Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
54 commits
Select commit Hold shift + click to select a range
888ab8b
Add deferred tool schema loading
kwakayama Jul 30, 2026
2a12ce7
Secure hosted deferred tool loading
kwakayama Jul 30, 2026
1e4c47e
Benchmark deferred tool schema loading
kwakayama Jul 30, 2026
b708e1c
Document deferred tool schema loading
kwakayama Jul 30, 2026
c8eb920
Fix eager tool exposure in child forks
kwakayama Jul 30, 2026
38ffd2b
Tighten deferred tool loading review gates
kwakayama Jul 30, 2026
80ceed6
Close deferred tool loading review findings
kwakayama Jul 30, 2026
c092bc7
Enforce native deferred tool authority
kwakayama Jul 30, 2026
d4a8560
Fix deferred loading compatibility fixtures
kwakayama Jul 30, 2026
44c7394
Replace flattened runtime tool inventory
kwakayama Jul 30, 2026
ffe38c7
Close progressive tool loading review gaps
kwakayama Jul 30, 2026
82d2c23
Resolve progressive loading review blockers
kwakayama Jul 30, 2026
c8b4755
Track benchmark provenance in clean checkouts
kwakayama Jul 30, 2026
acadaa9
Align run event writer token contract
kwakayama Jul 30, 2026
a2641bf
Add benchmark request context harness
kwakayama Jul 30, 2026
2441ce8
Refresh clean live tool-loading evidence
kwakayama Jul 30, 2026
f66f3bd
Gate provider-native tool search rollout
kwakayama Jul 30, 2026
d3cde2e
Refresh attested tool-loading evidence
kwakayama Jul 30, 2026
dc3ccad
test(eval): refresh live tool-loading evidence
kwakayama Jul 30, 2026
57cd5ae
test(eval): attest collision-safe redaction
kwakayama Jul 30, 2026
06fb4f8
test(eval): freeze final privacy evidence
kwakayama Jul 30, 2026
d917fcd
refactor(agent): narrow deferred tool loading to framework durability
kwakayama Jul 31, 2026
cdf61c5
fix(agent): accept toolLoading markdown frontmatter
kwakayama Jul 31, 2026
8bf92dd
test(agent): retain deterministic tool search cases
kwakayama Jul 31, 2026
a7ef8e7
feat(agent): execute deferred tools through framework search
kwakayama Jul 31, 2026
d43cd7a
test(agent): prove deferred tool authority catalogs
kwakayama Jul 31, 2026
74f68de
feat(agent): persist tool exposure before continuation
kwakayama Jul 31, 2026
843bbdf
fix(agent): preserve valid checkpoint tool ids
kwakayama Jul 31, 2026
54ac13e
feat(agent): restore tool exposure through trusted hosted runs
kwakayama Jul 31, 2026
302fe22
test(agent): align run event writer contract fixture
kwakayama Jul 31, 2026
b28f9b9
fix(agent): bind writer token to configured service
kwakayama Jul 31, 2026
12eebef
test(eval): cover framework tool search with existing mocks
kwakayama Jul 31, 2026
b490574
test(eval): retain durable canary run identities
kwakayama Jul 31, 2026
86a3004
fix(eval): retain created durable canary run ids
kwakayama Jul 31, 2026
7382046
docs(agent): document framework tool search durability
kwakayama Jul 31, 2026
64a3f62
test(agent): add direct framework tool search verifier
kwakayama Jul 31, 2026
ca7d41b
fix(agent): harden direct tool search verifier
kwakayama Jul 31, 2026
dec004d
fix(eval): identify durable canaries as studio client
kwakayama Jul 31, 2026
9cf7542
docs(eval): refresh durable canary references
kwakayama Jul 31, 2026
c3c94ec
fix(eval): forward durable canary model
kwakayama Jul 31, 2026
d80b0d0
docs(eval): refresh durable model references
kwakayama Jul 31, 2026
d36a214
fix(agent): honor direct providers in hosted execution
kwakayama Jul 31, 2026
1cee96a
fix(agent): derive tool discovery from tools selector
kwakayama Jul 31, 2026
a606d8e
fix(agent): explain deferred tool discovery
kwakayama Jul 31, 2026
710048d
fix(agent): make deferred discovery resilient
kwakayama Jul 31, 2026
a21037b
Merge remote-tracking branch 'origin/main' into codex/framework-tool-…
kwakayama Jul 31, 2026
84d91ee
fix(agent): harden deferred tool discovery
kwakayama Jul 31, 2026
41a9f5a
fix(agent): close hosted tool exposure gaps
kwakayama Jul 31, 2026
15749e2
fix(agent): filter hosted skill tools for all selectors
kwakayama Jul 31, 2026
86725a5
fix(agent): render hosted skills after tool authorization
kwakayama Jul 31, 2026
02aa747
chore(release): bump veryfront to 0.1.1179
kwakayama Jul 31, 2026
8d7eec2
test(agent): align direct skill metadata with tool visibility
kwakayama Jul 31, 2026
d65863e
docs(agent): refresh hosted tool policy reference
kwakayama Jul 31, 2026
417703d
test(agent): tighten deferred tool discovery proof
kwakayama Jul 31, 2026
File filter

Filter by extension

Filter by extension


Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 3 additions & 2 deletions deno.json
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
{
"name": "veryfront",
"version": "0.1.1178",
"version": "0.1.1179",
"license": "Apache-2.0",
"nodeModulesDir": "auto",
"minimumDependencyAge": {
Expand Down Expand Up @@ -473,7 +473,8 @@
"lint:ban-test-only": "deno run --allow-read scripts/lint/ban-test-only.ts",
"lint:sanitizer-baseline": "deno run --allow-read scripts/lint/check-sanitizer-baseline.ts",
"lint:skipped-tests": "deno run --allow-read scripts/lint/check-skipped-tests-baseline.ts",
"test:scripts": "deno test --config=scripts/test.deno.json --no-check --allow-read --allow-write --allow-run scripts/ci/publish-npm-packages.test.ts scripts/build/dnt-polyfill.test.ts scripts/build/generate-sbom.test.ts scripts/build/npm-dependency-sources.test.ts scripts/build/npm-extension-package-metadata.test.ts scripts/build/npm-package-metadata.test.ts scripts/build/npm-react-shims.test.ts scripts/docs/docs-coverage.test.ts scripts/docs/generate-api-reference.test.ts scripts/docs/guide-validation.test.ts scripts/lint/audit-core-deps.test.ts scripts/lint/audit-dependency-boundaries.test.ts scripts/lint/audit-extension-capabilities.test.ts scripts/lint/audit-extension-contracts.test.ts scripts/lint/audit-deps.test.ts scripts/lint/check-module-boundaries.test.ts scripts/lint/ban-test-only.test.ts scripts/lint/check-sanitizer-baseline.test.ts scripts/lint/check-skipped-tests-baseline.test.ts scripts/lint/check-coverage.test.ts scripts/security/audit-npm.test.ts scripts/security/submit-dependency-snapshot.test.ts",
"test:scripts": "deno test --config=scripts/test.deno.json --no-check --allow-read --allow-write --allow-run scripts/ci/publish-npm-packages.test.ts scripts/build/dnt-polyfill.test.ts scripts/build/generate-sbom.test.ts scripts/build/npm-dependency-sources.test.ts scripts/build/npm-extension-package-metadata.test.ts scripts/build/npm-package-metadata.test.ts scripts/build/npm-react-shims.test.ts scripts/docs/docs-coverage.test.ts scripts/docs/generate-api-reference.test.ts scripts/docs/guide-validation.test.ts scripts/lint/audit-core-deps.test.ts scripts/lint/audit-dependency-boundaries.test.ts scripts/lint/audit-extension-capabilities.test.ts scripts/lint/audit-extension-contracts.test.ts scripts/lint/audit-deps.test.ts scripts/lint/check-module-boundaries.test.ts scripts/lint/ban-test-only.test.ts scripts/lint/check-sanitizer-baseline.test.ts scripts/lint/check-skipped-tests-baseline.test.ts scripts/lint/check-coverage.test.ts scripts/security/audit-npm.test.ts scripts/security/submit-dependency-snapshot.test.ts && deno task test:tool-search-live",
"test:tool-search-live": "VF_DISABLE_LRU_INTERVAL=1 deno test --no-check -A tests/agent/verify-tool-search-live.test.ts",
"test:cross-runtime": "deno run --allow-all src/platform/compat/cross-runtime.test.ts",
"test:node": "node ./tests/node/run-tests.mjs 'src/**/*.test.ts'",
"test:bun": "node ./tests/bun/run-tests.mjs src/",
Expand Down
481 changes: 242 additions & 239 deletions docs/api-reference/veryfront/agent.md

Large diffs are not rendered by default.

33 changes: 17 additions & 16 deletions docs/api-reference/veryfront/eval.md
Original file line number Diff line number Diff line change
Expand Up @@ -203,7 +203,7 @@ import { assertCompleted, assertNoMalformedCreateFileToolCalls, buildAgentServic
| Name | Description | Source |
|------|-------------|--------|
| `DEFAULT_AGENT_SERVICE_EVAL_ENDPOINT` | Default local AG-UI endpoint used by agent-service evals. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service.ts#L22) |
| `DEFAULT_DURABLE_RUN_CANARY_TIMEOUT_MS` | Default value for durable run canary timeout ms. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/environment.ts#L16) |
| `DEFAULT_DURABLE_RUN_CANARY_TIMEOUT_MS` | Default value for durable run canary timeout ms. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/environment.ts#L17) |
| `DEFAULT_LIVE_EVAL_AREA_TAG_RULES` | Default value for live eval area tag rules. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/live-evals/metadata.ts#L37) |
| `DEFAULT_LIVE_EVAL_ENDPOINT` | Default value for live eval endpoint. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/live-evals/environment.ts#L17) |
| `DEFAULT_LIVE_EVAL_OPTIONAL_JUDGE_CASE_PREFIXES` | Default value for live eval optional judge case prefixes. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/live-evals/metadata.ts#L30) |
Expand All @@ -230,8 +230,8 @@ import { assertCompleted, assertNoMalformedCreateFileToolCalls, buildAgentServic
| `containsSkillLoad` | Contains skill load helper. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/live-evals/runner.ts#L484) |
| `countStepStartedEvents` | Count step started events helper. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/live-evals/runner.ts#L489) |
| `createAgentServiceEvalAdapter` | Create an `EvalAgentAdapter` that executes examples against an AG-UI agent-service endpoint. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service.ts#L553) |
| `createDurableRunCanaryApiClient` | Create durable run canary API client. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L260) |
| `createDurableRunCanaryRunner` | Create durable run canary runner. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L539) |
| `createDurableRunCanaryApiClient` | Create durable run canary API client. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L271) |
| `createDurableRunCanaryRunner` | Create durable run canary runner. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L554) |
| `createDurableRunTokenGrowthCanaryCase` | Create a two-turn durable run canary for historical tool-input token growth. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/token-growth.ts#L39) |
| `createFailedEvalResult` | Result returned from create failed eval. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/live-evals/result.ts#L80) |
| `createLiveEvalApiClient` | Create live eval API client. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/live-evals/api-client.ts#L272) |
Expand All @@ -251,10 +251,10 @@ import { assertCompleted, assertNoMalformedCreateFileToolCalls, buildAgentServic
| `hasEveryLiveEvalTag` | Check whether every live eval tag is present. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/live-evals/report.ts#L28) |
| `hasFinished` | Check whether finished is present. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/live-evals/runner.ts#L479) |
| `listOpenLiveEvalInputRequests` | List open live eval input requests. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/live-evals/api-client.ts#L504) |
| `parseDurableRunCanaryRunSummary` | Parses durable run canary run summary. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L114) |
| `parseDurableRunCanaryRunSummary` | Parses durable run canary run summary. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L115) |
| `printRuntimeConfidencePreflight` | Print runtime confidence preflight helper. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/live-evals/preflight.ts#L40) |
| `resolveAgentServiceEvalEnvironment` | Resolve environment values for live agent-service eval execution. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service.ts#L477) |
| `resolveDurableRunCanaryEnvironment` | Resolves durable run canary environment. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/environment.ts#L19) |
| `resolveDurableRunCanaryEnvironment` | Resolves durable run canary environment. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/environment.ts#L20) |
| `resolveLiveEvalEnvironment` | Resolves live eval environment. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/live-evals/environment.ts#L20) |
| `resolveLiveEvalRequestedCaseIds` | Resolves live eval requested case IDs. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/live-evals/report.ts#L89) |
| `runDurableRunCanaryCli` | Run durable run canary cli. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/cli-runner.ts#L47) |
Expand All @@ -278,19 +278,20 @@ import { assertCompleted, assertNoMalformedCreateFileToolCalls, buildAgentServic
| `BuildAgentServiceEvalRequestBodyInput` | Input accepted by `buildAgentServiceEvalRequestBody`. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service.ts#L59) |
| `BuildLiveEvalCaseMetadataInput` | Input payload for build live eval case metadata. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/live-evals/metadata.ts#L23) |
| `BuildLiveEvalRequestBodyInput` | Input payload for build live eval request body. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/live-evals/request.ts#L19) |
| `DurableRunCanaryApiClient` | Public API contract for durable run canary API client. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L249) |
| `DurableRunCanaryApiClient` | Public API contract for durable run canary API client. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L260) |
| `DurableRunCanaryApiConfig` | Configuration used by durable run canary API. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L10) |
| `DurableRunCanaryCase` | Public API contract for durable run canary case. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L373) |
| `DurableRunCanaryCase` | Public API contract for durable run canary case. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L392) |
| `DurableRunCanaryCliCaseFactoryInput` | Input payload for durable run canary cli case factory. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/cli-runner.ts#L17) |
| `DurableRunCanaryCreateRootRunInput` | Input payload for durable run canary create root run. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L21) |
| `DurableRunCanaryCreateRootRunInput` | Input payload for durable run canary create root run. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L22) |
| `DurableRunCanaryEnvironment` | Public API contract for durable run canary environment. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/environment.ts#L7) |
| `DurableRunCanaryMessage` | Message shape for durable run canary. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L50) |
| `DurableRunCanaryPreparedCase` | Public API contract for durable run canary prepared case. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L358) |
| `DurableRunCanaryResult` | Result returned from durable run canary. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L346) |
| `DurableRunCanaryRunnerConfig` | Configuration used by durable run canary runner. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L380) |
| `DurableRunCanaryRunSummary` | Public API contract for durable run canary run summary. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L55) |
| `DurableRunCanarySendUserMessageInput` | Input payload for durable run canary send user message. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L27) |
| `DurableRunCanaryStartRunInput` | Input payload for durable run canary start run. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L33) |
| `DurableRunCanaryExecution` | Execution metadata retained for each durable run canary prompt. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L357) |
| `DurableRunCanaryMessage` | Message shape for durable run canary. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L51) |
| `DurableRunCanaryPreparedCase` | Public API contract for durable run canary prepared case. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L376) |
| `DurableRunCanaryResult` | Result returned from durable run canary. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L363) |
| `DurableRunCanaryRunnerConfig` | Configuration used by durable run canary runner. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L399) |
| `DurableRunCanaryRunSummary` | Public API contract for durable run canary run summary. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L56) |
| `DurableRunCanarySendUserMessageInput` | Input payload for durable run canary send user message. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L28) |
| `DurableRunCanaryStartRunInput` | Input payload for durable run canary start run. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L34) |
| `DurableRunTokenGrowthCanaryCaseInput` | Input payload for create durable run token-growth canary case. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/token-growth.ts#L14) |
| `LiveEvalApiClient` | Public API contract for live eval API client. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/live-evals/api-client.ts#L123) |
| `LiveEvalApiContext` | Context for live eval API. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/live-evals/api-client.ts#L11) |
Expand Down Expand Up @@ -333,4 +334,4 @@ import { assertCompleted, assertNoMalformedCreateFileToolCalls, buildAgentServic

| Name | Description | Source |
|------|-------------|--------|
| `getDurableRunCanaryMessageSchema` | Zod schema for get durable run canary message. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L40) |
| `getDurableRunCanaryMessageSchema` | Zod schema for get durable run canary message. | [source](https://github.com/veryfront/veryfront-code/blob/main/src/eval/agent-service/durable-run-canaries/runner.ts#L41) |
29 changes: 29 additions & 0 deletions docs/evidence/deferred-tool-discovery-hi-anthropic.json
Original file line number Diff line number Diff line change
@@ -0,0 +1,29 @@
{
"schemaVersion": 1,
"kind": "deferred-tool-discovery-hi",
"prompt": "hi",
"agentId": "veryfront-agent",
"agentConfiguration": "prd-all-scoped-example",
"catalogProfile": "deterministic-64-tool-verifier-fixture",
"provider": "anthropic",
"model": "anthropic/claude-opus-4-6",
"measuredAt": "2026-07-31T21:29:04.642Z",
"provenance": "direct-provider-framework-fallback",
"usageSource": "paired-response.usage.promptTokens",
"baselineInputTokens": 5276,
"effectiveInputTokens": 648,
"reductionPercent": 87.718,
"baselineModelSteps": 1,
"baselineExposedToolCount": 64,
"modelSteps": 1,
"toolCalls": 0,
"authorizedToolCount": 64,
"initiallyExposedTools": [
"tool_search"
],
"thresholds": {
"maximumEffectiveInputTokens": 10000,
"minimumReductionPercent": 60
},
"passed": true
}
Loading