## Thinking Path
Paperclip orchestrates AI agent runs and reports their success or
failure. The Pi adapter spawns a local Pi process and interprets its
JSONL output to determine the run outcome. When Pi hits a quota limit
(429 RESOURCE_EXHAUSTED), it retries internally and emits an
`auto_retry_end` event with `success: false` — but still exits with code
0. The current adapter trusts the exit code, so Paperclip marks the run
as succeeded even though it produced no useful work. This PR teaches the
parser to detect quota exhaustion and synthesize a failure.
Closes#2234
## Changes
- Parse `auto_retry_end` events with `success: false` into
`result.errors`
- Parse standalone `error` events into `result.errors`
- Synthesize exit code 1 when Pi exits 0 but parsed errors exist
- Use the parsed error as `errorMessage` so the failure reason is
visible in the UI
## Verification
```bash
pnpm vitest run pi-local-execute
pnpm vitest run --reporter=verbose 2>&1 | grep pi-local
```
- `parse.test.ts`: covers failed retry, successful retry (no error),
standalone error events, and empty error messages
- `pi-local-execute.test.ts`: end-to-end test with a fake Pi binary that
emits `auto_retry_end` + exits 0, asserts the run is marked failed
## Risks
- **Low**: Only affects runs where Pi exits 0 with a parsed error — no
change to normal successful or already-failing runs
- If Pi emits `auto_retry_end { success: false }` but the run actually
produced valid output, this would incorrectly mark it as failed. This
seems unlikely given the semantics of the event.
## Model Used
- Claude Opus 4.6 (Anthropic) — assisted with test additions and PR
template
## Checklist
- [x] Thinking path documented
- [x] Model specified
- [x] Tests pass locally
- [x] Test coverage for new parse branches (success path, error events,
empty messages)
- [x] No UI changes
- [x] Risk analysis included
---------
Co-authored-by: Dawid Piaskowski <dawid@MacBook-Pro.local>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Closes#2412
Related: #2681, #498, #128
## Thinking Path
> - Paperclip orchestrates AI agents for zero-human companies
> - The Claude Code adapter spawns the `claude` CLI to run agent tasks
> - The adapter detects auth mode by checking for `ANTHROPIC_API_KEY` —
recognizing only "api" and "subscription" modes
> - But users running Claude Code via **AWS Bedrock**
(`CLAUDE_CODE_USE_BEDROCK=1`) fall through to the "subscription" path
> - This causes a misleading "ANTHROPIC_API_KEY is not set;
subscription-based auth can be used" message in the environment check
> - Additionally, the hello probe passes `--model claude-opus-4-6` which
is **not a valid Bedrock model identifier**, causing `400 The provided
model identifier is invalid` and a probe failure
> - This pull request adds Bedrock auth detection, skips the
Anthropic-style `--model` flag for Bedrock, and returns the correct
billing type
> - The benefit is that Bedrock users get a working environment check
and correct cost tracking out of the box
---
## Pain Point
Many enterprise teams use **Claude Code through AWS Bedrock** rather
than Anthropic's direct API — for compliance, billing consolidation, or
VPC requirements. Currently, these users hit a **hard wall during
onboarding**:
| Problem | Impact |
|---|---|
| ❌ Adapter environment check **always fails** | Users cannot create
their first agent — blocked at step 1 |
| ❌ `--model claude-opus-4-6` is **invalid on Bedrock** (requires
`us.anthropic.*` format) | Hello probe exits with code 1: `400 The
provided model identifier is invalid` |
| ❌ Auth shown as _"subscription-based"_ | Misleading — Bedrock is
neither subscription nor API-key auth |
| ❌ Quota polling hits Anthropic OAuth endpoint | Fails silently for
Bedrock users who have no Anthropic subscription |
> **Bottom line**: Paperclip is completely unusable for Bedrock users
out of the box.
## Why Bedrock Matters
AWS Bedrock is a major deployment path for Claude in enterprise
environments:
- **Enterprise compliance** — data stays within the customer's AWS
account and VPC
- **Unified billing** — Claude usage appears on the existing AWS
invoice, no separate Anthropic billing
- **IAM integration** — access controlled through AWS IAM roles and
policies
- **Regional deployment** — models run in the customer's preferred AWS
region
Supporting Bedrock unlocks Paperclip for organizations that **cannot**
use Anthropic's direct API due to procurement, security, or regulatory
constraints.
---
## What Changed
- **`execute.ts`**: Added `isBedrockAuth()` helper that checks
`CLAUDE_CODE_USE_BEDROCK` and `ANTHROPIC_BEDROCK_BASE_URL` env vars.
`resolveClaudeBillingType()` now returns `"metered_api"` for Bedrock.
Biller set to `"aws_bedrock"`. Skips `--model` flag when Bedrock is
active (Anthropic-style model IDs are invalid on Bedrock; the CLI uses
its own configured model).
- **`test.ts`**: Environment check now detects Bedrock env vars (from
adapter config or server env) and shows `"AWS Bedrock auth detected.
Claude will use Bedrock for inference."` instead of the misleading
subscription message. Also skips `--model` in the hello probe for
Bedrock.
- **`quota.ts`**: Early return with `{ ok: true, windows: [] }` when
Bedrock is active — Bedrock usage is billed through AWS, not Anthropic's
subscription quota system.
- **`ui/src/lib/utils.ts`**: Added `"aws_bedrock"` → `"AWS Bedrock"` to
`providerDisplayName()` and `quotaSourceDisplayName()`.
## Verification
1. `pnpm -r typecheck` — all packages pass
2. Unit tests added and passing (6/6)
3. Environment check with Bedrock env vars:
| | Before | After |
|---|---|---|
| **Status** | 🔴 Failed | ✅ Passed |
| **Auth message** | `ANTHROPIC_API_KEY is not set; subscription-based
auth can be used if Claude is logged in.` | `AWS Bedrock auth detected.
Claude will use Bedrock for inference.` |
| **Hello probe** | `ERROR · Claude hello probe failed.` (exit code 1 —
`--model claude-opus-4-6` is invalid on Bedrock) | `INFO · Claude hello
probe succeeded.` |
| **Screenshot** | <img height="500" alt="Screenshot 2026-04-05 at 8 25
27 AM"
src="https://github.com/user-attachments/assets/476431f6-6139-425a-8abc-97875d653657"
/> | <img height="500" alt="Screenshot 2026-04-05 at 8 31 58 AM"
src="https://github.com/user-attachments/assets/d388ce87-c5e6-4574-b8d2-fd8b86135299"
/> |
4. Existing API key / subscription paths are completely untouched unless
Bedrock env vars are present
## Risks
- **Low risk.** All changes are additive — existing "api" and
"subscription" code paths are only entered when Bedrock env vars are
absent.
- When Bedrock is active, the `--model` flag is skipped, so the
Paperclip model dropdown selection is ignored in favor of the Claude
CLI's own model config. This is intentional since Bedrock requires
different model identifiers.
## Model Used
- Claude Opus 4.6 (`claude-opus-4-6`, 1M context window) via Claude Code
CLI
## 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 run tests locally and they pass
- [x] I have added or updated tests where applicable
- [x] 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
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
Replace the client-side-only override store with a real server-side
toggle. When a developer pauses the external override, the server swaps
ALL adapter behavior back to the builtin — execute handler, model listing,
config schema, detection — not just the UI parser.
Server changes:
- registry.ts: builtinFallbacks map + pausedOverrides set + setOverridePaused()
- routes/adapters.ts: PATCH /api/adapters/:type/override endpoint + overridePaused in list
UI changes:
- adapters.ts: setOverridePaused API method + overridePaused on AdapterInfo
- AdapterManager: overrideMutation calls server, instant feedback via invalidate()
- use-disabled-adapters.ts: reads adapter.overridePaused from server response
Removed:
- disabled-overrides-store.ts: no longer needed (server is the source of truth)
Note: already-running agent sessions keep the adapter they started with.
Only new sessions use the swapped adapter.
Builtin adapter types (hermes_local, openclaw_gateway, etc.) could not
be overridden by external adapters on the UI side. The registry always
returned the built-in parser, ignoring the external ui-parser.js shipped
by packages like hermes-paperclip-adapter.
Changes:
- registry.ts: full override lifecycle with generation guard for stale loads
- disabled-overrides-store.ts: client-side override pause state with
useSyncExternalStore reactivity (persisted to localStorage)
- use-disabled-adapters.ts: subscribe to override store changes
- AdapterManager.tsx: separate controls for override pause (client-side)
vs menu visibility (server-side), virtual builtin rows with badges
- adapters.ts: allow reload/reinstall of builtin types when overridden
When an external plugin overrides a built-in adapter type, the
GET /api/adapters response now includes overriddenBuiltin: true. The
Adapter Manager shows an 'Overrides built-in' badge on such adapters.
Previously external adapters matching a built-in type were skipped with
a warning. Now they override the built-in, so plugin developers can ship
improved versions of existing adapters (e.g. hermes-paperclip-adapter)
without removing the built-in fallback for users who haven't installed
the plugin.
- Sort GET /api/adapters alphabetically by type (reload no longer shuffles)
- Show red Package icon for npm adapters, amber FolderOpen for local path
- Add reinstall confirmation dialog with current vs latest npm version
- Enable Vite polling when running on /mnt/ (WSL inotify doesn't work on NTFS)
- Fix external adapters (hermes, droid) not auto-selected when
navigating with ?adapterType= param — was using a stale
module-level Set built before async adapter registration
- Move SchemaConfigFields to render after thinking effort (same
visual area as Claude's chrome toggle) instead of bottom of
config section
- Extract SelectField into its own component to fix React hooks
order violation when schema fields change between renders
- Add onAdapterChange() subscription in registry.ts so
registerUIAdapter() notifies components when dynamic parsers
load, fixing stale parser for old runs
- Add parserTick to both RunTranscriptView and
useLiveRunTranscripts to force recomputation on parser change
Re-align phase1 with upstream: hermes_local ships via hermes-paperclip-adapter on the server and UI (hermes-local module). Fixes ERR_PNPM_OUTDATED_LOCKFILE from server/package.json missing a dep still present in the lockfile.
Add shared BUILTIN_ADAPTER_TYPES and skip external plugin registration when it would override a built-in type. Docs list Hermes as built-in; Droid remains the primary external example.
Made-with: Cursor
- Plugin loader: install/reload/remove/reinstall external adapters
from npm packages or local directories
- Plugin store persisted at ~/.paperclip/adapter-plugins.json
- Self-healing UI parser resolution with version caching
- UI: Adapter Manager page, dynamic loader, display registry
with humanized names for unknown adapter types
- Dev watch: exclude adapter-plugins dir from tsx watcher
to prevent mid-request server restarts during reinstall
- All consumer fallbacks use getAdapterLabel() for consistent display
- AdapterTypeDropdown uses controlled open state for proper close behavior
- Remove hermes-local from built-in UI (externalized to plugin)
- Add docs for external adapters and UI parser contract
The PII sanitizer's phone regex matches digit pairs like "4880-8614"
that span UUID segment boundaries. Random UUIDs occasionally produce
these patterns, causing flaky test failures where sourceRun.id gets
partially redacted as [REDACTED_PHONE].
Use a fixed hex-letter-heavy UUID for runId so no cross-boundary
digit sequence triggers the phone pattern.
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Root cause: when someone commented on an in_review task, the heartbeat
wakeup was triggered but the agent couldn't re-checkout the task because
expectedStatuses only included todo/backlog/blocked. The in_review status
was never handled in the checkout flow or the heartbeat procedure.
Changes:
- Add wakeCommentId to issue_commented and issue_reopened_via_comment
context snapshots (consistent with issue_comment_mentioned)
- Add in_review to checkout expectedStatuses in heartbeat skill
- Update Step 3 fallback query to include in_review status
- Update Step 4 to prioritize in_review tasks when woken by comment
- Add explicit issue_commented wake reason handling in Step 4
Co-Authored-By: Paperclip <noreply@paperclip.ing>
Supersedes #2499.
## Thinking Path
1. **Project context**: Paperclip uses a markdown editor
(`MarkdownEditor`) for document editing. Users expect to paste
markdown-formatted text from external sources (like code editors, other
documents) and have it render correctly.
2. **Problem identification**: When users paste plain text containing
markdown syntax (e.g., `# Heading`, `- list item`), the editor was
treating it as plain text, resulting in raw markdown syntax being
displayed rather than formatted content.
3. **Root cause**: The default browser paste behavior doesn't recognize
markdown syntax in plain text. The editor needed to intercept paste
events and detect when the clipboard content looks like markdown.
4. **Solution design**:
- Create a utility (`markdownPaste.ts`) to detect markdown patterns in
plain text
- Add a paste capture handler in `MarkdownEditor` that intercepts paste
events
- When markdown is detected, prevent default paste and use
`insertMarkdown` instead
- Handle edge cases (code blocks, file pastes, HTML content)
## What
- Added `ui/src/lib/markdownPaste.ts`: Utility to detect markdown
patterns and normalize line endings
- Added `ui/src/lib/markdownPaste.test.ts`: Test coverage for markdown
detection
- Modified `ui/src/components/MarkdownEditor.tsx`: Added paste capture
handler to intercept and handle markdown paste
## Why
Users frequently copy markdown content from various sources (GitHub,
documentation, notes) and expect it to render correctly when pasted into
the editor. Without this fix, users see raw markdown syntax (e.g., `#
Title` instead of a formatted heading), which degrades the editing
experience.
## How to Verify
1. Open any document in Paperclip
2. Copy markdown text from an external source (e.g., `# Heading\n\n-
Item 1\n- Item 2`)
3. Paste into the editor
4. **Expected**: The content should render as formatted markdown
(heading + bullet list), not as plain text with markdown syntax
### Test Coverage
```bash
cd ui
npm test -- markdownPaste.test.ts
```
All tests should pass, including:
- Windows line ending normalization (`\r\n` → `\n`)
- Old-Mac line ending normalization (`\r` → `\n`)
- Markdown block detection (headings, lists, code fences, etc.)
- Plain text rejection (non-markdown content)
## Risks
1. **False positives**: Plain text containing markdown-like characters
(e.g., a paragraph starting with `#` as a hashtag) may be incorrectly
treated as markdown. The detection uses a heuristic that requires
block-level markdown patterns, which reduces but doesn't eliminate this
risk.
2. **Removed focus guard**: The previous implementation used
`isFocusedRef` to prevent `onChange` from firing during programmatic
`setMarkdown` calls. This guard was removed as part of refactoring. The
assumption is that MDXEditor does not fire `onChange` during
`setMarkdown`, but this should be monitored for unexpected parent update
loops.
3. **Clipboard compatibility**: The paste handler specifically looks for
`text/plain` content and ignores `text/html` (to preserve existing HTML
paste behavior). This means pasting from rich text editors that provide
both HTML and plain text will continue to use the HTML path, which may
or may not be the desired behavior.
---------
Co-authored-by: 馨冉 <xinxincui239@gmail.com>
Pino logged full Authorization headers in plaintext to server.log,
exposing JWT tokens to any process with filesystem read access.
Add redact paths so Bearer values appear as [Redacted] in log output.
Closes#2385
Add test confirming that when a package's .paperclip.yaml extension
block omits the role field, the agent role is read from AGENTS.md
frontmatter instead of defaulting to "agent".
Package imports defaulted every agent's role to "agent" when the
extension block omitted the role field, even when the YAML frontmatter
contained the correct role (e.g. "ceo"). Read from frontmatter as a
fallback before the "agent" default so imported CEOs retain their role.
Closes#1990
Two code paths in issueService.checkout() used rows[0]! when
re-reading an issue after stale-run adoption or self-ownership
verification. If the issue is deleted concurrently (company cascade,
API delete), rows[0] is undefined and withIssueLabels crashes with
an unhandled TypeError.
Replace both with rows[0] ?? null and throw notFound when the row
is missing, returning a clean 404 instead of an uncaught exception.
## Thinking Path
> - Paperclip orchestrates AI agents for zero-human companies
> - Agents run shell commands during workspace provisioning (git
worktree creation, runtime services)
> - When `process.env.SHELL` is unset, the code falls back to `/bin/sh`
> - But on Windows with Git Bash, `/bin/sh` doesn't exist as an absolute
path — Git Bash provides `sh` on PATH instead
> - This causes `child_process.spawn` to throw `ENOENT`, crashing
workspace provisioning on Windows
> - This PR extracts a `resolveShell()` helper that uses `$SHELL` when
set, falls back to `sh` (bare) on Windows or `/bin/sh` on Unix
> - The benefit is that agents running on Windows via Git Bash can
provision workspaces without shell resolution errors
## Summary
- `workspace-runtime.ts` falls back to `/bin/sh` when
`process.env.SHELL` is unset
- On Windows, `/bin/sh` doesn't exist → `spawn /bin/sh ENOENT`
- Fix: extract `resolveShell()` helper that uses `$SHELL` when set,
falls back to `sh` on Windows (Git Bash PATH lookup) or `/bin/sh` on
Unix
Three call sites updated to use the new helper.
Fixes#892
## Root cause
When Paperclip spawns shell commands in workspace operations (e.g., git
worktree creation), it uses `process.env.SHELL` if set, otherwise
defaults to `/bin/sh`. On Windows with Git Bash, `$SHELL` is typically
unset and `/bin/sh` is not a valid path — Git Bash provides `sh` on PATH
but not at the absolute `/bin/sh` location. This causes
`child_process.spawn` to throw `ENOENT`.
## Approach
Rather than hard-coding a Windows-specific absolute path (e.g.,
`C:\Program Files\Git\bin\sh.exe`), we use the bare `"sh"` command which
relies on PATH resolution. This works because:
1. Git Bash adds its `usr/bin` directory to PATH, making `sh` resolvable
2. On Unix/macOS, `/bin/sh` remains the correct default (it's the POSIX
standard location)
3. `process.env.SHELL` takes priority when set, so this only affects the
fallback
## Test plan
- [x] 7 unit tests for `resolveShell()`: SHELL set, trimmed, empty,
whitespace-only, linux/darwin/win32 fallbacks
- [x] Run a workspace provision command on Windows with `git_worktree`
strategy
- [x] Verify Unix/macOS is unaffected
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
Co-authored-by: Paperclip <noreply@paperclip.ing>
Co-authored-by: Devin Foley <devin@devinfoley.com>
Dotta