mirror of
https://github.com/garrytan/gstack.git
synced 2026-09-26 22:51:47 +02:00
v1.87.6.0 fix: make checks reliable and everyday validation faster (#2898)
* fix: acknowledge seeded plans before invoking review skills * fix: distinguish current plan input from conversation history * fix: keep hermetic plan reviews on manual permissions * fix: distinguish tool discovery from file permission ownership * fix: preserve initial plan mode in observation tests * fix: wait for scope decisions before writing review findings * fix: carry autoplan decisions consistently into review artifacts * test: retain native failure context in periodic assertions * fix: advance active file permissions before queued questions * fix: finish red-team attempts before retry and cleanup * fix: finalize plan format captures and judges before retry * fix: cancel setup-gbrain SDK attempts before fixture cleanup * test: select periodic consumers of the bounded attempt helper * fix native Bash permission cards and queued questions * fix: preserve independent decisions and review scope Keep CEO approach, engineering scope and outside-review choices from approving independent remedies together. Carry declared contracts through DX polish and resolve new gaps before editing the plan. Regenerate every host and retain existing stop boundaries. Validation: 654 focused tests passed across nine files; all-host generation passed. Full free and periodic validation pending. Co-Authored-By: OpenAI Codex <noreply@openai.com> * fix: require approval before design plan amendments Align the Design review philosophy and rating recipe with its section protocol: resolve one proposed fix, then apply only that approved decision and retain honest scores for declined fixes. Validation: 469 focused tests passed across four files; all-host generation passed. Co-Authored-By: OpenAI Codex <noreply@openai.com> * fix: observe native question completion before transcript persistence Match owned completion hooks to submitted choices, reject conflicting or late answers, and retain bounded failure evidence. * test: recognize review posture in acknowledged native questions Require the selected mode acknowledgement, a completed follow-up question, and its current decoded display while preserving existing posture assertions. * fix: preserve settled CEO choices and isolate pending remedies Resolve established approach gates with cited authority and keep independent fixes out of unrelated option commitments and plan amendments. * fix: carry approved DX choices through later review steps Choose documentation approaches within the accepted scope and map resolved confusion points without reopening them through a bulk menu. * test: handle native settings-file edit prompts Keep one-time owned-file approvals and retain the actual sampled Autoplan permission frame with its matching barrier state. * test: accept standard CEO reply directives with tuning footers Recognize the exact trailing preference footer and letter-list directive while preserving current-display and exact acknowledgement checks. * test: scope split reviewers to their generated plan artifacts * test: observe native Bash permissions and invocation results * test: handle owned Bash prompts during mode preference checks * test: preserve synchronous subprocess rejection in Codex fixture * Fix periodic review handoff navigation Recognize review-first and explicit manual-next-step labels while preserving exact action families, manual preference, and ambiguous-menu rejection. Co-authored-by: OpenAI Codex <noreply@openai.com> * Bind pending file permissions to distinct current targets Allow one captured file request to own the complete current dialog while unrelated file work is pending. Preserve same-path ambiguity, exact input ownership, and one-time grant checks. Co-authored-by: OpenAI Codex <noreply@openai.com> * Make paired CEO verification choices genuinely unresolved Start the positive control with proposed manual checks so its unchanged oracle measures two new coverage decisions. Preserve runtime contracts, targets, count bounds, and all assertions. Co-authored-by: OpenAI Codex <noreply@openai.com> * Keep CEO review options and verification within approved scope Audit every offered option for independent add-ons and keep new verification depth pending until accepted. Preserve already requested coverage and trace plan changes to the actual decision. Co-authored-by: OpenAI Codex <noreply@openai.com> * Assemble DX review artifacts before appending the final report Keep early DX evidence above decisions, update artifact sections in place, and append the report using the actual current file suffix. Re-read after deleting an existing report before choosing the append anchor. Co-authored-by: OpenAI Codex <noreply@openai.com> * Keep outside plan reviews exclusive and invocation-owned Follow one preflight-selected backend, terminate failed Codex work before fallback, and allocate extra prompt/output files uniquely. Consume only the current invocation’s completed output. Co-authored-by: OpenAI Codex <noreply@openai.com> * Select periodic completion evaluations for report writer changes Register the shared review resolver for eight missing consumers and regress selection for all nine completion cases without changing their IDs or tiers. Co-authored-by: OpenAI Codex <noreply@openai.com> * Keep permission ambiguity fixtures on the same normalized target Use distinct raw spellings of one target in the four negative fixtures so they exercise the normalized duplicate-owner guard after exact current-file disambiguation. Preserve the existing exception, no-input, diagnostic and cleanup assertions. Co-authored-by: OpenAI Codex <noreply@openai.com> * Clarify preserved contracts in engineering review fixture Co-authored-by: OpenAI Codex <noreply@openai.com> * Recognize the offered DX follow-up handoff Co-authored-by: OpenAI Codex <noreply@openai.com> * Check independent commitments before presenting review options Co-authored-by: OpenAI Codex <noreply@openai.com> * Keep Codex review output and status in one shell invocation Co-authored-by: OpenAI Codex <noreply@openai.com> * Distinguish seeded plans from reports written by a test attempt Co-authored-by: OpenAI Codex <noreply@openai.com> * Recover clipped Autoplan file approvals with bounded viewport resizing Co-authored-by: OpenAI Codex <noreply@openai.com> * Recover clipped Bash approvals before binding the complete command Co-authored-by: OpenAI Codex <noreply@openai.com> * Isolate setup message tests from the shared checkout Run the real installer in a temporary payload with private config, require successful completion, and guard source and binary contents and mtimes. Co-authored-by: OpenAI Codex <noreply@openai.com> * Fix periodic native permission and report completion handling Match the pinned CLI's soft wraps and clipped headings without granting from incomplete frames. Retire completed file requests, retain mode annotations, and ask section captures for a short final acknowledgement after their full report is saved. Co-authored-by: OpenAI Codex <noreply@openai.com> * Preserve review approvals and validate DX comparison artifacts Keep independent remedies and approved amendments explicit. Give the synthetic DX review its existing documentation and validate peer comparison as required analysis alongside four native decisions. Add positive and negative semantic calibrations while preserving review counts, model budgets and prompt size limits. Co-authored-by: OpenAI Codex <noreply@openai.com> * Make the five-finding CEO fixture's application boundary explicit Materialize the request adapter and service composition used by the synthetic payment application. Explicitly declare the revised unregistered-event and mail-telemetry assumptions while preserving uncaught handler errors, the original invoice path and all five unresolved findings. Co-authored-by: OpenAI Codex <noreply@openai.com> * Keep CEO state-path checks scoped to directory preparation Co-authored-by: OpenAI Codex <noreply@openai.com> * Use checked ports and bounded cleanup in pair-agent tests Discover the daemon port from its owned state file, retain startup diagnostics, and await failed-start cleanup. Add occupied-port, early-exit, deadline, and foreign-state regressions while preserving the existing HTTP assertions and hook budgets. Co-authored-by: Codex <noreply@openai.com> * Preserve queued edit identity and recover clipped Bash permissions Distinguish separately queued unfinished edits from mutation of one native tool ID. Keep grants bound to an exact owned request and reject reused IDs, ambiguous inputs, and competing owners. Support the pinned renderer's literal em dash and request a repaint when only the Bash card's top rule is clipped. Grants still require the complete fresh card and an exact native acknowledgment. Validation: 413 integrated parser/event tests passed; private repaint controls and joint source review passed. Full canonical suite and native periodic rerun remain pending. Co-authored-by: Codex <noreply@openai.com> * Keep periodic reviews within their approved contracts and deliverables Carry exact approvals through engineering review, preserve declared contracts when amending CEO plans, and keep prioritization at the requested decision level. Materialize the revised synthetic SDK reference contract while retaining the five original documentation gaps. Accept the observed semicolon in the finite DX handoff menu and register the direct source dependencies used by the engineering cases. Regenerate canonical review documents without changing model budgets, retries, count bands, or native completion assertions. Validation: all-host generation and 275 review, fixture, selection and parity tests passed. Full free-suite and native periodic validation remain pending. Co-authored-by: Codex <noreply@openai.com> * Keep Eng approval cadence and independence guards explicit * Accept ordinary punctuation in manual review handoffs * Recover file permissions alongside queued Bash calls * Carry approved DX work through later review findings * Clarify the synthetic auth internal failure decision * Bound the periodic DX fixture to onboarding changes * Recognize native Design review handoff labels * Hold scope in the integration-choice review fixture * Carry approved Design decisions through review evidence * Capture listener state when feedback reload fails * Exclude workspace caches before checking deprecated flags * Verify Design UI scope against a seeded review plan * Clarify plan review decisions and outside-voice approval flow * Reject setup menus in the Design UI gate * docs: require focused repair validation before final acceptance * fix: separate review commitments within existing prompt budgets * docs: align generation and contributor validation guidance * fix: advance native review prompts and count acknowledged findings * chore: bump version and changelog (v1.87.1.0) Co-Authored-By: OpenAI Codex <noreply@openai.com> * chore: enforce cheap checks and side-effect-free validation previews * fix: handle owned Fetch permissions and oversized native cards * test: ground review fixtures in independent executable contracts * fix: preserve review decisions and verify reports before completion * test: construct the synthetic credential URL without a scanner false positive * test: materialize DX examples and verify their actual local behavior * fix: clarify CEO review decisions and execution order * fix: clarify review workflow ordering and select Design quality checks * Fix review decision gates and incomplete evaluation fixtures Persist CEO and engineering commitment ledgers before menus, preserve exact approvals, and distinguish implementation structure from feature scope. Route Autoplan through the canonical CEO Step 0 ordering. Classify DX findings before requesting approval and ground runtime claims in actual evidence. Complete neutral non-target fixture contracts and accept the captured Design handoff purpose without relaxing its ownership or acknowledgment checks. Record runtime-capability verification in AGENTS.md validation discipline. Validation: 1,335 focused tests passed across 21 files; build, all-host freshness, skill validation (647 artifacts / 107 tracked), and credential checks passed. Prior paid failures are preserved; behavioral acceptance remains pending. * Fix review decision boundaries and owned Read prompts Preserve exact approvals across review options, compare consistent DX milestones, and keep proposed implementation separate from review evidence. Bind modern Read prompts to one immutable native request and wait for its result. Retain captured regression verdicts, correct fixture error names, improve import probe diagnostics, and record focused-first validation discipline in AGENTS.md. * Clarify CEO and engineering review decisions Use explicit decision steps, one engineering ledger, and clear scope/write transitions. Preserve exact approvals and distinguish pending test requirements. Keep unrelated generated content unchanged. * Fix review decision ordering and native evaluation interactions * Clarify engineering decisions and test artifact order * Clarify pending choices and approvals in CEO reviews * Make CEO review phases sequential and clarify completion * Fix Design board submission intent matching * Seed an existing browser test baseline for Autoplan * Document decision-log payloads before state initialization * Preserve exact review scope and decide one change before drafting options * Require input identity before repeating passing model judges * Honor permitted storage throughout CEO review completion * Match complete native permission text within the pinned renderer contract * Align review approvals, independent choices, and bounded validation * fix: preserve reopened approvals and declare fixture interfaces * fix: isolate review artifacts and audit complete questions * fix: match detector artifact permissions to configured storage * fix: complete native permissions and review fixture workflows * fix: order CEO review work and separate engineering guarantees * fix: preserve native validation and separate review choices * fix: clarify review decisions and judge complete report context * fix: constrain review judgments and retain parse failures * fix: compare each affected value before review decisions * fix: make engineering review decisions and completion order explicit * fix: give the complete Autoplan evaluation a bounded chain budget * fix(cso): diagnose forbidden Docker endpoints before tool lookup * fix(reviews): reconcile workflow contracts and generated artifacts after main integration * fix(evals): migrate retained regressions to the native review harness * fix(tests): close native harness and workflow integration regressions * fix(evals): preserve complete permission context and native menu contracts * fix(tests): capture synchronous command output without pipe drain stalls * fix(reviews): clarify decision and completion ordering * fix(reviews): separate decision readiness from final completion checks * refactor(reviews): consolidate decision rules and completion branches * fix(plan-eng-review): order preparation and clarify decision routing * fix(plan-eng-review): restore size and question-format guard parity * fix(plan-eng-review): clarify scope phases and blocked completion * fix(plan-eng-review): unify review flow and report destination * fix(plan-eng-review): define bootstrap and question stage ownership * fix(plan-eng-review): clarify review structure and design lookup * fix(plan-eng-review): render report examples and show saved decisions * fix: consolidate Eng review decisions and select their evaluations * test: cover overlapping terminal attachments and clean merged runner type * fix: preserve Office Hours relationship closings during review updates * fix: retain pasted review targets across slash invocations * docs: preserve validation traces and correct release scope * test: cover pasted targets in both review skills * fix: validate report artifacts before recording success * fix: redact source roots at CSO report boundaries * fix: bind native Design questions before answering * test: select report privacy and native recovery regressions * test: bind rejection predicate in extracted observers * fix: bind complete boxed native questions * test: keep the Design UI fixture on native review * fix: preserve review decisions and evaluation completion outcomes * fix: clarify CEO approval and report completion order * fix: align native review evaluation ownership and completion * fix: bind review evaluators to native decisions and owned artifacts * fix: validate review decisions against native outcomes * fix: preserve review evidence and Autoplan phase handoffs * test: bind review evidence to owned decisions and completion * fix: retain owned native history across compaction * fix(evals): validate current review decisions and setup choices * fix: bind Autoplan reviews and phase completion to current amended input * fix: reconcile native review evidence and close Autoplan phases * test: recognize owned whole-candidate complexity decisions * test: preserve report freshness for approved investigation handoffs * fix: recognize scoped review findings and isolate dual voice fixtures * fix: make review handoffs and question dispatch self-contained * test: recognize complete CEO decisions and procedural pauses * fix: bind current CEO comparison options and risk intervals * test: bind engineering decisions and completion to owned evidence * fix: publish Autoplan phase reports before continuing tools * test: verify actual Autoplan dual-review dispatch evidence * test: select dual review when shared evidence fixtures change * fix: clarify plan review decisions and completion gates * fix: make CEO review decisions and return paths explicit * test: keep Autoplan prompt files inside attempt state * test: preserve source whitespace across permission dialog wraps * fix: publish Autoplan phase reports before continuing * test: recognize current CEO comparisons and reject inactive records * fix: reconcile engineering decision states before completion * test: recognize complete Design decisions and reports * test: verify current engineering decisions before navigation * Recognize source-owned component reduction choices * fix: recognize current CEO ledger and commitment grids * test: supply RequestPolicy context to Eng count fixture * fix: save complete engineering decisions before asking * fix: bind Autoplan publication to the complete phase readback * chore: prepare 1.87.5.0 reliability release * fix: clarify engineering review completion and preserve log failures * fix: bind CEO saved choices and current section ancestry * fix(evals): bind review execution and completion evidence * fix(plan-ceo-review): verify complete decisions before asking * fix(evals): preserve complete engineering choice records * fix(evals): preserve complete review outcomes and bounded fixtures * fix(autoplan): publish phase reports before advancing * fix(plan-ceo-review): validate option fields before asking * fix(plan-eng-review): verify current decisions after answers * fix(evals): bind review decisions and bound fixture scope * fix(plan-ceo-review): verify decision rows and edit saved checkpoints * fix(evals): bind review evidence and scope document lookup * fix(plan-eng-review): update resolution state with its answer * fix(reviews): preserve complete questions through dispatch * fix(evals): recognize completed mode declarations * fix(evals): define cache consistency at wrapper completion * fix(evals): validate owned initial scope and completed review handoffs * fix: assemble complete CEO decision fields before saving * fix: authenticate automatic mode decisions without guessing selectors * fix: bind engineering coverage to approved regression contracts * fix(evals): supply review helpers to native Eng capture * fix(plan-eng-review): preserve the full selected option scope * fix(evals): recognize owned engineering seed and regression evidence * fix(evals): bind engineering retry reports to native approvals * docs: clarify release guarantees (v1.87.5.0) Co-Authored-By: OpenAI Codex <noreply@openai.com> * fix(evals): recognize owned engineering decisions and handoffs * fix(evals): bind engineering decisions and completion evidence * fix(tests): align review contracts and selection fixtures * fix(skills): restore review prompt size limits * fix(plan-eng-review): clarify review execution and completion * fix(evals): preserve configured retries through all supervision layers * Clarify Engineering decisions and report completion * Keep native decision assertions within their source boundary * fix: recognize owned engineering decisions and completed navigation * fix: bind completed auto decisions to their current review * fix: recognize explicit CEO source attribution * fix: dispatch verified CEO decisions without recomposing fields * test: expose existing execution deadlines to review actors * fix: distinguish CEO decision records from incidental headings * test: bind split-scope choices to the registered native actor * test: connect reviewed regressions to required evaluation coverage * Clarify CEO decision routing and completion stages * test: expose existing section review deadlines to fixture actors * test: recognize complete native CEO pacing inventories * test: exclude answered history from current CEO payloads * test: detect phase entry through owned skill HOME aliases * test: validate native review completion and owned report permissions * fix: make Autoplan close packets carry the parent handoff steps * test: assess source-bound HOLD decisions within the existing deadline * fix: keep CEO native decision fields under one formatting authority * test: register integrated review and permission dependencies * test: align native review adapters and finding coverage Preserve explicit AUTO decisions, apply native single-select defaults, and bind complete cropped questions and report permissions to their owned requests. Require seeded review findings instead of crediting setup menus. Keep captured failure controls and additive selection dependencies. The integrated candidate passed 3,099 focused tests across 65 files; affected paid validation remains required before publication. * fix(autoplan): require phase reports before advancing * fix(evals): bind setup and evidence to complete attempts * fix(evals): bind native answers and pending writes to fixture scope Preserve complete option rows when native descriptions wrap, retain current owned Write arguments before journal publication, and keep engineering and DX answers within their declared fixture interfaces. Add captured free regressions without increasing model budgets or relaxing completion checks. * fix(autoplan): verify phase reports across native tool paths Guard owned methodology reads and reviewer dispatches, detect complete driver loads through Bash, and distinguish report-only edits from implementation changes. Follow authenticated native UUID ancestry when journal writes arrive out of order and verify earlier native content for cached phase reads. Keep current close acknowledgment and parent publication in order, require CEO entry before later phases, and register captured failure regressions. * fix(evals): honor native input and collection lifecycles Match complete native Edit panes and truncated question borders, reject stderr close before EOF, and stop the CEO split fixture once its acknowledged scope decisions are collected. Keep semantic validation, process failures, report requirements, and absolute deadlines authoritative. Add captured-event and real-process regressions with selection dependencies. Focused checks pass; final integrated paid and full-suite acceptance remain pending. * fix(autoplan): retain native session ownership across directory changes Recover missed native UUID ancestry through the existing strict graph while preserving ordinary event order and legacy scoping. Bind publication hooks to Claude's original project directory while retaining current cwd for requested file paths. Captured public-event regressions, existing caller checks, and a pinned native CLI loopback verify both fixes. Preserve failed attempts and require fresh paid and final full-suite acceptance. * docs: align evaluation limits and completion version * fix(autoplan): allow authenticated phase reads during journal streaming * fix(evals): bind clipped native questions and owned edit dialogs * fix: preserve overlay retries and bounded cleanup * fix: recognize owned planning preludes in native questions * docs: explain overlay scheduling and cleanup guarantees * fix: require fresh publication after Autoplan phase reruns * Release gstack 1.87.6 * fix: preserve CI paths, process identity, and test deadlines * fix: keep informational setup commands independent of install probes * fix: clarify plan review decisions and bound source audit reports * Fix remaining Windows identity and native path CI failures * Clarify CEO review decision and reviewer-result routing * test: accept no-install planner in retry supervision * fix(ceo-review): make review decisions and report completion explicit * perf(test): add fast PR gates, input-keyed judge reuse and isolated free shards * fix(test): start isolated CEO smoke from its existing project plan * fix(test): repair CI fixture races and preserve retry evidence * fix(ceo-review): clarify approvals, depth and saved completion --------- Co-authored-by: OpenAI Codex <noreply@openai.com>
This commit is contained in:
Garry Tan
co-authored by
OpenAI Codex
OpenAI Codex
parent
35dd014c58
commit
636175d349
+80
@@ -0,0 +1,80 @@
|
||||
# eval-sdk
|
||||
|
||||
Synthetic product documentation for this review fixture. The eval-sdk implementation
|
||||
is not included or installed here; the commands below describe its assumed existing
|
||||
interface. They are not claims that this fixture can execute an SDK evaluation.
|
||||
|
||||
Evaluate an application's outputs against caller-supplied cases. Python 3.10 or
|
||||
later; install the assumed package with `pip install eval-sdk`. The library is
|
||||
`eval_sdk`; the companion command is `eval-sdk`.
|
||||
|
||||
Confirm the installed package version with `python -m pip show eval-sdk` and
|
||||
the interpreter with `python --version`; neither command starts an evaluation.
|
||||
|
||||
During beta the published API and configuration contract remain compatible.
|
||||
Breaking changes need a versioned migration guide and deprecation notice for two
|
||||
minor releases before removal in a breaking release. See [upgrades](docs/reference-v1.md#upgrades).
|
||||
|
||||
## Quick start
|
||||
|
||||
This neutral example is mirrored in the getting-started guide. The existing
|
||||
product's offline release checks verify both copies and their output contract.
|
||||
The fixture does not run those product checks.
|
||||
|
||||
```python
|
||||
from eval_sdk import evaluate
|
||||
import json
|
||||
|
||||
def target(inputs):
|
||||
return {"ready": inputs["enabled"]}
|
||||
|
||||
def exact_match(actual, expected):
|
||||
# This application's structured-output rule.
|
||||
if not isinstance(actual, dict):
|
||||
return 0.0
|
||||
return float(actual == expected)
|
||||
|
||||
cases = [{"inputs": {"enabled": True}, "expected": {"ready": True}}]
|
||||
result = evaluate(target, cases, exact_match)
|
||||
print(json.dumps([{"score": case.score, "actual": case.actual, "expected": case.expected}
|
||||
for case in result.cases], sort_keys=True))
|
||||
```
|
||||
|
||||
Shown application output (JSON from the documented structured fields, not SDK repr):
|
||||
|
||||
```text
|
||||
[{"actual": {"ready": true}, "expected": {"ready": true}, "score": 1.0}]
|
||||
```
|
||||
|
||||
The application supplies the metric and decides what scores are acceptable;
|
||||
the SDK has no default quality bar. Fixture checks reproduce this text with an
|
||||
explicit assumed-contract double; they do not execute the absent SDK.
|
||||
|
||||
To use this same example in pytest, put it inside `test_ready()` in a `test_*.py`
|
||||
file and add the application's own assertion:
|
||||
|
||||
```python
|
||||
assert all(case.score == 1.0 for case in result.cases)
|
||||
```
|
||||
|
||||
This assertion is the application's exact-match acceptance rule, not an SDK
|
||||
default. See the [API and pytest reference](docs/reference-v1.md#api-and-pytest).
|
||||
The example remains one ordinary passing case; it has no staged regression.
|
||||
|
||||
Before substituting a real callable, read [deadlines and provider costs](docs/reference-v1.md#configuration).
|
||||
The SDK cannot cap spending by arbitrary application code; that code must use a
|
||||
bounded provider client or enforce its own limits. The [worked application client](docs/getting-started.md#bounded-application-calls)
|
||||
shows separate request timeouts, retry limits and cost reservations; the reference
|
||||
distinguishes these from SDK-managed configuration.
|
||||
|
||||
**Current first-run requirement:** both the library and CLI block the first eval
|
||||
for the mandatory five-minute compatibility/conformance check. There is no skip.
|
||||
The diagnostic report is not consumed by evaluation. No first-run duration has
|
||||
been measured, and no time-to-hello-world promise is made here.
|
||||
|
||||
[Getting started and free-text example](docs/getting-started.md).
|
||||
[Stuck while getting started?](docs/feedback.md).
|
||||
[CLI, configuration, errors, and upgrades](docs/reference-v1.md).
|
||||
|
||||
This is ordinary documentation, with no interactive demo or designed aha sequence.
|
||||
The beta launch still has no selected primary developer persona or peer-DX study.
|
||||
@@ -0,0 +1,19 @@
|
||||
# Getting-started feedback
|
||||
|
||||
The assumed existing open-source SDK uses its public repository's issue templates
|
||||
and pinned getting-started thread in Discussions. README already links this page.
|
||||
Use either path; ordinary CONTRIBUTING and support routes remain available.
|
||||
|
||||
Find the installed SDK version with `python -m pip show eval-sdk` and the Python
|
||||
version with `python --version`; neither triggers the first-run check.
|
||||
|
||||
The existing friction template and thread ask for:
|
||||
|
||||
- the step where you got stuck and the SDK/Python versions;
|
||||
- what you expected and what happened, with a redacted minimal reproducer;
|
||||
- an optional estimate of time spent at that step.
|
||||
|
||||
Do not attach secrets, API keys, private prompts or unredacted application data.
|
||||
This is voluntary support feedback through existing channels, not telemetry.
|
||||
Reports are not aggregated into a measured onboarding benchmark or a TTHW target.
|
||||
No new hosted service, automatic collection or mandatory feedback is proposed.
|
||||
@@ -0,0 +1,199 @@
|
||||
# Getting started
|
||||
|
||||
These are documentation examples for the assumed existing SDK. The SDK source,
|
||||
package and release-check implementation are absent from this review fixture.
|
||||
Do not run an install or infer a successful execution from these documents.
|
||||
|
||||
Install the assumed Python package with `pip install eval-sdk`. The mandatory
|
||||
five-minute first-run compatibility check applies to both CLI and library evals,
|
||||
including these examples, with no skip. It is not needed by the evaluator itself.
|
||||
|
||||
## Neutral first evaluation
|
||||
|
||||
```python
|
||||
from eval_sdk import evaluate
|
||||
import json
|
||||
|
||||
def target(inputs):
|
||||
return {"ready": inputs["enabled"]}
|
||||
|
||||
def exact_match(actual, expected):
|
||||
# This application's structured-output rule.
|
||||
if not isinstance(actual, dict):
|
||||
return 0.0
|
||||
return float(actual == expected)
|
||||
|
||||
cases = [{"inputs": {"enabled": True}, "expected": {"ready": True}}]
|
||||
result = evaluate(target, cases, exact_match)
|
||||
print(json.dumps([{"score": case.score, "actual": case.actual, "expected": case.expected}
|
||||
for case in result.cases], sort_keys=True))
|
||||
```
|
||||
|
||||
Shown application output (JSON from the documented structured fields, not SDK repr):
|
||||
|
||||
```text
|
||||
[{"actual": {"ready": true}, "expected": {"ready": true}, "score": 1.0}]
|
||||
```
|
||||
|
||||
Fixture checks reproduce this text with an explicit assumed-contract double and
|
||||
keep the README copy synchronized. They do not run the absent SDK or measure
|
||||
onboarding duration.
|
||||
|
||||
## Caller-owned metric for free text
|
||||
|
||||
This separate reference example supplies a real callable, cases and metric. The
|
||||
simple whitespace-insensitive metric demonstrates the API; applications choose
|
||||
their own metric and acceptance rule. It is not a production quality threshold.
|
||||
|
||||
```python
|
||||
from eval_sdk import evaluate
|
||||
import json
|
||||
|
||||
def text_metric(actual, expected):
|
||||
return float(" ".join(actual.split()) == " ".join(expected.split()))
|
||||
|
||||
def prose_target(inputs):
|
||||
return inputs["reply"]
|
||||
|
||||
cases = [
|
||||
{"inputs": {"reply": "The lamp is green."}, "expected": "The lamp is green."},
|
||||
]
|
||||
result = evaluate(prose_target, cases, text_metric)
|
||||
print(json.dumps([{"score": case.score, "actual": case.actual, "expected": case.expected}
|
||||
for case in result.cases], sort_keys=True))
|
||||
```
|
||||
|
||||
Shown free-text application output (the same explicit field projection):
|
||||
|
||||
```text
|
||||
[{"actual": "The lamp is green.", "expected": "The lamp is green.", "score": 1.0}]
|
||||
```
|
||||
|
||||
The public example returns one normal matching result with score 1.0. Fixture
|
||||
checks reproduce this text with the contract double, not the absent SDK. Separately,
|
||||
the existing product's offline checks exercise this complete callable/metric path
|
||||
with matching and mismatching prose and verify scores 1.0 and 0.0 plus the latter
|
||||
case's expected/actual failure summary.
|
||||
Structured result fields retain full values; displayed summaries may truncate.
|
||||
This reference check already exists in the revised synthetic baseline. It adds
|
||||
no launch gate, evaluator default, telemetry or designed onboarding delight beat.
|
||||
No executable SDK or assertion of its execution is supplied in this fixture.
|
||||
|
||||
## Bounded application calls
|
||||
|
||||
This is a complete **application-owned** example, separate from the SDK. The
|
||||
local transport below is free and makes no network requests. To substitute a
|
||||
paid transport, first establish a **verified upper bound** on its charge per
|
||||
invocation; this example reserves two cents per attempt. A provider without such
|
||||
a bound cannot use that reservation as a spending guarantee.
|
||||
|
||||
Save as `fixture_transport.py`:
|
||||
|
||||
```python
|
||||
import json
|
||||
import sys
|
||||
|
||||
inputs = json.load(sys.stdin)
|
||||
print(json.dumps({"ready": inputs["enabled"]}))
|
||||
```
|
||||
|
||||
Save as `bounded_client.py`:
|
||||
|
||||
```python
|
||||
import json
|
||||
import math
|
||||
import subprocess
|
||||
import threading
|
||||
|
||||
class BoundedClient:
|
||||
def __init__(self, command, *, timeout_seconds, max_attempts,
|
||||
total_cents, attempt_cents):
|
||||
if (not math.isfinite(timeout_seconds) or timeout_seconds <= 0
|
||||
or any(type(n) is not int for n in (max_attempts, total_cents, attempt_cents))
|
||||
or max_attempts < 1 or total_cents < 0 or attempt_cents < 1):
|
||||
raise ValueError("Use a positive timeout, finite attempts and integer-cent bounds")
|
||||
self.command = list(command)
|
||||
self.timeout_seconds, self.max_attempts = timeout_seconds, max_attempts
|
||||
self.total_cents, self.attempt_cents = total_cents, attempt_cents
|
||||
self.reserved_cents = 0
|
||||
self.lock = threading.Lock()
|
||||
|
||||
def __call__(self, inputs):
|
||||
for attempt in range(self.max_attempts):
|
||||
with self.lock:
|
||||
if self.reserved_cents + self.attempt_cents > self.total_cents:
|
||||
raise RuntimeError("Application spending limit reached before request")
|
||||
self.reserved_cents += self.attempt_cents
|
||||
try:
|
||||
response = subprocess.run(self.command, input=json.dumps(inputs),
|
||||
text=True, capture_output=True, check=True, timeout=self.timeout_seconds)
|
||||
return json.loads(response.stdout)
|
||||
except (subprocess.TimeoutExpired, subprocess.CalledProcessError) as error:
|
||||
if isinstance(error, subprocess.CalledProcessError) and error.returncode != 75:
|
||||
raise # Only the application's explicit temporary-failure status retries.
|
||||
if attempt + 1 == self.max_attempts:
|
||||
raise RuntimeError("Application attempt limit reached") from error
|
||||
```
|
||||
|
||||
Each transport process gets a per-attempt timeout and at most two attempts below.
|
||||
`subprocess.run` kills and waits for a timed-out direct child. This local transport
|
||||
starts no descendant processes. Killing it **does not prove that a remote provider cancelled**
|
||||
a request: its reservation is **not refunded**, even on timeout or failure. The
|
||||
shared counter refuses an attempt before the six-cent total would be exceeded;
|
||||
concurrent calls in this process share that counter. Separate application processes
|
||||
would need a shared external spending limit.
|
||||
|
||||
Use the application client in the callable (save these files together):
|
||||
|
||||
```python
|
||||
import sys
|
||||
from eval_sdk import evaluate
|
||||
from bounded_client import BoundedClient
|
||||
|
||||
client = BoundedClient([sys.executable, "fixture_transport.py"],
|
||||
timeout_seconds=2, max_attempts=2, total_cents=6, attempt_cents=2)
|
||||
|
||||
def target(inputs):
|
||||
return client(inputs)
|
||||
|
||||
def metric(actual, expected):
|
||||
return float(isinstance(actual, dict) and actual == expected)
|
||||
|
||||
cases = [{"inputs": {"enabled": True}, "expected": {"ready": True}}]
|
||||
result = evaluate(target, cases, metric, deadline_seconds=20, max_cost_usd=0.25)
|
||||
```
|
||||
|
||||
The client's timeout, attempts and reservation govern its own transport. The
|
||||
`evaluate` keywords still govern only SDK-managed scheduling/provider requests;
|
||||
they neither interrupt this application client nor add to its six-cent allowance.
|
||||
The local client/files run in fixture checks, including timeouts, retries and
|
||||
refusal before overspending. The `evaluate` call is checked with an explicit
|
||||
contract double because the SDK is absent. This is reference safety code, not a
|
||||
new metric default, launch gate, first-run benchmark or onboarding delight step.
|
||||
|
||||
## Handling errors
|
||||
|
||||
The assumed SDK's existing release checks produce this malformed-case example:
|
||||
|
||||
```text
|
||||
SDK_E001: case 0 is missing 'expected'
|
||||
Cause: CaseValidationError at cases[0].expected
|
||||
Next: add the expected output for this case and retry.
|
||||
Reference: docs/reference-v1.md#sdk-e001
|
||||
```
|
||||
|
||||
This is an authored synthetic output contract, not output obtained by executing
|
||||
the SDK here. Error codes, originating causes, actionable next steps, secret
|
||||
redaction, and versioned reference anchors are existing contracts.
|
||||
|
||||
## Next steps
|
||||
|
||||
- Use the same callable and cases in [pytest](reference-v1.md#api-and-pytest), with
|
||||
the application's own acceptance assertion.
|
||||
- Before substituting a provider-backed callable, configure [deadlines and cost
|
||||
limits](reference-v1.md#configuration). Arbitrary application requests require
|
||||
their own bounds; SDK-managed limits do not intercept them.
|
||||
- Run the [noninteractive CLI](reference-v1.md#cli) locally or in CI; the same
|
||||
invocation and exit codes apply to both.
|
||||
- Find [error codes](reference-v1.md#errors), the [beta upgrade contract](reference-v1.md#upgrades),
|
||||
and the existing [support path](feedback.md).
|
||||
@@ -0,0 +1,160 @@
|
||||
# Existing SDK reference, version 1
|
||||
|
||||
These are explicitly authored contracts for the revised synthetic fixture.
|
||||
The SDK, package, and release-check implementation are absent. SDK invocation
|
||||
examples describe its assumed interface; those calls have not been executed against
|
||||
the SDK here. Fixture checks execute the local application files and explicit
|
||||
contract doubles. This reference supplies baseline documentation, not launch remedies.
|
||||
|
||||
## API and pytest
|
||||
|
||||
`evaluate(target, cases, metric)` invokes the application's callable on each
|
||||
case's `inputs` and calls its metric with actual and expected outputs.
|
||||
`result.cases` contains per-case `score`, `actual`, and `expected` fields.
|
||||
The application supplies its acceptance rule; no score is a universal pass bar.
|
||||
|
||||
```python
|
||||
from eval_sdk import evaluate
|
||||
|
||||
def test_ready():
|
||||
def target(inputs):
|
||||
return {"ready": inputs["enabled"]}
|
||||
|
||||
def exact_match(actual, expected):
|
||||
return float(isinstance(actual, dict) and actual == expected)
|
||||
|
||||
cases = [{"inputs": {"enabled": True}, "expected": {"ready": True}}]
|
||||
result = evaluate(target, cases, exact_match)
|
||||
assert all(case.score == 1.0 for case in result.cases)
|
||||
```
|
||||
|
||||
`python -m pytest` runs this application-owned test. Public type hints and
|
||||
`py.typed` ship; the assumed release checks type-check the examples and run their
|
||||
documented outputs. This reference test remains a neutral passing example.
|
||||
|
||||
## Configuration
|
||||
|
||||
The existing optional library keywords are `deadline_seconds`, `max_cost_usd`,
|
||||
and `reporter` (`"auto"`, `"on"`, or `"off"`). For example:
|
||||
|
||||
```python
|
||||
result = evaluate(target, cases, metric, deadline_seconds=20,
|
||||
max_cost_usd=0.25, reporter="on")
|
||||
```
|
||||
|
||||
The deadline stops new case scheduling and is forwarded to SDK-managed provider
|
||||
requests. Their request timeouts and finite retries remain bounded by it.
|
||||
The cost ceiling covers only requests through that managed provider client.
|
||||
It cannot interrupt arbitrary application code or cap requests made by a separate
|
||||
client inside `target`; configure that client's timeout, retries, and spending
|
||||
limit before substituting the callable. The [worked application client](getting-started.md#bounded-application-calls)
|
||||
materializes all three bounds with a local transport and explains its verified
|
||||
per-attempt cost assumption. These boundaries apply locally and in CI.
|
||||
|
||||
Before work the CLI reports case count, deadline, and cost ceiling (or "none set")
|
||||
on stderr. The library does so on a TTY by default; `reporter` overrides that
|
||||
choice. Library reporting never writes to stdout. Values and scores are not
|
||||
persisted in a shared cache. No settings define an onboarding-time target.
|
||||
|
||||
## CLI
|
||||
|
||||
The existing noninteractive invocation uses the application's importable target
|
||||
and metric plus a JSON list of cases. The following complete files are explicit
|
||||
synthetic baseline examples; the SDK/CLI is absent, so fixture checks validate the
|
||||
files, import paths and arguments with an assumed-contract double, not the real CLI.
|
||||
This documents the shown JSON-list form only, not any other possible SDK format.
|
||||
|
||||
Save as `app.py`:
|
||||
|
||||
```python
|
||||
def target(inputs):
|
||||
return {"ready": inputs["enabled"]}
|
||||
|
||||
def metric(actual, expected):
|
||||
return float(isinstance(actual, dict) and actual == expected)
|
||||
```
|
||||
|
||||
Save as `cases.json`:
|
||||
|
||||
```json
|
||||
[
|
||||
{"inputs": {"enabled": true}, "expected": {"ready": true}}
|
||||
]
|
||||
```
|
||||
|
||||
Run with the assumed SDK from the directory containing both files:
|
||||
|
||||
```bash
|
||||
eval-sdk run --target app:target --cases cases.json --metric app:metric --deadline-seconds 20 --max-cost-usd 0.25 --no-input
|
||||
```
|
||||
|
||||
The CLI accepts the same bounds and prints readable per-case results. Exit 0
|
||||
means evaluation completed, not that an application's quality bar was met;
|
||||
the application-owned pytest assertion enforces that bar. Usage or malformed
|
||||
inputs exit 2; execution failures exit 1, with the actionable error on stderr.
|
||||
`eval-sdk --help` lists these options and noninteractive behavior.
|
||||
|
||||
Both CLI and library still block their first evaluation for the existing
|
||||
mandatory five-minute conformance check. There is no skip. This reference does
|
||||
not bypass, remove, or time that prerequisite.
|
||||
|
||||
## Errors
|
||||
|
||||
Errors expose a stable code, original cause, actionable next step, and versioned
|
||||
reference anchor. Secret values are redacted; displayed values may truncate
|
||||
while structured fields retain full values. Existing release checks verify the
|
||||
code-to-anchor mapping and example output.
|
||||
|
||||
### SDK E001
|
||||
|
||||
Missing required case input: identify the case index and missing `inputs` or
|
||||
`expected` field, then add it and retry. See the shown malformed-case output in
|
||||
[getting started](getting-started.md#handling-errors).
|
||||
|
||||
### SDK E002
|
||||
|
||||
The application metric returned a non-number: change it to return a numeric
|
||||
score. The application still chooses its own acceptable score.
|
||||
|
||||
The following E002 and E003 blocks are newly authored synthetic output contracts,
|
||||
not captured output from the absent SDK or its release checks. They make the
|
||||
existing code/cause/next-step/reference contract concrete without changing it.
|
||||
|
||||
```text
|
||||
SDK_E002: case 0 metric returned a non-number
|
||||
Cause: MetricTypeError at cases[0].score (received str)
|
||||
Next: return a numeric score from the application's metric and retry.
|
||||
Reference: docs/reference-v1.md#sdk-e002
|
||||
```
|
||||
|
||||
### SDK E003
|
||||
|
||||
A configured deadline or managed-provider cost limit was reached: inspect the
|
||||
reported bound and cause, then reduce the cases or explicitly change that bound.
|
||||
Unmanaged application requests have the separate limits described above.
|
||||
|
||||
Deadline example:
|
||||
|
||||
```text
|
||||
SDK_E003: evaluation deadline reached before case 2
|
||||
Cause: DeadlineExceeded at deadline_seconds=20
|
||||
Next: reduce the cases or explicitly choose a longer evaluation deadline.
|
||||
Reference: docs/reference-v1.md#sdk-e003
|
||||
```
|
||||
|
||||
Managed-provider cost example:
|
||||
|
||||
```text
|
||||
SDK_E003: managed-provider cost ceiling reached before case 2
|
||||
Cause: ManagedProviderCostLimit at max_cost_usd=0.25
|
||||
Next: reduce managed-provider work or explicitly choose a higher managed-provider limit.
|
||||
Reference: docs/reference-v1.md#sdk-e003
|
||||
```
|
||||
|
||||
## Upgrades
|
||||
|
||||
Beta releases preserve the published API and configuration contract. Breaking
|
||||
changes require a versioned migration guide and call-site `DeprecationWarning`
|
||||
naming the replacement, removal version, and migration anchor. Removal requires
|
||||
two minor releases of notice and a breaking release. No AST migration tool,
|
||||
plugin, new hosted documentation service, or new CI provider is part of the beta.
|
||||
Reference in New Issue
Block a user