# Description of Changes
Redesign policy running so the server is in charge of policy IDs and
running, to make it impossible to have the frontend miss the results.
This solves a minor bug that we currently have in policies, where if you
load a file and then refresh while the policy is running, you'll never
receive the outputted file.
## What & why
Production reports of policy enforcement "hanging" traced to
large/many-page documents: the watermark step's flatten-to-image
(`convertPDFToImage`) on a 500+ page PDF takes minutes, exceeding both
the client poll cap and the backend per-step timeout. This makes the
slow case graceful instead of looking broken, and makes load-shedding
non-fatal.
### Poll runs to completion (no false "hang")
The client poll loop used a flat ~150s cap that was **shorter than the
backend's 300s per-step timeout**, so it abandoned long-but-healthy runs
mid-flight. The budget is now sized to the backend's real worst case —
`stepCount × per-step timeout + grace`, learned from the first status
report — so the client always polls long enough to surface the run's
**actual** terminal state (success or the backend's real error) rather
than a misleading client-side timeout.
### Per-step progress
The activity feed now shows `Enforcing… · step n/m` (from
`currentStep`/`stepCount`), so a slow step shows movement instead of a
dead spinner.
### Soft-retry on queue rejection
Under load the shared `JobQueue` rejects runs ("queue full"), which
previously surfaced as a hard failure needing a manual Retry. The
backend now tags that rejection with a stable `POLICY_QUEUE_FULL`
errorCode; the client treats it as transient backpressure and
**auto-retries the file in place** with exponential backoff (≈4s→64s, ~2
min), shown as a soft "Busy — retrying…" row, falling back to the manual
Retry only once the retry budget is spent.
## Testing
- **Frontend unit tests** (30 pass across the policies suite), including
a new `usePolicyAutoRun.retry.test.tsx` that drives the real controller
orchestration (poll → `POLICY_QUEUE_FULL` → relabel → backoff → in-place
re-dispatch), plus poll-budget, step-progress, and activity-feed relabel
cases.
- **Backend** `PolicyEngineTest` case asserting a queue-rejected run
carries the `POLICY_QUEUE_FULL` code.
- Typecheck clean on all three flavors (proprietary/saas/core); prettier
+ spotless clean.
- Poll-budget + progress + real-error surfacing were also verified live
end-to-end against a 599-page run (survived past the old cap, showed
step progress, reported the backend's real 300s-timeout failure,
recovered after a simulated network drop).
## Not included (follow-ups)
- The underlying flatten-to-image cost itself (bounded-memory/streaming
flatten, revisiting `convertPDFToImage` default and the 300s timeout) —
the real perf fix, deliberately out of scope here.
## Problem
We're getting 402s when an AI **agent** (chat) run hits the free
allowance / spending cap, but the frontend handles them poorly and never
pops the usage-limit modal.
The agent runs its tool calls **server-side** (loopback HTTP via
`PolicyExecutor`), so the 402 never reaches the `apiClient` interceptor
that pops the modal for direct calls. It was caught by the generic
tool-failure handler and flattened into a `CANNOT_CONTINUE` reason
string (`"The /api/v1/… tool failed: 402…"`), streamed as a `result`
event, and rendered as a scary chat bubble. This is the same gap the
policy auto-run path bridges (#6626) — one layer up.
## Fix
**Backend** (`proprietary`)
- `AiWorkflowResponse` gains `errorCode` + `errorSubscribed`.
- `AiWorkflowService` detects a downstream 401/402 entitlement sentinel
in its three tool-exec catch sites (`onToolCall`, `runPlan`,
`onConvertMarkdown`) and surfaces the structured code (+ `subscribed`)
on the terminal response instead of the raw failure text.
- Factored the 401/402 body extraction `PolicyEngine` already had into a
shared `DownstreamEntitlementError` util so the two server-side paths
can't drift.
**Frontend**
- New `usageLimitBridge` (`PAYG_LIMIT_REACHED_EVENT` +
`dispatchPaygLimitReached`) generalises the previously policy-only
bridge. Proprietary can't import the saas modal API (layering), so
server-side limit hits broadcast a window event the saas
`UsageLimitModalHost` opens the modal from. Migrated the policy path
onto it.
- `ChatContext` fires the matching modal (free → subscribe, subscribed →
raise cap) on the limit result **and** on a direct 402, replacing the
raw reason with a brief friendly line
(`chat.responses.usage_limit_reached`).
No Python engine changes — the charge/402 happens on the Java tool
endpoint that Java itself calls.
## Test plan
- [x] `:proprietary:compileJava` + `spotlessCheck` clean
- [x] `AiWorkflowServiceTest` + `PolicyEngineTest` green
- [x] eslint, proprietary + saas typechecks clean
- [ ] Manual: drive an agent run over the limit → brief line in chat +
the right modal (free vs cap)
> Note: proprietary test compilation is currently blocked on the
pre-existing `InitialSecuritySetupTest` 6-arg ctor break (unrelated,
tracked separately); verified locally by temporarily patching it.
Follow-up to #6604 (merged). Builds the Security policy out so it
actually enforces, driven from the editor.
## What it does
- **Run on upload or export** — a single choice in the wizard: enforce
when a file is uploaded, or just before it's exported.
- **Output** — enforced result is a **new version** of the file
(default) or a **new file**, with optional filename
prefix/suffix/auto-number ("Output filename" subsection; auto-number
only for new files).
- **Export enforcement** — exporting an export-mode file runs the policy
first and downloads the enforced result; never hard-blocks (on failure
the original downloads). For new-version policies the in-editor file is
versioned too. Covers every export path incl. multi-file ZIP. A toast
(glowing in the policy's accent while it runs) reports progress and
fades after ~10s.
- **Affordances** — a freshly enforced file briefly glows its policy
accent and carries a shield badge.
- **Config tidy-up** — removed the unwired Security setting fields + the
wizard's review step; "Upgrade to enterprise" on locked categories;
category accent in the detail/wizard headers.
## Notes
- Builds on the manual-only (client-driven) policy model from #6587
(`trigger: null`, metadata in `output.options`), adding the `runOn`
field + export-time enforcement.
- The page-editor merge-export (no single source file) enforces +
downloads but doesn't version in place.
## Verification
typecheck (core + proprietary), eslint, prettier; proprietary suite
(105) green; flows checked in-app.
Follow-up to #6598 (squash-merged into `SaaS`). These are the policy
refinements made after that merge, against the current `SaaS` tip.
## Changes
- **Simplify Security config + plain-language info buttons** — Redact
config reduced to the PII field; Sanitise has no config
(JavaScript-removal only) with a non-technical info button; per-tool
info buttons reworded to match the tool-steps style.
- **Hide 'Flatten PDF pages to images' from the watermark policy
config** — new `PolicyWatermarkConfig` wrapping the watermark settings
with the flatten checkbox gated off.
- **Flatten-to-image on by default for redact + watermark** — both
normalise `convertPDFToImage: true` on mount.
- **Self-heal a stale backing folder** — `ensurePolicyFolder` recreates
a backing folder whose `folderId` no longer resolves (preferring the
backend's stored automation), instead of hanging Edit Settings on a
permanent "Loading…".
- **Version the input file on 'new version' output mode** — completed
runs whose policy output mode is `new_version` replace the input file
with a versioned child (origin tool `automate`) rather than adding a
separate file; falls back to a new file if the input is gone.
`outputMode` is plumbed through `PolicyState`, the local-cache default,
and backend reconciliation.
## Verification
- `typecheck:proprietary` + `typecheck:core` clean
- policy + hooks vitest: 17 passing
- eslint + prettier clean on all changed files
## Summary
Adds the **Policies** feature (proprietary, behind the
`POLICIES_ENABLED` flag): backend-driven enforcement that runs a fixed
tool pipeline on documents, docked in the right tool sidebar alongside
Tools.
## Highlights
- **Policy catalog** — 5 categories; **Security** is wired (redact PII +
sanitize), the others are marked "Coming soon".
- **Backend as source of truth** — policies persist via the Policies
engine (`/api/v1/policies`), one policy per category, with a local cache
+ offline fallback.
- **Auto-run** — enabled policies run on every uploaded file: dispatch →
poll → import outputs into the workspace.
- **Security redact config** — PII preset dropdown + custom word/regex
entry + advanced options; tool params map to the backend endpoint
fields.
- **Activity feed** with retry on failures; **file badges** showing
which policies ran on a file (sidebar + files page), tinted to the
policy colour.
- Reuses the **Watched Folders** engine for each policy's backing
folder; policy-owned folders are filtered out of the Watched Folders UI.
## Notes
- Gated by `POLICIES_ENABLED` (true in proprietary, false in core) —
unreachable in the open-source build.
- Frontend-only diff; depends on the backend Policies engine and the
merged Watched Folders feature.
James BruntonandGitHub
Reece BrowneandGitHub
ConnorYohandGitHub