Files
Guilhem eb238e3f0b fix: stop the AI chat destroying secret variables on edit (#10616)
* fix: stop the AI chat destroying secret variables on edit

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: clear stale staged secret values and state the draft-staging rule

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* docs: condense the pending-secret invariant to its field

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: refuse empty and oauth-managed secret values, keep drawer-staged ones in the draft

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: resolve a variable deploy's secret from one draft snapshot

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* refactor: make the variable draft the single source of a staged secret

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* docs: drop stale in-memory secret invariants from comments and the eval

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: stop null account/expires_at leaking into variable drafts and diffs

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: report when a variable deploy leaves the secret value unchanged

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* docs: scope the variable-value readability claims to the chat

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* docs: correct the secret-draft invariant in the diff masking comment

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* docs: record why a non-secret value is resent on a partial update

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: stop "Load secret value" discarding a staged secret

The audit-logged load writes the deployed secret into the draft row the
variable drawer shares with the AI chat, so offering it while that row
already stages a value silently replaces it — and the deploy that follows
carries the old value with no sign the staged one was lost.

The gate that hid the action already existed but keyed on
`isEncryptedDraftValue`, which only holds once a draft has round-tripped
through the server. A value staged in the same tab is still plaintext, so
it slipped through. Key on "anything staged" instead; clearing stays
explicit via Reset.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: extend the variable draft's empty-value sentinel past secrets

Two gaps in the chat's variable write path, both from treating "the draft
cannot carry this value" as meaning only "the value is secret".

`variableToDraftState` drops the value of an OAuth-managed variable so a
refreshed live token is never pinned into a draft, leaving '' behind. The
deploy body resent that '' verbatim for a non-secret one, wiping the token
the refresh flow owns. The sentinel now covers every value the draft is not
allowed to hold, which also removes the divergence from
`VariableEditor.save` and the shared deployer.

Making a variable secret when it holds no value produced a secret draft
staging '', a deploy body with no `value`, and the backend's "cannot change
is_secret without updating value too" — the sibling create path already
answers that case with guidance, so answer it here too.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

* fix: gate the Secret toggle's secret load on the staged value too

The toggle calls `onLoadSecret` on every change so an is_secret flip has a
value to send, but that load overwrites the shared draft row — the same
discard the button gate just closed, reached by a different control.

It now loads only when the row stages nothing, which is exactly when the
flip needs a value fetched. With a value already staged there is one to
send, and it is the one the user or the chat put there.

Blocking the load costs the side effect that used to mask a worse bug: for
a deployed variable, the load replaced an `$encrypted:` marker with real
plaintext before save. Without it, un-securing a marker would store the
marker string as the value, since the deploy endpoints only decrypt it while
is_secret stays true. So the toggle is disabled outright while a marker is
staged — Reset first. That closes the marker case for draft-only variables
as well, where no load could ever have masked it.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-12 13:13:41 +02:00
..

AI Evals

Small benchmark runner for the Windmill AI generation modes:

  • cli
  • flow
  • script
  • app
  • global

The benchmark always tests the current production prompts, tools, and guidance in this checkout.

Each attempt runs:

  1. the real production path
  2. deterministic validation
  3. LLM judging

Install

cd ai_evals
bun install

Frontend modes also require frontend dependencies:

cd frontend
bun install

Commands

List model aliases:

cd ai_evals
bun run cli -- models

List cases:

cd ai_evals
bun run cli -- cases
bun run cli -- cases flow

Run benchmarks:

cd ai_evals
bun run cli -- run flow
bun run cli -- run flow flow-test4-order-processing-loop --model opus
bun run cli -- run flow flow-test0-sum-two-numbers --models haiku,opus,4o
bun run cli -- run flow flow-test0-sum-two-numbers --runs 3 --verbose
bun run cli -- run flow --record
GEMINI_API_KEY=... bun run cli -- run app app-test1-counter-create --model gemini-3-flash-preview
WMILL_AI_EVAL_BACKEND_URL=http://127.0.0.1:8000 bun run cli -- run flow --backend-validation preview
bun run cli -- run global global-test1-script-create
bun run cli -- run cli bun-hello-script

Public CLI surface:

  • models
  • cases [mode]
  • run <mode> [caseIds...]

run options:

  • --runs <n>: repeat each case n times
  • --output <path>: custom result JSON path
  • --model <alias>: choose the model under test
  • --models <a,b,c>: run the same cases sequentially against several model aliases
  • --verbose: stream assistant output for frontend runs
  • --skip-judge: skip LLM judge scoring for the run
  • --execution-only: only require the model/proxy/frontend loop to complete; skip validators, tool expectations, backend artifact validation, and judge scoring
  • --record: append a compact tracked summary line to ai_evals/history/<mode>.jsonl for full-suite runs only
  • --backend-validation <mode>: optional backend smoke validation (off or preview) for script and flow evals

Models

Use bun run cli -- models to see the current aliases.

Today:

  • haiku
  • sonnet
  • opus
  • 4o
  • gpt-5.5
  • gemini-3-flash-preview
  • gemini-3.1-pro-preview
  • deepseek-v4-flash
  • deepseek-v4-pro

Notes:

  • the command also prints accepted alias spellings such as gpt-4o, gpt-55, claude-opus-4.6, and claude-haiku-4.5
  • frontend modes (flow, script, app, global) can use Anthropic, OpenAI, Gemini, and DeepSeek-backed aliases
  • cli mode always uses the Anthropic agent SDK, so only Anthropic aliases are valid there
  • the judge model is separate and currently defaults to claude-sonnet-4-6; use --skip-judge for deterministic-only runs

Case Format

Cases live in one YAML file per mode under ai_evals/cases/.

Minimal shape:

- id: flow-test0-sum-two-numbers
  prompt: |-
    Create a flow that takes two numbers, `a` and `b`, and returns their sum.
  initial: ai_evals/fixtures/...
  expected: ai_evals/fixtures/...

Optional fields:

  • initial: starting state fixture
  • expected: expected artifact fixture
  • validate: extra deterministic validation rules
  • runtime.backendPreview: optional real backend preview config for smoke validation

For flow mode, validate can express requirements such as:

  • accepted input schema shapes
  • required results.* reference validity
  • required module/code/input characteristics

For app mode, validate can express narrow hard requirements such as:

  • required frontend file paths or backend runnable keys
  • minimum backend runnable counts
  • required backend runnable types
  • minimum datatable / datatable-table counts
  • specific required datatable tables

For global mode, validate can express draft-level requirements such as:

  • required draft type/path/language
  • required or forbidden snippets in draft values
  • required or forbidden draft counts
  • forbidden draft paths

Global initial fixtures can also seed liveEditorDrafts with type, storagePath, effectivePath, and value fields. These drafts emulate the currently open script, flow, or raw app editor so cases can test prompts that refer to "this" or the "current" item.

Global initial fixtures can seed workspace.variables with { path, value, is_secret, description?, labels?, ws_specific? } entries so cases can read and edit variables that already exist in the workspace. The mock mirrors the real get_variable, decrypt-by-default included: a secret's value is withheld only when the caller explicitly passes decryptSecret: false, and omitting the flag returns the decrypted value, exactly as against a real backend. The chat's read path passes decryptSecret: false, so a case can verify it never invents a value it was not shown. Seed a recognizable secret (the existing fixture uses sk_live_do_not_leak_me) and assert it via valueExcludes to catch a leak.

toolExpect.toolCallArgs entries additionally support fieldMustBeAbsent: true: no recorded call to that tool may pass the field at all (an explicit null counts as passing it). Use it for partial-update tools, where supplying a field the model could not have read is itself the failure — e.g. write_variable.value on a secret variable.

Global (and flow) initial fixtures can seed workspace.datatables so the list_datatables, get_datatable_table_schema, and exec_datatable_sql tools return seeded data during evals. Each entry is { datatable_name, schemas: { <schema>: { <table>: { columns, rows? } } } }. SQL runs through a small in-memory engine (datatableSqlEngine.ts), not a real database. Writes are stateful within a case: CREATE/DROP/INSERT/UPDATE/ DELETE mutate the seeded datatable in place, so a later list_datatables, get_datatable_table_schema, SELECT, or information_schema query reflects them — this is what stops a model from looping when it re-queries to verify a write. The engine is best-effort: SELECT returns all rows of the referenced (or first) table with no WHERE filtering/projection/joins, WHERE on UPDATE/DELETE supports col = value predicates joined by AND, and anything unparseable is a no-op success. So validate datatable cases through tool-use and SQL-argument assertions (requiredToolsUsed, stringIncludesAnyOf) — not through exact returned row values. An empty/absent datatables seed makes list_datatables return [], which is what the "no datatable configured" blocking cases rely on.

Set WMILL_AI_EVAL_DISABLE_ACTIVE_EDITOR_CONTEXT=1 to run those cases with the old behavior where the live editor is only discoverable through list_workspace_items.

App fixtures can also include an optional datatables.json file at the fixture root.

For flow mode, an initial fixture can also include a benchmark workspace catalog of existing scripts and flows. That lets the real search_workspace and get_runnable_details tools discover reusable workspace runnables during evals.

If --backend-validation preview is enabled:

  • script evals run a real backend script preview in an isolated temp workspace
  • flow evals run a real backend flow preview only for cases that define runtime.backendPreview
  • flow cases with initial.workspace fixtures seed those scripts and flows into the preview workspace before preview
  • when WMILL_AI_EVAL_BACKEND_WORKSPACE is set, ai_evals creates or reuses that workspace as a dedicated test workspace, clears managed eval assets under f/evals/* before each preview run, and then reseeds the current case fixtures

Supported backend env vars:

  • WMILL_AI_EVAL_BACKEND_VALIDATION=preview
  • WMILL_AI_EVAL_BACKEND_URL=http://127.0.0.1:8000
  • WMILL_AI_EVAL_BACKEND_EMAIL=admin@windmill.dev
  • WMILL_AI_EVAL_BACKEND_PASSWORD=changeme
  • WMILL_AI_EVAL_BACKEND_WORKSPACE=integration-tests to reuse an existing workspace on CE installs with low workspace limits

Frontend modes require a reachable Windmill backend and send model requests through the workspace AI proxy at /api/w/{workspace}/ai/proxy. At startup, ai_evals checks the resolved backend URL and fails early with setup guidance if the backend cannot be reached or login fails.

For frontend modes:

  • ai_evals creates a temporary backend workspace, or creates/reuses WMILL_AI_EVAL_BACKEND_WORKSPACE when it is set
  • it upserts a provider resource under f/evals/ai/<provider>
  • frontend requests go through /api/w/{workspace}/ai/proxy

Results And Artifacts

Every run writes:

  • a summary JSON under ai_evals/results/
  • generated artifacts in a sibling directory

If --record is used, the CLI also appends one compact JSON line to:

  • ai_evals/history/flow.jsonl
  • ai_evals/history/script.jsonl
  • ai_evals/history/app.jsonl
  • ai_evals/history/global.jsonl
  • ai_evals/history/cli.jsonl

Each recorded line contains:

  • run metadata (createdAt, gitSha, mode, runModel, judgeModel)
  • suite totals (caseCount, attemptCount, passedAttempts, passRate, averageDurationMs, averagePassedDurationMs, averageJudgeScore)
  • average token usage (averageTokenUsagePerAttempt, averageTokenUsagePerPassedAttempt)
  • per-case metrics under cases[] (averageDurationMs, averagePassedDurationMs, averageJudgeScore, averageTokenUsagePerAttempt, averageTokenUsagePerPassedAttempt, pass rate)
  • failedCaseIds

The CLI headline duration and token averages use passed attempts only. All-attempt averages are still recorded to make failures auditable without letting failed attempts skew success cost comparisons.

Example:

  • summary: ai_evals/results/2026-04-09T09-40-33.051Z__flow.json
  • artifacts: ai_evals/results/2026-04-09T09-40-33.051Z__flow/

Typical artifacts by mode:

  • flow: flow.json
  • script: script.json plus the generated script file
  • app: app.json plus frontend/backend files
  • global: global-drafts.json
  • cli: assistant-output.txt, trace.json, wmill-invocations.jsonl, plus generated workspace files
  • backend-validated attempts also include backend-preview.json

Layout

  • cases/: one YAML file per mode
  • fixtures/: initial and expected fixtures
  • core/: shared loading, model resolution, validation, judging, and result writing
  • modes/: one runner per mode
  • history/: optional tracked pass-rate history written by run --record, one JSONL file per mode
  • results/: local benchmark output and artifacts

Notes

  • Frontend modes reuse the production frontend chat code through the Vitest bridge.
  • Global mode evaluates the production global AI tools and validates the resulting AI draft store.
  • CLI mode creates an isolated workspace, writes the current checkout guidance into it, and benchmarks the real skills / AGENTS.md flow.
  • CLI mode now also records a structured trace of invoked skills, tool calls, proposed wmill commands, and any attempted wmill executions.
  • Frontend progress streams live while the benchmark is running.
  • Deterministic validators should stay focused on real correctness constraints, not one exact implementation shape.