembly, encoding and the write after it. It is charted because it is the only duration every artifact can support. An export served from a cached index is skipped rather than plotted, because then the interval measures the age of the cache and not any work done.

Generated 2026-09-13 22:59 UTC from layer_coverage in the analyzer’s graph export · built by qa at 48eeac7b, analysis and output code unchanged through efc5882c
Source: qa-telemetry-v1 · semantic index built 2026-09-13 22:45 UTC — supplied export passed graph validation; producer freshness is caller-supplied. An export served from an existing cache is not evidence of a code change.
Coverage means a source-linked answer exists — not that it is safe, correct, or confirmed.
Regenerate: uv run python tools/qa_coverage_page.py in the Cascade repo.