## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Planning work relies on issue documents, request-confirmation interactions, and inline plan annotations > - Agents can be woken after a plan comment, annotation, or confirmation decision > - The wake payload needs enough plan-review context for the agent to act on the specific feedback instead of losing the thread and falling back to broad refetches > - This pull request adds bounded plan-review context to wake payloads and heartbeat context > - It also teaches the adapter wake prompt renderer to surface those open plan annotations and interaction results directly > - The benefit is that agents can continue plan review and plan acceptance flows with the relevant comments in hand while keeping wake payloads bounded and company-scoped ## Linked Issues or Issue Description No matching public GitHub issue was found. ### Subsystem affected Cross-cutting: `server/`, `packages/shared`, and `packages/adapter-utils`. ### Problem or motivation Plan-review continuations can wake an agent after a plan comment, inline annotation, or request-confirmation decision without enough inline context about the open plan annotations or accepted/rejected confirmation target. That makes scoped wakes less reliable because the agent may need to refetch broad issue history before it can tell what feedback should be incorporated. ### Proposed solution Include bounded, company-scoped plan review context in wake payloads and heartbeat context. The context includes open `plan` annotation threads, recent annotation comments, truncation metadata, and plan-confirmation interaction target/result details. Render that information in the adapter wake prompt so agents see the relevant plan-review feedback immediately. ### Alternatives considered Relying on agents to fetch the full issue thread after every plan-review wake was rejected because it is slower, harder to audit, and easier to mishandle when the wake is meant to be scoped to a specific comment, annotation, or interaction result. ### Roadmap alignment This supports the roadmap areas for Agent Reviews and Approvals, Deep Planning, and Enforced Outcomes by making plan approval continuations explicit and actionable. ### Additional context The implementation keeps payload size bounded with per-thread, per-comment, and total-body limits. Resolved annotation threads are intentionally omitted so the wake focuses on feedback still needing action. ## What Changed - Added shared `PlanReviewContext` types for plan annotation threads, comments, interaction targets, and continuation results. - Added server-side plan review context assembly for open `plan` annotation threads with bounded thread/comment/body limits. - Included plan review context in heartbeat context and scoped wake payloads for planning, annotation, comment, and plan-confirmation interaction wakes. - Rendered plan annotation deltas, open plan comments, interaction results, and accepted target revisions in adapter wake prompts. - Added focused regression coverage for scoped plan review context, wake prompt rendering, annotation filtering, and safe standard-mode annotation wakes. - Addressed Greptile feedback by bounding the plan-comment DB fetch and removing unused plan review context input fields. ## Verification - `pnpm run preflight:workspace-links` - `pnpm exec vitest run --project @paperclipai/adapter-utils packages/adapter-utils/src/server-utils.test.ts` - `pnpm exec vitest run --project @paperclipai/server --no-file-parallelism --maxWorkers=1 server/src/__tests__/document-annotations-service.test.ts server/src/__tests__/issue-thread-interaction-routes.test.ts server/src/__tests__/issues-goal-context-routes.test.ts` - `pnpm --filter @paperclipai/shared typecheck` - `pnpm --filter @paperclipai/adapter-utils typecheck` - `pnpm --filter @paperclipai/server typecheck` - `git diff --check public-gh/master...HEAD` - GitHub PR checks are green on `36d0ac6a5dce27b9d62e201bf6d9829170c5974e`n- Rebased onto current `paperclipai/paperclip:master` and confirmed GitHub reports the PR as mergeable - Greptile Review completed successfully after 2 comments were addressed and resolved; 0 unresolved review threads remain ## Risks - Medium: wake payloads now include additional plan-review data, so limits and truncation behavior need to stay conservative as annotation volume grows. - Low migration risk: no database schema or migration changes. - Low repository hygiene risk: this PR does not touch `pnpm-lock.yaml`, `.github/workflows`, or media assets. > 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 OpenAI Codex using `gpt-5` as a coding agent with shell/tool execution. Reasoning mode and exact context window were not exposed by the runtime. ## 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>
@paperclipai/adapter-utils
Shared utilities for Paperclip adapters: process spawning, environment injection, sandbox/SSH transport, workspace sync, and the round-trip helpers that move code between the local execution-workspace cwd and wherever the agent actually runs.
For the adapter-author guide see
docs/adapters/creating-an-adapter.md
and the in-repo notes at packages/adapters/AUTHORING.md.
No-remote-git contract
The local execution-workspace cwd is the only persistence boundary across runs. No adapter may depend on a git remote for cross-run state.
Adapters that run the agent on a different host should use the SSH round-trip
helpers in src/ssh.ts:
prepareWorkspaceForSshExecution({ spec, localDir, remoteDir })— bundles the local cwd (tracked files, dirty edits, untracked additions, and the git history needed to reconstruct it) toremoteDirbefore the run starts. Runs with nogit remoteconfigured.restoreWorkspaceFromSshExecution({ spec, localDir, remoteDir, ... })— syncs the remote cwd back intolocalDirafter the run, including any new commits the agent created. Also runs with nogit remoteconfigured.
prepareRemoteManagedRuntime in
src/remote-managed-runtime.ts wraps both
calls for adapters that want a per-run remote workspace and an automatic
restoreWorkspace() finally hook.
The invariant is pinned by the no-remote-git contract case in
src/ssh-fixture.test.ts, which asserts that a
remote-only commit propagates to the local worktree through the
prepare → restore round-trip with no git remote configured at any point. Do
not regress that test.