mirror of
https://github.com/paperclipai/paperclip.git
synced 2026-10-07 07:23:08 +02:00
DottaandPaperclip
4cd7b40255
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work.
> - Native recovery must distinguish a fresh user request from replay of
failed work.
> - Older runs can lose their process fields before a local stop receipt
exists.
> - A suspended durable session can still prove that the exact runner
and provider session are idle.
> - The message admission path ignored that evidence and kept new user
messages blocked.
> - This pull request uses the existing exact-state verifier for those
historical runs.
> - The user can start one fresh turn while the old history and unknown
outcomes remain intact.
## Linked Issues or Issue Description
**What happened?**
A user sent a new message after a native run exhausted recovery.
Paperclip saved the message but said the previous run had no verified
stop record. The old runner was suspended, with no active provider turn
or pending output. Its process fields had been cleared before stop
receipts were added.
**Expected behavior**
A new user message starts a fresh turn when the exact retained session
proves it is suspended and the other execution gates pass.
**Steps to reproduce**
1. Retain a failed native run with a terminal controller, cleared
process fields, and no process receipt events.
2. Retain its exact suspended runner state and idle provider state. Keep
its recovery hold.
3. Send a new user comment. Before this fix, admission returns no
successor.
**Paperclip version or commit**
Reproduced on master at d351e08de.
**Deployment mode**
Self-hosted server. The regression uses an embedded PostgreSQL test
database and real durable state files.
Related work: Refs #13270, Refs #13338. This adds compatibility for
older stopped runs. Refs #13332 concerns separate recovery-hold scope
rules.
## What Changed
- Add exact suspended-state evidence to explicit native message
admission for runs that predate process receipts.
- Reuse the existing failed-retry verifier for run, runner, workspace,
provider identity, and pending-work checks.
- Reject this fallback if any server-authored process receipt or launch
event exists.
- Add a red/green regression with real message admission, duplicate
delivery, dry-run behavior, and blocked-state cases.
- Document the new-message recovery rule.
## Verification
- Red: the historical suspended-state regression failed on master
because admission returned null. Eight rejection cases passed.
- Green: 469 tests passed across explicit native continuation and native
session execution.
- Full repository `pnpm -r typecheck` and `pnpm build` passed locally.
- The complete Vitest CI matrix and all browser test shards passed. The
duplicate local `pnpm test:run` was stopped after these CI results; it
did not complete locally.
- The first runner verification worker received an infrastructure
shutdown signal during compilation. The single retry passed.
- Greptile reviewed commit `6836d7310` at 5/5 with no findings.
- Inspected the affected server's database and durable state read-only.
It has the historical missing-PID shape and an exact suspended runner
with no active provider turn, pending tools, or undelivered output.
## Risks
- Incorrect idle evidence could allow overlapping work. The fallback
requires an exact suspended root and rejects active or pending work,
mismatched identities, missing files, and newer process evidence.
- Normal task, controller, lease cleanup, decision, and active-run gates
remain in force.
- This does not resume old provider actions or reset recovery attempts.
Unknown outcomes remain unknown.
- No schema change or deployment is included.
## Model Used
OpenAI Codex, GPT-6. The session does not expose the exact model
snapshot or context-window size. Used reasoning, repository inspection,
code editing, shell execution, and test tools.
## 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
- [x] I have updated relevant documentation to reflect my changes
- [x] I have considered and documented any risks above
- [x] All Paperclip CI gates are green
- [x] 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: Paperclip <noreply@paperclip.ing>