3 people 3634c4a845 fix(chat): log the underlying error when a publication becomes delivery_unknown (#18875) ...

## Thinking Path

> - Paperclip is the open source app people use to manage AI agents for
work
> - Chat connectors (Slack and others) publish agent messages to
external conversations through the publication worker
> - When the provider accepts a message but the local commit after the
send fails, the publication becomes `delivery_unknown`
> - A `delivery_unknown` row blocks the later publications in the same
conversation until an operator resolves it
> - The log line for this case contains only the generic reason. It
drops the error that the post-send commit threw
> - Operators cannot see why the commit failed, so they cannot tell a
race from a lease loss or another cause
> - This pull request adds the underlying error, redacted and
length-limited, to the existing warning for this case only
> - The benefit is that operators and maintainers can find the real
cause of stalled chat threads

## Linked Issues or Issue Description

Refs #18874

I searched open and closed issues and pull requests for
`delivery_unknown`, Slack publication failures and "could not confirm
its durable result". I found no duplicate. #18874 describes the bug and
the proposed behavior fix. This pull request does not change behavior.
It only makes the cause visible.

## What Changed

- `server/src/services/chat-channels/publications/failures.ts`: when
`providerAccepted` is true, the `chat publication failed` warning gets a
`cause` field. The field contains the underlying error message, redacted
with `redactSensitiveText` and limited to `MAX_ERROR_TEXT`, like the
existing `error` field.
- The stored `redactedError`, the disposition and the operator
resolution flow do not change.
- `server/src/__tests__/chat-run-milestone-work.test.ts`: the existing
`rechecks resolved prompt ownership after delivery ...` test now spies
on `logger.warn`. It asserts that `delivery_unknown` logs the `cause`,
and that `retry` and `failed` do not log it.

## Verification

- Ran in a `node:24-bookworm-slim` container on this branch:
- `npx vitest run src/__tests__/chat-run-milestone-work.test.ts` (in
`server/`): 72 passed.
  - `pnpm run typecheck` (in `server/`): passed.
- Negative check: I removed only the new `cause` field and ran the test
again. The two `delivery_unknown` cases failed. The `retry` and `failed`
cases passed.

## Risks

- Low risk. The change adds one field to one warning log line, only for
publications that the provider already accepted.
- The new field uses the same redaction and length limit as the existing
`error` field, so it does not add a new path for secrets into the log.

> For core feature work, check [`ROADMAP.md`](ROADMAP.md) first and
discuss it in `#dev` before opening the PR. Feature PRs that overlap
with planned core work may need to be redirected — check the roadmap
first. See `CONTRIBUTING.md`.

## Model Used

- Anthropic Claude Opus 5.5 (`claude-opus-5-5`), 1M context window, in
Claude Code with tool use (shell, file edits, code search). A human
reviewed and requested the change.

## Checklist

- [x] I have included a thinking path that traces from project context
to this change
- [x] I have specified the model used (with version and capability
details)
- [x] I have checked ROADMAP.md and confirmed this PR does not duplicate
planned core work
- [x] I have searched GitHub for duplicate or related PRs and linked
them above
- [x] I have either (a) linked existing issues with `Fixes: #` / `Closes
#` / `Refs #` OR (b) described the issue in-PR following the relevant
issue template
- [x] I have not referenced internal/instance-local Paperclip issues or
links (only public GitHub `#NNN` / `github.com/paperclipai/paperclip`
URLs)
- [x] My branch name describes the change (e.g. `docs/...`, `fix/...`)
and contains no internal Paperclip ticket id or instance-derived details
- [x] I have run tests locally and they pass
- [x] I have added or updated tests where applicable
- [ ] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [ ] All Paperclip CI gates are green
- [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
- [x] I will address all Greptile and reviewer comments before
requesting merge

Co-authored-by: lyuboslavmedarov <285605601+lyuboslavmedarov@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

2026-11-04 17:38:39 -06:00

..

scripts

perf(chat): reuse bound-run authorization query templates (#18966)

2026-11-04 08:32:31 -06:00

src

fix(chat): log the underlying error when a publication becomes delivery_unknown (#18875)

2026-11-04 17:38:39 -06:00

test-baselines

…

CHANGELOG.md

…

package.json

…

tsconfig.json

…

vitest.backup-qualification.config.ts

…

vitest.config.ts

…