mirror of
https://github.com/paperclipai/paperclip.git
synced 2026-10-12 08:19:54 +02:00
bf3eecdd2d888fe548c59e459ab25a66520d652b
17
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 Devin FoleyandPaperclip
|
eab93fd4a0 |
fix: checkpoint adapter usage and preserve unknown prices (#14991)
Co-Authored-By: Paperclip <noreply@paperclip.ing> |
||
|
|
233c12f029 |
feat: add kimi-local adapter for Kimi Code CLI (CLI + ACP engines) (#9967)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Local agent adapters (`claude_local`, `gemini_local`, `grok_local`, …) are the integration surface that lets Paperclip run coding CLIs on the host machine > - The Kimi Code CLI (`kimi`, Moonshot AI) has a documented non-interactive mode, `kimi -p --output-format stream-json` with session resume via `kimi -r`, but Paperclip has no built-in adapter for it > - So Kimi users (especially Kimi membership / OAuth subscribers) cannot onboard their CLI to Paperclip agent teams > - This pull request adds a complete built-in `kimi_local` adapter (both execution engines, session management, instructions + skills delivery, thinking-effort control, environment test, UI and CLI modules, docs) following the established `gemini_local`/`grok_local` package pattern > - Kimi Code ships an ACP server (`kimi acp`), so the adapter runs on Paperclip's shared acpx engine by default (streaming transcript with live tool status, like `claude_local`/`gemini_local`) and falls back to a headless CLI lane (`kimi -p --output-format stream-json`) when ACP prerequisites are unavailable > - The benefit is that Kimi Code becomes a first-class Paperclip agent lane: selectable in the UI, resumable across heartbeats, with the same operating context (instruction bundle, skills, effort) and streaming transcript the other local adapters get ## Linked Issues or Issue Description - Supersedes #9880 (same branch; expanded from the CLI-only lane into a complete adapter with the default ACP engine lane, control-plane skill install, and live transcript wiring) - Refs #9879 (adapter request for Kimi Code CLI, filed with this PR) - Refs #163 (original Kimi support request) Duplicate/related prior PRs, per the dedup search (both appear stale: no updates or maintainer review since May 2026, and both target an older Kimi CLI interface; calling them out for reviewer context per CONTRIBUTING.md): - Refs #6276 (`feat: add kimi-local adapter`): targets an older array-based content format (`{type: think}`/`{type: text}` blocks), not the current documented stream-json schema - Refs #5202 (`feat(adapter): add Kimi CLI local adapter with Wire protocol support`): builds on a `--wire` JSON-RPC interface that current Kimi Code CLI (0.27.0) no longer documents; the current documented headless interface is `-p --output-format stream-json` This PR is a fresh implementation against current master and the currently documented/verified Kimi CLI behavior (see Verification). Happy to fold in anything useful from the earlier attempts if a reviewer prefers. ## What Changed - **New adapter package** `packages/adapters/kimi-local` (`@paperclipai/adapter-kimi-local`), modeled on `gemini-local`/`grok-local`: - `src/server/execute.ts`: spawns `kimi -p <prompt> --output-format stream-json` (argv array, no shell), `-m <model>` only when configured, `-r <sessionId>` when the stored session cwd matches the run cwd, automatic fresh-session retry on unrecoverable-session errors, headless-safe env (`CI=1`, `NO_COLOR=1`, `KIMI_CODE_NO_AUTO_UPDATE=1`, `TERM=dumb`; user-configured values win), full remote (ssh/sandbox) execution lane with runtime install via `@moonshot-ai/kimi-code` - **Instruction bundle delivery**: the prompt path directive now names the sibling instruction files (`./HEARTBEAT.md`, `./SOUL.md`, `./TOOLS.md`) alongside the prepended entry file, and local runs pass `--add-dir <instructions-dir>` so Kimi can actually open them (matching `claude_local`). Without this, only the entry file reached Kimi and agents improvised the operating workflow that `HEARTBEAT.md` documents - **Thinking effort**: a configured `effort` is forwarded as the `KIMI_MODEL_THINKING_EFFORT` operational override (Kimi has no per-invocation effort flag). It is only sent for models that advertise `support_efforts` (currently `kimi-code/k3`) to avoid provider rejections, and `medium` maps to `high` since Kimi has no medium tier (`low`/`high`/`max` pass through) - **Skills delivery**: desired Paperclip skills are delivered via Kimi's `--skills-dir` flag from a dedicated per-run directory (a local snapshot, or the synced snapshot on remote targets), so skills load reliably and in isolation. Paperclip never overwrites the shared `$KIMI_CODE_HOME/skills` home, so skills installed by the operator or other agents are left intact. `--skills-dir` is only passed when at least one skill is desired, so unconfigured agents keep Kimi's default skill discovery - **Live run status**: the adapter now forwards each streamed stream-json line to `onEvent` (assistant `content` as an assistant snippet, `tool_calls` as tool-name events), which drives the issue-thread activity indicator (`currentToolName` / `lastAssistantSnippet` / `lastEventAt`). Previously the adapter only wrote the raw run log, so the issue thread showed a stale "no output for N s" line with no tool or reasoning context while Kimi worked. Tool results are omitted so the last meaningful "Using X" / snippet is not overwritten by a generic label - `src/server/parse.ts`: parses the verified Kimi stream-json event shapes (`assistant` text, `assistant.tool_calls` with JSON-string arguments, `tool` results, trailing `meta.session.resume_hint` for session-id capture) plus failure classifiers (`kimi_auth_required`, transient network, unrecoverable session). A signaled exit (null exit code, not a timeout) is now reported as a failure rather than coalesced to success, and the error message names the terminating signal - `src/server/skills.ts`: lists/syncs Paperclip skills for the adapter's skill-management surface - `src/server/test.ts`: environment test covering CLI resolution + `kimi --version`, cwd check, auth detection (OAuth credential dirs, keyed `[providers.*]` in config.toml, or the `KIMI_MODEL_NAME` + `KIMI_MODEL_API_KEY` env pair), and a live hello probe - `src/ui/` (stdout-line parser for transcripts, config builder) and `src/cli/` (stream event formatter) modules - Root metadata: three managed model aliases (`kimi-code/kimi-for-coding`, `kimi-code/kimi-for-coding-highspeed`, `kimi-code/k3`), effort-capable-model metadata (`EFFORT_CAPABLE_MODELS`, effort mapping helpers), `agentConfigurationDoc` - Tests: 101 tests across parse, execute (args building, resume gating, retry, auth error code, timeout, signaled-exit failure, effort forwarding/gating/mapping, `--add-dir` instructions directive, `--skills-dir` gating, `onEvent` runtime-event forwarding), ACP engine (engine resolution, acpx config build, node-version gate), ACP transcript delegation, environment test, UI parse/build-config - **ACP engine lane (default)** (`src/server/acp.ts` + shared `adapter-utils/acpx-engine`): Kimi Code ships an ACP server (`kimi acp`), so `kimi_local` now runs on Paperclip's shared acpx engine by default, matching `claude_local`/`codex_local`/`gemini_local`. The issue-thread transcript streams live (assistant text deltas, tool calls with a `pending`->`completed` status lifecycle) instead of the CLI lane's bursty complete-message output. Registered `kimi_local -> "kimi"` in `ACPX_ADAPTER_AGENT_IDS` and resolved the built-in agent command to `kimi acp`; `execute.ts` dispatches to the ACP executor first with an automatic CLI fallback when ACP prerequisites fail (`engine=acp` requires ACP, `engine=cli` pins the headless lane); `index.ts` falls back to the shared acpx session codec; the UI/CLI delegate `acpx.*` events to the shared acpx transcript parser and event formatter. The headless CLI lane (above) remains as the fallback - **Registration** (one entry each, mirroring existing adapters): server adapter registry + `BUILTIN_ADAPTER_TYPES`, `AGENT_ADAPTER_TYPES` (shared), UI adapter registry + display registry (`Kimi Code`, Moon icon) + capabilities defaults, CLI adapter registry, `Dockerfile` (package copy + `npm install --global @moonshot-ai/kimi-code@latest`), `vitest.config.ts` workspace, `scripts/release-package-manifest.json` - **Behavioral sets** mirroring `gemini_local` (Kimi resumes sessions the same way): `GIT_SENSITIVE_LOCAL_ADAPTER_TYPES`, `SESSIONED_LOCAL_ADAPTERS` (heartbeat + recovery), `REMOTE_MANAGED_ADAPTERS`, ssh/sandbox execution-target allow-lists, `ADAPTER_DEFAULT_RULES_BY_TYPE` (`timeoutSec: 0`, `graceSec: 15`), and `LEGACY_SESSIONED_ADAPTER_TYPES` + `ADAPTER_SESSION_MANAGEMENT` in adapter-utils - **UI touch-points**: New Agent default-model branch, AgentConfigForm command map (`kimi_local: "kimi"`) + model defaults + a Kimi-specific thinking-effort option list (`Low`/`High`/`Max`, reflecting Kimi's tiers rather than borrowing Claude's), OnboardingWizard (command map, model default, `kimi login` / `KIMI_MODEL_NAME + KIMI_MODEL_API_KEY` auth hints, manual-debug command line), InviteLanding enabled adapters - **Control-plane skill install** (`cli/src/commands/client/agent.ts`): `paperclipai agent local-cli` seeded the Paperclip control-plane skills into `~/.codex/skills` and `~/.claude/skills` so Codex/Claude agents auto-discover the API reference every run. Kimi had no equivalent target, so `kimi_local` agents began each session without the control-plane skill and rediscovered routes (e.g. the company-scoped `POST /api/companies/{companyId}/issues`) by trial and error. Added `~/.kimi-code/skills` (honoring `KIMI_CODE_HOME`) as a third install target for parity. Independent of the per-run `--skills-dir` delivery, which only applies to explicitly configured skills. - **Docs**: `docs/adapters/kimi-local.md` (prerequisites, auth options, config fields including `effort`, session resume, instruction bundle, skills delivery, control-plane skill install) + a row in `docs/adapters/overview.md` Out of scope (deliberately): model profiles, built-in agent `allowedAdapterTypes` additions. ## Verification\n\nCurrent-master rebase verification (OpenAI Codex, 2026-08-03): 13 focused files / 231 tests pass; adapter-utils, server, UI, CLI, and Kimi adapter typechecks pass; full repository build and UI token gates pass. The branch is conflict-free against master at head `1249df117c5e12e5771b9a570a6340866450619e`.\n\nAutomated (all from repo root, pnpm 9.15.4, Node 22): - `vitest run packages/adapters/kimi-local`: 89/89 pass (includes coverage for the instruction `--add-dir` directive, effort forwarding/gating/mapping, `--skills-dir` gating, the signaled-exit failure path, and `onEvent` runtime-event forwarding with cross-chunk line buffering) - `vitest run server/src/__tests__/adapter-registry.test.ts server/src/__tests__/adapter-routes.test.ts server/src/services/heartbeat-stop-metadata.test.ts ui/src/adapters/adapter-display-registry.test.ts`: 37/37 pass - `vitest run cli/src/__tests__/skills.test.ts`: 13/13 pass (the control-plane skill install target follows the existing Codex/Claude install path, whose symlink logic is unchanged) - `vitest run packages/shared`: 307/307 pass; `vitest run packages/adapter-utils`: pass except one pre-existing, unrelated failure (`mcp-isolation.integration.test.ts` requires Claude CLI ≥ 2.1.207; host has 2.1.185, fails identically on unmodified master) - `pnpm --filter @paperclipai/adapter-kimi-local typecheck|build`, plus typecheck of `server`, `ui`, `cli`, `adapter-utils`: all clean - `pnpm install --frozen-lockfile`: passes (the PR diff itself contains no lockfile changes, per repo policy; verified against a locally regenerated lockfile) - `node scripts/check-no-git-push.mjs` and `node scripts/check-forbidden-tokens.mjs`: pass - CI note: the `policy` job's release-bootstrap step is expected to stay red until a maintainer bootstraps the first npm publish of `@paperclipai/adapter-kimi-local`; see the CI Note for Maintainers comment. All other contributor-actionable checks are green. Manual end-to-end (real Kimi CLI 0.27.0, OAuth login, dev server on an isolated instance): 1. Server `GET /api/adapters` lists `kimi_local` as builtin with correct capability flags; models endpoint returns the three Kimi models 2. `POST .../adapters/kimi_local/test-environment`: all checks pass, including a live `kimi -p` hello probe 3. Created a `kimi_local` agent and invoked two heartbeats: run 1 spawned `kimi -p ... --output-format stream-json`, Kimi used its `Read` tool, produced the expected answer, and the session id was captured from the `session.resume_hint` meta event; run 2 resumed the **same** Kimi session (`sessionIdBefore == sessionIdAfter`) via `-r` 4. UI: adapter appears in the New Agent dropdown; selecting it shows the Kimi command placeholder, the three models, and the Kimi config fields; the run transcript renders Kimi tool calls via the adapter's stdout parser The instruction-bundle, thinking-effort, and `--skills-dir` changes landed after the manual run above. They are covered by the unit tests listed under Automated, and the Kimi CLI flags they rely on (`--add-dir`, `--skills-dir`, `KIMI_MODEL_THINKING_EFFORT`) were confirmed against the installed Kimi Code CLI 0.27.0 (`kimi --help`, config-file thinking-effort docs). Screenshots (assets branch on the fork, not part of the diff):       ## Risks - Low risk to existing behavior: the change is additive, one new workspace package plus single-entry registrations alongside existing adapters; no existing adapter code paths are modified. - The adapter invokes the locally installed `kimi` CLI; like other local adapters, run behavior depends on the host's Kimi version. The parser is written against the documented/verified 0.27.0 stream-json schema and degrades gracefully (malformed lines are skipped, failures surface as run errors). - `--skills-dir` overrides Kimi's auto-discovery of user and project skills for the run. This is intentional (paperclip-managed agents get a reproducible, isolated skill set), and it is only passed when at least one Paperclip skill is desired, so unconfigured agents keep default discovery. - Thinking effort is only forwarded to models that advertise `support_efforts` (currently `kimi-code/k3`); `EFFORT_CAPABLE_MODELS` must be extended when more Kimi models gain support, otherwise a configured effort is silently ignored for them. - `Dockerfile` now installs `@moonshot-ai/kimi-code@latest` globally alongside the other agent CLIs, so image size increases slightly. - Maintainer action needed for the npm bootstrap gate: the `policy` job's release-bootstrap step fails until the first npm publish of `@paperclipai/adapter-kimi-local` (the gate from #5146 that every new adapter package has passed through). Enrollment with `publishFromCi: true` is required by the manifest validator (dropping the entry, `false`, or `private` are all rejected), so this is intentionally left to a maintainer. Remaining CI lanes are expected to run once it is done. ## Model Used\n\n- **Current-master rebase, conflict adaptation, and registry-parity coverage:** OpenAI, **GPT-5 Codex** (Codex agent; exact serving model ID and context-window size were not exposed to the runtime), with repository, shell, Git, and GitHub tooling. It preserved Hawik’s commit authorship, reconciled ACPX and environment-capability changes, added current registry tests, and ran the verification above.\n- **Adapter implementation and initial review:** Moonshot AI, **Kimi K3 Coding** (latest), via **Kimi Code CLI v0.27.0** (`kimi-code/k3` alias, 1M-token context window, thinking mode, agentic tool use). The CLI agent explored the repo, wrote the adapter implementation (delegated to a coder sub-agent of the same model), ran tests, and drafted the first version of this PR body. A second model-driven review pass (read-only, same model) audited the diff for security/correctness before submission; its findings (shell-quoting hardening, auth-detection false positive, session-compaction registration, test gaps) were fixed and are included. - **Harness-context fixes and review responses:** Anthropic, **Claude Opus 4.8** (`claude-opus-4-8`) via Claude Code. Diagnosed from run logs that Kimi received only the entry instructions file (not the `HEARTBEAT.md`/`SOUL.md`/`TOOLS.md` bundle) and that `effort` was never wired, then implemented the instruction `--add-dir` delivery, `KIMI_MODEL_THINKING_EFFORT` forwarding, and `--skills-dir` skill delivery, added the accompanying tests and docs, and addressed the automated review comments (preserving external skills on remote sync, treating a signaled exit as a failure). Also extended the `paperclipai agent local-cli` installer to seed the control-plane skills into `~/.kimi-code/skills` for Codex/Claude parity, wired `onEvent` runtime events so the issue-thread activity indicator reflects Kimi's tool and reasoning output live, and built the ACP engine lane (`kimi acp` via the shared acpx engine, default) so the transcript streams with live tool status like the other ACP adapters. The Kimi CLI flags, subcommand, and env var relied on here were verified against the installed Kimi Code CLI 0.27.0. - All CLI behaviors claimed here (`-p`, `--output-format stream-json`, `-r` resume, event shapes, `--add-dir`, `--skills-dir`, `KIMI_MODEL_THINKING_EFFORT`) were verified empirically against the installed Kimi CLI, not assumed. ## 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 *(only the release-bootstrap step remains red, pending the maintainer npm publish described in Risks)* - [x] Greptile is 5/5 with no open P2s, recommendations, or follow-ups *(will address all Greptile comments as they arrive)* - [x] I will address all Greptile and reviewer comments before requesting merge --- ## Maintainer Addendum (2026-08-20) The shared acpx-engine and issue-chat changes (run-summary segmentation, placeholder tool-event coalescing, `ISSUE_CHAT_TRANSCRIPT_MAX_VISIBLE_ENTRIES` 30 → 400, live-reasoning UI) have been **extracted to #11761** so the cross-adapter behavior changes review and revert independently — both commits there preserve @hawikk's authorship. This PR is now the kimi-specific adapter only (60 files, +3,793/−8, essentially pure addition); the only shared-engine touch left is the `kimi acp` command resolution. `publishFromCi` is `true` — the package name is bootstrapped on npm. --------- Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com> Co-authored-by: Paperclip <noreply@paperclip.ing> Co-authored-by: Dotta <bippadotta@protonmail.com> Co-authored-by: Devin Foley <devin@paperclip.ing> |
||
|
Nicky LeachandPaperclip
|
6b7e0814a0 |
feat(acp): stream Daytona sandbox agent output and remove the host output poll (#11049)
## Thinking Path > - Paperclip is the open source app that manages AI agents for work > - Sandbox providers let agents run in remote and isolated environments > - Daytona session commands need a path that sends agent output to the host without host polling > - Host polling adds delay and repeats provider output work > - This pull request adds typed execute.log notifications and a log sink for incremental output > - This pull request adds an optional ACP session stream with final-result replay protection > - The benefit is lower output delay while the default flags keep current behavior unchanged ## Linked Issues or Issue Description **Subsystem affected** Cross-cutting. The change spans the plugin SDK, Daytona provider, adapter utilities, and server execution services. **Problem or motivation** The Daytona ACP bridge polls a host output file while an agent command runs. This adds delay and can repeat work. The host also needs a safe route for provider output chunks. **Proposed solution** Add a typed `execute.log` notification with host-issued invocation correlation. Add an ordered log sink to the environment execute path. Add an optional ACP session-log path that parses newline-delimited JSON frames and removes the host output poll for that path. **Alternatives considered** Keep the output-file poll as the only path. This keeps the current behavior but does not provide timely output. The new path stays behind flags, so the existing path remains the default fallback. **Roadmap alignment** This change supports the shipped Cloud / Sandbox agents milestone in `ROADMAP.md`, including Daytona support. ## What Changed - Add the typed `execute.log` worker-to-host notification and company-scoped host route. - Add ordered `stdout` and `stderr` chunk delivery before the final execute result. - Add the Daytona session log sink and the optional ACP streamed session path. - Add monotonic frame handling so live and final output reach the host once. - Keep `useLogStream` and `streamAgentSessionOutput` off by default. - Add unit and integration coverage for the notification, execution target, runtime, and Daytona paths. ## Verification - Run adapter-utils tests: 445 tests pass locally. - Run server environment tests: 73 tests pass locally. - Run Daytona plugin tests: 131 tests pass locally. - Run TypeScript checks for shared, adapter-utils, and server. - Review the pull request checks after GitHub completes them. - All required GitHub checks pass on the current head. ## Risks The new paths change output delivery only when a feature flag enables them. The final execute result remains available for parsing and fallback. The main risk is a provider stream or frame-order error; the final-result parser limits that risk. ## Model Used OpenAI Codex, GPT-5, tool use and code execution. The runtime did not supply a context-window value. ## 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 No operator documentation change applies because both new flags remain disabled by default. - [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> |
||
|
Nicky LeachandPaperclip
|
c647b8cc2e |
feat(acpx-engine): give sandbox.exec spans real parents (#10852)
## Thinking Path > - Paperclip uses spans and traces to show how work moves through agents and tools > - sandbox.exec spans need a real parent so the trace tree matches the work tree > - Wrong parent links make execution history hard to read and hard to debug > - This pull request adds a single task.run root span and re-parents live work to the nearest active span > - The change keeps detached work under the closest live span instead of the HTTP root > - The benefit is a clear trace tree for sandbox.exec work and better execution diagnosis ## Linked Issues or Issue Description **What happened?** sandbox.exec spans attached to the wrong parent or to no live parent in some paths. **Expected behavior** Each sandbox.exec span should attach to the nearest live span. **Steps to reproduce** 1. Run work that creates sandbox.exec spans during startup and callback bridge paths. 2. Inspect the trace tree. 3. Observe an orphaned span or a span with the wrong parent. **Paperclip version or commit** `672e9de9c8b004aebc1f08e24b612ab067735ad1` **Deployment mode** Local dev. **Additional context** The branch adds the task.run root span, parents sandbox.startup to it, and re-parents detached bridge work to the nearest live span. ## What Changed - Added a task.run root span for the run tree. - Re-parented sandbox.startup, agent.turn, and detached bridge work to the nearest live span. - Added end-to-end trace-tree assertions for the full parent chain. - Added negative coverage so sandbox.exec does not parent to the HTTP root. ## Verification - Focused Vitest suite passed: `packages/adapter-utils/src/acpx-engine/execute.test.ts`, `packages/adapter-utils/src/acpx-engine/startup-timing.test.ts`, `packages/adapter-utils/src/execution-target-sandbox.test.ts`, `packages/adapter-utils/src/sandbox-callback-bridge.test.ts`, and `server/src/__tests__/environment-execution-target.test.ts`. - Result: 5 files passed, 204 tests passed. - The submitted branch also reported `adapter-utils` checks, `server` seam checks, and `tsc` exit 0 in the handoff state. ## Risks - This change can alter trace tree shape in tools that read parent spans. - A missed bridge path could still point to the wrong live span. - Low risk for runtime behavior, because the change only changes span parent attribution. ## Model Used OpenAI GPT-5, tool-use capable. ## 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 - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [ ] I will address all Greptile and reviewer comments before requesting merge --------- Co-authored-by: Paperclip <noreply@paperclip.ing> |
||
|
Nicky LeachandPaperclip
|
2ab797dcbe |
fix(sandbox): reset step store for long-lived bridge work (#10813)
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work
> - Bridge workers keep startup setup separate from long-lived runtime
work
> - The callback bridge must keep queue-directory setup inside the
startup step
> - The long-lived poll loop must run with no active startup step
> - This pull request keeps that boundary in the right place
> - The benefit is correct span parents and correct runtime exec
metadata
## Linked Issues or Issue Description
**Bug**
**What happened?**
Long-lived bridge continuations kept a stale startup step store during
the queue-directory setup path.
**Expected behavior**
Runtime exec spans should start with no active startup step.
**Steps to reproduce**
1. Start a bridge lane.
2. Let the startup step end.
3. Run later runtime exec work on the same lane.
**Paperclip version or commit**
|
||
|
Nicky LeachandPaperclip
|
6a69a3d6b0 |
refactor(observability): stop writing detailed per-step timing to the run log (#10776)
## Thinking Path > - Paperclip tracks agent work for the team. > - The startup timing path already emits detailed spans. > - The run log repeats the same detailed timing. > - That makes the event larger than it needs to be. > - This pull request removes the redundant per-step timing fields from the run-log payload. > - The benefit is a smaller log with the same detail still available in traces. ## Linked Issues or Issue Description No public GitHub issue or open pull request matches this change. I described the enhancement below. **What existing behavior does this improve?** `run.startup.step` event output. **Subsystem affected** Cross-cutting (multiple of the above) **Current behavior** `run.startup.step` writes `roundTrips`, `providerExecMs`, `providerGetMs`, `createRuntimeMs`, and `ensureSessionMs`. The same detail already exists in the spans. **Proposed behavior** `run.startup.step` keeps only `step`, `durationMs`, and `outcome`. The heartbeat lifecycle timestamps stay unchanged. **Reason and benefit** This change removes redundant data from the run log. It keeps the useful detail in trace spans. It also makes the payload smaller and easier to read. The revert path stays clear because the removed data has one producer chain. **Breaking changes** Yes. Consumers that read the removed fields must switch to span data or the remaining event fields. The heartbeat lifecycle timestamps do not change. **Additional context** I searched GitHub for related open PRs and issues. I found no match. ## What Changed - Removed the redundant per-step timing fields from the run-log payload. - Removed the now-dead producer chain that fed those fields. - Kept the span-level timing data and the heartbeat lifecycle timestamps. ## Verification - `adapter-utils` typecheck clean - `server` typecheck clean - `startup-timing` suite: 30 passed - `adapter-utils` `execute` suite: 99 passed - `server` `environment-execution-target` suite: 21 passed ## Risks - Consumers that still read the removed fields will need a code change. - This is a one-way-door data removal for the run log. ## Model Used OpenAI Codex, GPT-5, tool use enabled. ## 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 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> |
||
|
Nicky LeachandPaperclip
|
af6b32d82a |
feat(observability): add provider OTel spans - cache-hit flag, plugin tracer, Daytona pack/transfer spans (#10764)
## Thinking Path > - Paperclip uses pull requests to ship control-plane changes with review and audit. > - This change extends sandbox provider telemetry. > - The first PR added startup and exec spans. > - This PR adds provider spans for cache hits, plugin tracer flow, and Daytona file sync steps. > - The host must keep the trust boundary and reject unsafe span data. > - The result is fuller OTel coverage with bounded labels and safe parentage. ## Linked Issues or Issue Description **Problem or motivation** Sandbox provider telemetry lacks an explicit cache-hit signal, plugin span context, and file-sync pack and transfer spans. The current proxy for a cache hit is fragile. The plugin SDK also needs a safe tracer surface and a per-call parent context. **Proposed solution** Add an explicit `cacheHit` flag from the Daytona sandbox handle lookup. Expose `ctx.tracer` on the plugin context and pass a `traceparent` string with each call. Record provider spans on the host with allowlist clamping and capability checks. Wrap Daytona file sync pack and transfer steps in spans. **Alternatives considered** Keep the old `providerGetMs == 0` proxy. Reject that path because the cache-hit decision belongs at the handle lookup, not in a timing proxy. The host stays the trust boundary for worker span data. It clamps labels, rejects bad parent data, and gates the worker-to-host span RPC by capability. A security review completed before this PR opened. ## What Changed - Added an explicit `cacheHit` flag from the Daytona sandbox handle lookup. - Added `ctx.tracer` on the plugin context and a `traceparent` field in the per-call context. - Added host-side span recording with attribute clamping and capability checks. - Wrapped Daytona file sync pack and transfer steps in spans. - Added tests for the host trust boundary, the plugin tracer no-op path, and the Daytona span paths. ## Verification - `pnpm --filter @paperclipai/plugin-daytona exec vitest run` - `pnpm --filter @paperclipai/plugin-sdk exec vitest run` - `pnpm --filter @paperclipai/server exec vitest run plugin-host-services environment-execution-target instrumentation plugin-worker-manager` - `tsc --noEmit` for server, plugin-sdk, and adapter-utils ## Risks - Span data now crosses the worker boundary, so the host allowlist and capability gate must stay strict. - The `traceparent` path must stay valid and host-owned. - Daytona file sync spans must not add new round trips. ## Model Used OpenAI Codex, GPT-5, tool use. ## 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 linked existing issues with `Fixes:` / `Closes:` / `Refs:` or described the issue in the PR body - [x] I have not referenced internal or instance-local Paperclip issues or links - [x] My branch name describes the change 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> |
||
|
Nicky LeachandPaperclip
|
c09ea7112b |
feat(observability): add granular OpenTelemetry spans for sandbox startup and execution (#10758)
## Thinking Path > - Paperclip coordinates AI agent work and records execution data. > - The sandbox startup path and the host-to-sandbox execution path need clearer OpenTelemetry spans. > - The trace data now shows wait time, critical path data, and bounded labels. > - This pull request adds bounded spans and closed attribute helpers for sandbox startup and execution. > - The result is better trace data with low cardinality and no secret leakage. ## Linked Issues or Issue Description - No public GitHub issue exists for this branch. - Related PR: #10536 - This PR extends the sandbox OpenTelemetry work with exec spans, root timing, and bounded labels. ## What Changed - Added a closed span-attribute contract for sandbox startup data. - Added a bounded label helper for command and region values. - Threaded the active step context into the execution path. - Added a `sandbox.exec` span with true timestamps and bounded attributes. - Added skipped-step and root-span timing data with low-cardinality context. - Moved handshake sub-times and bridge batch data onto spans. ## Verification - `pnpm --filter @paperclipai/adapter-utils run typecheck` - `pnpm --filter @paperclipai/server run typecheck` - `pnpm --filter @paperclipai/adapter-utils exec vitest run` - `pnpm --filter @paperclipai/server exec vitest run environment-execution-target` - `git log --oneline origin/master..origin/feat/sandbox-startup-otel-spans` - `git diff origin/master...origin/feat/sandbox-startup-otel-spans --stat` ## Risks - Span attribute rules could still miss a future field if new code skips the shared helper. - Low-cardinality labels may hide some detail, but that is the intended tradeoff. - Tracing stays fail-open, so a missing tracer still hides data rather than stopping work. ## Model Used - OpenAI Codex, GPT-5, tool use enabled. The runtime did not expose the context window size. ## 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 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] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [x] I will address all Greptile and reviewer comments before requesting merge The current head still has one failing required check: `e2e`. The PR stays unready for board handoff until that check passes. --------- Co-authored-by: Paperclip <noreply@paperclip.ing> |
||
|
Nicky LeachandPaperclip
|
9f7565f4ce |
feat: activate OpenTelemetry spans on the sandbox start path (#10536)
## Thinking Path > - Paperclip moves agent work through sandboxed execution and control-plane services. > - The sandbox start path now has a no-op span seam. > - This change turns that seam on when OTLP export is configured. > - It keeps the default path unchanged when export is off. > - The result is structured startup traces with low-cardinality attributes and explicit parent links. > - The benefit is better observability without changing normal behavior. ## Linked Issues or Issue Description No public GitHub issue exists for this change. ### Problem The sandbox start path has a tracer seam, but it stays a no-op unless the OTLP export path is active. ### Proposed solution Enable the server tracer on sandbox bring-up, open a root span, parent each startup boundary to that root, and keep the export path opt-in behind `OTEL_EXPORTER_OTLP_ENDPOINT`. ### Alternatives considered - Keep the start path as a no-op. I rejected that path because it leaves sandbox start opaque when OTLP export is already configured. - Add broad attributes for commands and paths. I rejected that path because the span allowlist must stay low-cardinality. ### Roadmap alignment This follows the current OTel sandbox-start work and keeps the default path unchanged. ## What Changed - Add a root sandbox startup span and child spans for each named startup boundary. - Keep concurrent bridge spans parented to the root span. - Inject the server tracer through the adapter deps without OpenTelemetry imports in the engine. - Attach host-received provider duration attributes only when the values are finite. - Keep span attributes inside the allowlist and keep command, path, id, and error text out of span data. ## Verification - The pushed branch already passed `pnpm --filter @paperclipai/adapter-utils exec tsc --noEmit`. - The pushed branch already passed `pnpm exec vitest run packages/adapter-utils/src/acpx-engine/`. - The pushed branch already passed `pnpm --filter @paperclipai/server exec vitest run src/__tests__/environment-execution-target.test.ts src/__tests__/instrumentation.test.ts`. - The pushed branch already passed `pnpm --filter @paperclipai/server exec tsc --noEmit`. ## Risks - OTel export changes trace volume when the endpoint is set. - The allowlist limits trace detail, so new fields need care. - The change stays no-op when OTLP export is off. ## Model Used - OpenAI GPT-5, tool use enabled. ## Checklist - [x] I have included a thinking path that traces from project context to this change - [x] I have specified the model used with 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 linked existing issues or described the issue in-PR - [x] I have not referenced internal/instance-local Paperclip issues or links - [x] My branch name describes the change and contains no internal 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> |
||
|
Devin FoleyandPaperclip
|
7a5a217d60 |
fix(server): gate sandbox/ssh execution targets by shared remote-managed adapter capability (#10459)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work. > - Paperclip can run an agent in a remote-managed environment, such as a sandbox provider or an SSH host. > - The server resolves an execution target for each run. The resolver kept its own hardcoded list of allowed adapters. > - The shared capability metadata in `packages/shared/src/environment-support.ts` already defines which adapters support remote-managed environments. The environment selector and the capabilities API use it. > - The two lists drifted. The UI offered sandbox environments to Grok Build (`grok_local`) agents, but the resolver refused them at run time. > - This pull request makes the resolver use the shared capability check for both the sandbox gate and the SSH gate. > - The benefit is one source of truth. The UI and the runtime now agree on which adapters can use remote-managed environments. ## Linked Issues or Issue Description No public GitHub issue exists for this bug. Inline description per the bug report template: **What happened?** A Grok Build (`grok_local`) agent was assigned a sandbox environment (a Daytona provider). The UI allowed the assignment. Every run and primary-model test then failed with the warning: `Adapter "grok_local" is not allowed in "<environment>" environments.` **Expected behavior** An adapter that the environment selector offers for a sandbox environment must also pass the runtime gate. The Grok Build run must start in the sandbox. **Steps to reproduce** 1. Create a sandbox environment (for example, with a Daytona provider plugin). 2. Create an agent that uses the `grok_local` adapter. 3. Set the agent's environment to the sandbox environment. The UI accepts this. 4. Run the agent, or run the primary-model test. The run fails with the adapter-not-allowed warning. **Paperclip version or commit** Reproduced on `master` at `0edb742f8d`. **Deployment mode** Local instance with a remote sandbox provider plugin. The same gate also applies to SSH environments. ## What Changed - `resolveEnvironmentExecutionTarget` in `server/src/services/environment-execution-target.ts` now gates the sandbox path with the shared `adapterSupportsRemoteManagedEnvironments()` helper. Before, it used a hardcoded six-adapter list that did not include `grok_local`. - The SSH path in the same file now uses the same shared helper. - New regression tests in `server/src/__tests__/environment-execution-target.test.ts`: sandbox target resolution for every remote-managed adapter (including `grok_local`), SSH target resolution for `grok_local`, and the null path for an adapter without remote-managed support. ## Verification - Run `node_modules/.bin/vitest run server/src/__tests__/environment-execution-target.test.ts`. All 10 tests pass, including the 3 new ones. - Confirm `grok_local` is in the `REMOTE_MANAGED_ADAPTERS` set in `packages/shared/src/environment-support.ts`. The resolver now reads the same set. - On a live local instance with this fix, a `grok_local` agent assigned to a Daytona sandbox environment no longer produces the adapter-not-allowed warning. ## Risks Low risk. The change routes two hardcoded checks through existing shared capability metadata. Behavior changes only where the lists had drifted: `grok_local`, and any future adapter added to the shared set, can now resolve sandbox and SSH execution targets. Adapters outside the shared set still return `null`. ## Model Used Claude Fable 5 (`claude-fable-5`) by Anthropic, with extended thinking and tool use, running in Claude Code. ## 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 - [ ] 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: Paperclip <noreply@paperclip.ing> |
||
|
Nicky LeachandPaperclip
|
030dd9d15c |
feat(adapter-utils): provider-delegable syncIn seam with ordered post-upload commands (#10340)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - The adapter/runtime layer has to move files into sandboxes safely and efficiently > - The current sync-in path needs a provider-delegable seam so providers can use their native upload transport when available > - The upload contract also needs ordered post-upload commands so extracted content can be finalized fail-fast after transfer > - The fallback path still has to preserve current behavior when the provider does not expose native sync verbs > - This pull request adds the contract and runtime seam for provider-delegable sync-in, plus the single-stream collapse flag > - The benefit is fewer round trips, a cleaner provider-owned upload path, and a compatible fallback for existing runners ## Linked Issues or Issue Description No public GitHub issue exists for this change. The underlying feature request is described below in the repository's feature-request format. ### Problem or motivation Paperclip needs a sync-in path that lets each provider choose the best available upload transport instead of forcing the harness to orchestrate uploads the same way every time. The runtime also needs a way to describe ordered post-upload commands so providers can finalize extracted content fail-fast after transfer. ### Proposed solution Extend the sync contract with ordered post-upload commands, forward that contract through the plugin and environment runtime layers, and make client syncIn always available. When a provider advertises native sync verbs, the client should delegate to that transport; otherwise it should fall back to the existing tarball/write/extract behavior and then run the post-upload commands in order. ### Alternatives considered Keeping upload orchestration entirely host-side would avoid a contract change, but it would block provider-specific transport optimizations and keep the harness responsible for a path the provider can do more efficiently. A separate post-upload API would add another surface without improving the existing sync flow. ### Roadmap alignment This work aligns with the broader runtime and adapter roadmap because it improves provider integration without changing the external product model. It is an additive contract change that preserves backward compatibility for providers that do not expose native sync verbs. ### Additional context The fallback path still needs to preserve existing observable behavior, including command ordering, cwd confinement, and fail-fast execution. The single-stream progress flag is part of the same transport improvement so smaller writes can collapse to a single round trip when the runner supports it. ## What Changed - Added ordered `postUploadCommands` support to the sync operation contract and SDK mirror. - Plumbed the sync-in contract through the plugin and environment runtime layers. - Implemented a runtime client `syncIn` path that delegates to native provider transport when available, otherwise uses the generic tarball/write/extract fallback. - Preserved fail-fast execution of ordered post-upload commands in the fallback path. - Flipped the sandbox runner's single-stream stdin progress flag to collapse small `writeFile` operations to a single round trip. - Added and updated tests for contract forwarding, fallback behavior, cwd rejection, fail-fast behavior, and single-stream collapse. ## Verification - `pnpm --filter @paperclipai/plugin-sdk exec vitest run protocol.postupload.test.ts` - `pnpm --filter @paperclipai/plugin-sdk exec vitest run environment-sync-negotiation.test.ts` - `pnpm --filter @paperclipai/adapter-utils exec vitest run command-managed-runtime.test.ts` - `pnpm --filter @paperclipai/server exec vitest run environment-execution-target.test.ts` - Local typecheck and targeted suite runs reported in the handoff passed before PR creation. ## Risks - The new fallback path could diverge from the previous inline upload behavior if the tarball/extract contract changes. - Provider-native sync handling may expose provider-specific edge cases if a runner advertises sync verbs but does not fully honor the contract. - The single-stream flag changes transport behavior for small uploads, so regressions would likely show up as round-trip or upload failures. ## Model Used OpenAI Codex (GPT-5, tool-using coding agent). ## 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 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> |
||
|
|
197718bc00 |
feat(acpx): per-step round-trip + provider-latency attribution for sandbox startup (#10222)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Sandbox-backed runs need more precise startup observability so operators can see where time is spent before an adapter is invoked > - Aggregate startup timing hides which boundary is actually slow, especially in remote execution where the bottleneck can move between host round-trips, provider-boundary latency, and handshake phases > - This pull request keeps the existing startup timing channel additive while attributing the latency to the specific startup steps that caused it > - The benefit is better diagnosis of sandbox startup regressions without changing control flow or introducing a schema migration ## Linked Issues or Issue Description Refs: #10204 This PR extends the existing sandbox run-startup timing observability with per-step round-trip and provider-latency attribution for the Daytona startup path. It keeps the event payload additive and free-form, and it leaves the control flow, database schema, and external adapter interfaces unchanged. ## What Changed - Added per-step round-trip counting for the host-to-sandbox execute seam - Added provider-boundary duration accumulation for the Daytona execute and re-fetch steps - Split the ACP handshake timing into `createRuntimeMs` and `ensureSessionMs` while preserving the warm-handle skip - Kept the startup timing payload additive and did not add a schema migration ## Verification - `adapter-utils` acpx-engine and startup-timing suites: pass - Daytona plugin suite: pass with mocked SDK and injected-clock duration assertions - `server` environment-execution-target suite: pass - `tsc --noEmit` for adapter-utils and server: pass - Git validation: fetched `origin/feat/sandbox-start-step-timing-attribution`, confirmed it matches the authorized submit SHA `53c573266e618af05c57a0720aaa9d9e0452de61`, and confirmed the branch contains only the expected commit on top of `origin/master` - Searched GitHub for duplicate or related PRs/issues; found one closely related merged PR and no open duplicate on this branch - Checked `ROADMAP.md`; the broad sandbox-agent roadmap section does not call out this specific startup-timing attribution work as a duplicate ## Risks - Low risk: the change is additive and only enriches existing timing data - Downstream consumers that assume aggregate-only startup timing may need to tolerate the additional per-step fields - The finer spawn/initialize/session split remains a follow-up in the external ACP client because that hook is not available here yet ## Model Used OpenAI GPT-5, tool-using coding agent ## 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 - [ ] All Paperclip CI gates are green - [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups - [ ] I will address all Greptile and reviewer comments before requesting merge --------- Co-authored-by: Harold Kim <harold@paperclip.ing> Co-authored-by: Paperclip <noreply@paperclip.ing> |
||
|
|
9cde4e128c |
feat: run ACP sessions in sandbox execution targets (#9390)
## Thinking Path > - Paperclip is the open source app people use to manage AI agents for work > - Agent adapters (Claude, Codex, Gemini) default to the ACP engine lane, which needs a live bidirectional stdio session with the agent process > - Sandbox execution targets only exposed one-shot command execution, so every ACP-capable adapter refused remote targets and fell back to the CLI lane with a "supports only the local Paperclip host" warning > - Running agents in sandboxes is a core deployment mode, and losing ACP there means losing streaming updates, structured events, and default-lane parity with local runs > - This pull request adds a provider-agnostic process-session bridge that relays the ACP stdio session into the sandbox over the existing sandbox runner contract, and updates the adapters to use it > - The benefit is that the default ACP lane now behaves the same on the local host and in any sandbox provider, with CLI fallback reserved for targets that genuinely cannot host a bidirectional session ## Linked Issues or Issue Description No existing public issue covers this; inline description following the feature request template: **Problem or motivation** Configuring an ACP-capable adapter (e.g. Claude) with a sandbox environment made every run fall back to the CLI lane with the warning "Claude ACP currently supports only the local Paperclip host, but this run targets a remote environment." The ACP engine only knew how to spawn a local subprocess, while sandbox providers only expose one-shot command execution — so there was no way to hold the bidirectional stdio session ACP requires. **Proposed solution** Add a process-session bridge in `adapter-utils`: a local ACPX-spawnable proxy script connects to a token-authenticated loopback TCP server, which relays JSON-framed stdin/stdout/stderr events to and from a small relay script executed inside the sandbox via the provider's ordinary runner. Claude/Codex/Gemini adapters now treat sandbox targets with a runner as ACP-capable, resolve agent commands against the remote target, and fall back to CLI only when the sandbox exposes no bidirectional path. The sandbox callback bridge injects a run-scoped API endpoint and bridge token so the agent inside the sandbox can reach Paperclip (including work-product handoffs) without ever receiving the host run JWT. **Alternatives considered** A provider-specific lane was prototyped first: Daytona minting SSH access metadata at lease time, converted into an SSH execution target. It was dropped because it only worked for providers able to advertise SSH, added per-provider surface area, and left every other sandbox provider on the CLI fallback. The merged design rides the one-shot runner contract all providers already implement; a regression test pins that sandbox targets stay on the bridge lane even when lease metadata advertises SSH access. **Roadmap alignment** Directly advances the "Cloud / Sandbox agents" roadmap item — agents running in remote and sandboxed environments keep the same control-plane behavior as local ones. No overlap with other planned core work. ## What Changed - `packages/adapter-utils/src/execution-target.ts`: new `startAdapterExecutionTargetProcessSessionBridge()` plus helpers — writes a token-authenticated local proxy script (spawnable by ACPX) and a remote relay script synced into the sandbox, with a loopback TCP server streaming JSON-framed stdio between them; events emitted before the ACP client attaches are buffered so none are lost. - `packages/adapter-utils/src/acpx-engine/execute.ts`: the ACP engine can execute against remote sandbox targets through the bridge instead of requiring a local subprocess, including remote cwd/env shaping. - `packages/adapter-utils/src/sandbox-callback-bridge.ts`: sandbox-scoped API bridging extended to allow work-product handoffs; the sandbox payload env carries a bridge token, never the host run JWT. - `packages/adapters/claude-local`, `codex-local`, `gemini-local` (`src/server/acp.ts`): default-lane selection no longer rejects all remote targets; command resolution is remote-aware (`ensureAdapterExecutionTargetCommandResolvable`, `resolveAdapterExecutionTargetCwd`); the fallback reason is now scoped to sandboxes that expose only one-shot execution. - `server/src/__tests__/environment-execution-target.test.ts`: pins that sandbox targets resolve to the bridge lane, including when lease metadata advertises SSH access. - Non-sandbox remote targets (e.g. SSH) keep the CLI lane: the ACP engine's remote transport is sandbox-only, so default-lane selection falls back for those targets across all three adapters, and tests covering CLI-specific remote behavior pin `engine: "cli"` explicitly. - The bridge authenticates loopback connections before they can own the session or receive buffered output (token required, idle unauthenticated peers dropped), and remote event writes are serialized so the exit event always lands after stdout/stderr have drained. - Daytona plugin: formatting-only residue from the earlier iteration; no functional change. ## Verification - `vitest run` over the touched suites — `packages/adapter-utils/src/acpx-engine/execute.test.ts`, `packages/adapter-utils/src/execution-target-sandbox.test.ts`, `packages/adapter-utils/src/sandbox-callback-bridge.test.ts`, the three adapter `acp.test.ts` files, and `server/src/__tests__/environment-execution-target.test.ts` — 102 tests pass. - End to end: with a Claude agent configured on a Daytona sandbox environment, the primary-model test now selects the default ACP lane (no fallback warning), and the full round trip (wake → sandbox execution → API bridge → comment post) was exercised twice from inside a live sandbox. ## Risks - Behavioral shift: adapters that previously always fell back to CLI on sandbox targets now default to ACP there; `engine=cli` still pins the CLI lane explicitly. - The bridge relays stdio as JSON lines over loopback TCP guarded by a per-session random token; the remote relay runs inside the sandbox under the provider's runner. Providers with slow one-shot execution will see higher session startup latency — the CLI fallback remains for genuinely incapable targets. - No schema or migration changes. ## Model Used - Claude Fable 5 (`claude-fable-5`, Anthropic) — extended thinking enabled, agentic tool use via the Claude Agent SDK harness; implementation iterated with local Vitest verification. ## 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 (no shipped docs describe the old local-only ACP limitation) - [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: Cody <noreply@paperclip.ing> Co-authored-by: Cody <cody@paperclip.local> |
||
|
Devin Foley
|
076067865f |
Migrate SSH environment callback to bridge (#5116)
> **Stacked PR (part 3 of 7).** Depends on: - PR #5114 - PR #5115 > Diff against `master` includes commits from earlier PRs in the stack — the new commit in this PR is the topmost one. ## Thinking Path > - Paperclip orchestrates AI agents for zero-human companies > - Agents executing on a remote SSH-backed environment need a way to call back into > the Paperclip control plane (run events, log streaming, signals) > - When the SSH host can't reach the Paperclip host (NAT, firewalls, or simply not > on the same network), the run silently fails or hangs — a recurring class of > failure during SSH testing > - In sandboxed environments we already solved this with a callback bridge that > tunnels back through the existing connection; SSH was the odd one out > - This PR migrates SSH execution to use the same callback bridge, so every > adapter's remote run uses one consistent reverse-channel. Per-adapter SSH glue > is deleted in favour of a shared `CommandManagedRuntimeRunner` built from the > SSH spec > - The benefit is fewer SSH-specific failure modes, a smaller code surface, and > one place to evolve the callback contract going forward ## What Changed - Added `createSshCommandManagedRuntimeRunner` in `packages/adapter-utils/src/ssh.ts` that adapts an SSH spec into a generic command-managed-runtime runner (with cwd, env, and timeout handling) - Removed `paperclipApiUrl` from `SshRemoteExecutionSpec`; the bridge URL now flows through the shared runner - Reworked `execution-target.ts` to use the SSH runner alongside sandbox runners via a unified `CommandManagedRuntimeRunner` interface - Simplified `remote-managed-runtime.ts` and `sandbox-managed-runtime.ts` to consume the shared runner abstraction - Deleted per-adapter SSH callback wiring from claude-local, codex-local, cursor-local, gemini-local, opencode-local, pi-local execute.ts files - Removed `environment-runtime-driver-contract.test.ts` (the contract is now enforced by `environment-execution-target.test.ts`) - Added/updated `execute.remote.test.ts` cases for each adapter to cover the SSH runner path ## Verification - `pnpm --filter @paperclipai/adapter-utils test` - `pnpm test -- execute.remote` (covers all six local adapters' SSH paths) - Manual QA: ran a claude-local agent against an SSH-backed environment, confirmed the agent successfully called back to `/api/agent-callback/*` endpoints during the run ## Risks - Refactor touches all six local adapters. If any adapter had subtle SSH-specific behaviour that wasn't captured in tests, it could regress. Mitigation: each adapter's `execute.remote.test.ts` was extended. - `paperclipApiUrl` removal from `SshRemoteExecutionSpec` is a breaking type change for any internal consumer. Verified no external plugins consume this type. - The new `CommandManagedRuntimeRunner` shape is a public surface in `@paperclipai/adapter-utils`; downstream plugins implementing custom runners may need updates, but no such plugins exist in this repo. ## Model Used - OpenAI GPT-5.4 (reasoning effort: high) via Codex CLI - Provider: OpenAI - Used to author the code changes in this PR ## 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 run tests locally and they pass - [x] I have added or updated tests where applicable - [ ] If this change affects the UI, I have included before/after screenshots — N/A - [ ] I have updated relevant documentation to reflect my changes — N/A - [x] I have considered and documented any risks above - [x] I will address all Greptile and reviewer comments before requesting merge |
||
|
Devin Foley
|
a7b45938b7 |
Let sandbox providers declare shell defaults (#5114)
## Thinking Path
> - Paperclip orchestrates AI agents for zero-human companies
> - Agents execute in sandboxed remote environments served by pluggable
sandbox
> providers (E2B today, more later)
> - Today every sandbox command runs under `sh -lc` regardless of what
the
> provider's container actually ships
> - That misses bash-only shell init on E2B (which ships bash) and
prevents
> future providers from declaring a different default — there's no way
for a
> provider to say "I have bash, use it"
> - This PR adds a `shellCommand` field to sandbox execution targets so
providers
> can declare their preferred shell ("bash" for E2B), threads it through
the
> sandbox-managed-runtime client, callback bridge, and execution-target
shell
> helper, and validates the value at the lease-metadata boundary
> - The benefit is that sandbox commands run under the right shell on
the right
> provider, and adding new sandbox providers only needs to declare a
shell
> preference
## What Changed
- Added `packages/adapter-utils/src/sandbox-shell.ts` exporting
`preferredShellForSandbox(shellCommand)` (returns `"bash"` if input is
`"bash"`,
else `"sh"`)
- Added `shellCommand?: "bash" | "sh" | null` to
`AdapterSandboxExecutionTarget`
and `CommandManagedRuntimeSpec`; threaded it through
`runAdapterExecutionTargetShellCommand`,
`prepareAdapterExecutionTargetRuntime`,
and `startAdapterExecutionTargetPaperclipBridge`
- `createCommandManagedRuntimeClient`, `prepareCommandManagedRuntime`,
and
`createCommandManagedSandboxCallbackBridgeQueueClient` now take an
optional
`shellCommand` and use `preferredShellForSandbox` to pick the shell
- `startSandboxCallbackBridgeServer` accepts a `shellCommand` for its
server
startup, readiness probe, and stop hook
- E2B sandbox plugin declares `shellCommand: "bash"` in `leaseMetadata`
- `resolveEnvironmentExecutionTarget` reads `shellCommand` from lease
metadata
(validating against `"bash" | "sh" | null`)
- `environment-runtime.ts` adds `"shellCommand"` to
`INTERNAL_PLUGIN_SANDBOX_CONFIG_KEYS`
so the field round-trips through internal plugin config without leaking
to
external plugin metadata
- Updated tests in `command-managed-runtime.test.ts`,
`execution-target-sandbox.test.ts`, `sandbox-callback-bridge.test.ts`,
`environment-execution-target.test.ts`
## Verification
- `pnpm --filter @paperclipai/adapter-utils test`
- `pnpm --filter @paperclipai/server test --
environment-execution-target`
- `pnpm --filter @paperclipai/sandbox-providers-e2b test`
- Manual QA: boot a Paperclip instance, create an E2B-backed
environment, run a
claude_local agent against it, and confirm the run completes (verifies
bash
shell semantics flow through the callback bridge end-to-end)
## Risks
- E2B sandbox commands now run under `bash -lc` instead of `sh -lc`.
Bash is a
strict superset for the commands we issue (no busybox-only flags in our
shell
scripts), so risk is low. The shellCommand field is opt-in via lease
metadata —
providers that don't declare it stay on `sh`.
- New optional field on `CommandManagedRuntimeSpec` and
`AdapterSandboxExecutionTarget`.
Consumers ignoring the field retain previous behaviour (sh).
- Lease metadata now carries an additional field. Existing leases
without
`shellCommand` resolve to `null` and fall back to sh — backwards
compatible.
## Model Used
- OpenAI GPT-5.4 (reasoning effort: high) via Codex CLI
- Provider: OpenAI
- Used to author the code changes in this PR
## 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 run tests locally and they pass
- [x] I have added or updated tests where applicable
- [ ] If this change affects the UI, I have included before/after
screenshots — N/A (no UI changes)
- [ ] I have updated relevant documentation to reflect my changes — N/A
- [x] I have considered and documented any risks above
- [x] I will address all Greptile and reviewer comments before
requesting merge
|
||
|
Devin Foley
|
a4ac6ff133 |
Add sandbox callback bridge for remote environment API access (#4801)
## Thinking Path > - Paperclip orchestrates AI agents for zero-human companies > - Agents can run inside sandboxed environments like E2B, which are isolated from the host network > - Sandboxed agents need to call back to the Paperclip API to report progress, post comments, and update issue status > - But sandbox environments cannot reach the Paperclip server directly because they run in isolated network namespaces > - This PR adds a callback bridge that proxies API requests from the sandbox to the Paperclip server, running as a local HTTP server on the host that forwards authenticated requests > - The bridge is started automatically when an adapter launches a sandbox execution, and torn down when the run completes > - The benefit is sandboxed agents can interact with the Paperclip API without requiring network-level access to the host, enabling E2B and similar providers to work end-to-end ## What Changed - Added `sandbox-callback-bridge.ts` in `packages/adapter-utils/` — a lightweight HTTP bridge server that accepts requests from sandbox environments and proxies them to the Paperclip API with authentication - Added request validation and security policy: the bridge only forwards requests to the configured API URL, validates content types, enforces size limits, and rejects non-API paths - Wired the bridge into all remote adapter execute paths (claude, codex, cursor, gemini, pi) — the bridge starts before the agent process and the bridge URL is passed via environment variables - Updated `environment-execution-target.ts` to prefer the explicit API URL from environment lease metadata for sandbox callback routing - Fixed Claude sandbox runtime setup to work with the bridge configuration - Added comprehensive test coverage for bridge request handling, policy enforcement, and sandbox execution integration - Fixed browser bundling — the bridge module is excluded from the frontend bundle via the adapter-utils index export ## Verification - `pnpm test` — all existing and new tests pass, including bridge unit tests and sandbox execution integration tests - `pnpm typecheck` — clean - Manual: configure an E2B environment, run an agent task, verify the agent can post comments and update issue status through the bridge ## Risks - Medium. This is a new network-facing component (HTTP server on localhost). The security policy restricts forwarding to the configured API URL only and validates all requests, but any proxy introduces attack surface. The bridge binds to localhost only and is scoped to the lifetime of a single agent run. ## Model Used Codex GPT 5.4 high via Paperclip. ## 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 run tests locally and they pass - [x] I have added or updated tests where applicable - [ ] If this change affects the UI, I have included before/after screenshots - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [x] I will address all Greptile and reviewer comments before requesting merge |
||
|
Devin Foley
|
70679a3321 |
Add sandbox environment support (#4415)
## Thinking Path > - Paperclip orchestrates AI agents for zero-human companies. > - The environment/runtime layer decides where agent work executes and how the control plane reaches those runtimes. > - Today Paperclip can run locally and over SSH, but sandboxed execution needs a first-class environment model instead of one-off adapter behavior. > - We also want sandbox providers to be pluggable so the core does not hardcode every provider implementation. > - This branch adds the Sandbox environment path, the provider contract, and a deterministic fake provider plugin. > - That required synchronized changes across shared contracts, plugin SDK surfaces, server runtime orchestration, and the UI environment/workspace flows. > - The result is that sandbox execution becomes a core control-plane capability while keeping provider implementations extensible and testable. ## What Changed - Added sandbox runtime support to the environment execution path, including runtime URL discovery, sandbox execution targeting, orchestration, and heartbeat integration. - Added plugin-provider support for sandbox environments so providers can be supplied via plugins instead of hardcoded server logic. - Added the fake sandbox provider plugin with deterministic behavior suitable for local and automated testing. - Updated shared types, validators, plugin protocol definitions, and SDK helpers to carry sandbox provider and workspace-runtime contracts across package boundaries. - Updated server routes and services so companies can create sandbox environments, select them for work, and execute work through the sandbox runtime path. - Updated the UI environment and workspace surfaces to expose sandbox environment configuration and selection. - Added test coverage for sandbox runtime behavior, provider seams, environment route guards, orchestration, and the fake provider plugin. ## Verification - Ran locally before the final fixture-only scrub: - `pnpm -r typecheck` - `pnpm test:run` - `pnpm build` - Ran locally after the final scrub amend: - `pnpm vitest run server/src/__tests__/runtime-api.test.ts` - Reviewer spot checks: - create a sandbox environment backed by the fake provider plugin - run work through that environment - confirm sandbox provider execution does not inherit host secrets implicitly ## Risks - This touches shared contracts, plugin SDK plumbing, server runtime orchestration, and UI environment/workspace flows, so regressions would likely show up as cross-layer mismatches rather than isolated type errors. - Runtime URL discovery and sandbox callback selection are sensitive to host/bind configuration; if that logic is wrong, sandbox-backed callbacks may fail even when execution succeeds. - The fake provider plugin is intentionally deterministic and test-oriented; future providers may expose capability gaps that this branch does not yet cover. ## Model Used - OpenAI Codex coding agent on a GPT-5-class backend in the Paperclip/Codex harness. Exact backend model ID is not exposed in-session. Tool-assisted workflow with shell execution, file editing, git history inspection, and local test execution. ## 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 run tests locally and they pass - [x] I have added or updated tests where applicable - [ ] If this change affects the UI, I have included before/after screenshots - [x] I have updated relevant documentation to reflect my changes - [x] I have considered and documented any risks above - [x] I will address all Greptile and reviewer comments before requesting merge |
