Repository navigation
refactor!: stop publishing 34 exports no consumer uses - #949
Merged
Merged
Conversation
Refresh the public-API census sweep (36 default-branch checkouts of every tangle-network repository whose package.json names this package, up from 31 on 2026-08-21) and act on its delete list. These symbols have no evidence in any channel: no consumer import, type bind, or mention, no in-repo production caller, test, script, example, or Markdown front door, and an organisation-wide GitHub code search finds them only here. Removed from the published entries; definitions stay where other modules of this package use them, and are deleted where nothing does (computeExperimentStats, roundTripRunRecord, runCostFloorUsd, isPolicyEdit, reportSupervisorRound, surfaceDigestEdits). Kept by the census rules: typed error constructors, wire schema and contract-version constants, runProposeReviewAsControlLoop, shuffleOrder, and every symbol another repository mentions (improvementVerdict, writeSupervisorRunReport, verifyLoopProvenanceRecord, ...). Files inside the analyst benchmark digest are untouched, so publicBenchmarkDistributions and traceAnalystFunctionGroup keep their definitions. BREAKING CHANGE: the 36 symbols listed in the pull request are no longer exported from any package entry point.
|
You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard. |
tangletools
previously approved these changes
Oct 6, 2026
tangletools
left a comment
Contributor
There was a problem hiding this comment.
✅ Auto-approved PR — b30f1d03
Blanket team auto-approval is intentional. This is not a code review.
No automated review runs on this PR. This approval rests on the rule above alone.
tangletools · auto-approval · reason: blanket_auto_approve · 2026-10-06T06:02:46Z
…e-cell It was also published by the profile-cell entry, which defines it; only agentProfileCellHash in the same file calls it. BREAKING CHANGE: agentProfileCellHashMaterial is no longer exported.
tangletools
approved these changes
Oct 6, 2026
tangletools
left a comment
Contributor
There was a problem hiding this comment.
✅ Auto-approved PR — 57fad7f8
Blanket team auto-approval is intentional. This is not a code review.
No automated review runs on this PR. This approval rests on the rule above alone.
tangletools · auto-approval · reason: blanket_auto_approve · 2026-10-06T06:04:46Z
This was referenced Oct 6, 2026
drewstone
added a commit
that referenced
this pull request
Oct 6, 2026
…951) test(gate): give the whole-repository collation scan a 60 s timeout The agent-eval v0.209.0 Publish failed in `verify` ([run 37425986741](https://github.com/tangle-network/agent-eval/actions/runs/37425986741)): `scripts/check-collation-ordering.test.mjs > the shipped repository > passes its own gate` took 6,783 ms on a loaded self-hosted runner and hit Vitest's 5,000 ms default. The other 6,142 tests passed. npm still serves 0.208.2, so #949's breaking change is not published.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Problem
docs/public-api.md(the censuspnpm api:censuswrites) lists published symbols with no consumer. Its consumer evidence was last swept on 2026-08-21 across 31 repositories. Every published symbol is surface the package must keep working.Change
tangle-networkrepositories whosepackage.jsonnames@tangle-network/agent-eval(adds chatgpt-plugins, discovery, factory, gtm, hospitality-agent, ph0ny, super-agent, tangle-tools, trustgate;loops,phony,skeletal-osno longer exist).scripts/public-api-consumers.jsonanddocs/public-api.mdare regenerated.docs/public-api.md: typed errors (HostedRequestError,SearchHistoryRequiredError), schema and version constants (*_VERSION,*_SCHEMA*,RUBRIC_VERSION_SCHEME,OTEL_AGENT_EVAL_SCOPE,TRACE_ANALYST_*,multishotGoldenVersions), the two documented exceptions (runProposeReviewAsControlLoop,shuffleOrder), and every symbol another repository mentions (improvementVerdict,writeSupervisorRunReport,writeSupervisorRunReportSafe,verifyLoopProvenanceRecord,verificationReportToRunRecord,renderPriorFindings,assertCodeSurfaceIdentity,codeSurfaceIdentityMaterial,judgeRealnessLlm,RUN_METRICS).publicBenchmarkDistributionsstays because its re-export sits inside the analyst benchmark digest.agentProfileCellHashMaterial,claimBrief,claimIntegritySystemOneQuestions,computeExperimentStats,DEFAULT_MAX_MODEL_TRACES,DEFAULT_PERMUTATIONS,EDIT_CREDIT_ESTIMATOR,EDIT_CREDIT_SIGNAL,EVIDENCE_AUTHORITY_KINDS,INDEPENDENT_EVIDENCE_AUTHORITY_KINDS,isPolicyEdit,isRunMetric,MANN_WHITNEY_EXACT_MAX_STATES,MANN_WHITNEY_EXACT_MAX_WORK,MAX_FIRST_FAILURE_IDS,META_SEARCH_SIGNAL,nearestNeighbours,NO_OP_ACTIONS,otlpRowsToTraceRunRecords,POLICY_EDIT_AXES,policyEditFromFinding,POLICY_EDIT_TARGET_SURFACES,profileTextLines,readHarborTask,readLabelEntries,REFEREE_VERDICTS,renderBatchReport,reportSupervisorRound,roundTripRunRecord,runCostFloorUsd,scorePolicyEditReadiness,surfaceDigestEdits,traceAnalystFunctionGroup,WILCOXON_EXACT_MAX_N(isRunMetric,REFEREE_VERDICTS,claimBrief,claimIntegritySystemOneQuestionswere published throughexport *andagentProfileCellHashMaterialthrough the./profile-cellentry that defines it, so they lose theirexportkeyword).Where nothing else in the package uses a definition, it is deleted:
computeExperimentStats(with its privatemedian/stddev),roundTripRunRecord,runCostFloorUsd,isPolicyEdit,reportSupervisorRound,surfaceDigestEdits.traceAnalystFunctionGroupkeeps its definition becausesrc/trace-analyst/tools.tsis inside the benchmark digest.Breaking change
Yes, for the 34 names above (commit carries
BREAKING CHANGE:so the next release notes it). Census after the change: 1,393 distinct published symbols (was 1,427 at main with the refreshed sweep),none212 (was 247).Blind spots
Consumers outside the organisation and private forks are invisible to every channel (census blind spot 2). Published
dist/of downstream npm packages was not separately unpacked; those packages pin older agent-eval versions and their default-branch source is in the sweep.Cost
Rollback: revert, or re-export a name if a consumer surfaces.
Measurement (measure.sh, base 06a4d03)
Checks (gtr, node 24.20.0)
pnpm install --frozen-lockfile; husky pre-commit (lint-staged +pnpm typecheck): pass.biome check src: pass.tsdown+pnpm openapi: pass.node scripts/verify-package-exports.mjs(packs and imports every entry): pass.3a994853…).pnpm test: 419 files, 6143 passed, 3 skipped.