85f4205de3
## Thinking Path
> - Paperclip is the open source app people use to manage AI agents for
work
> - Pull requests to this repository run a large CI matrix. The slowest
check sets the time a contributor waits for a green run.
> - After #15853, the twelve `server (n/12)` shards are the slowest
check in every fully green PR run created since its merge (518s to 548s;
`Build` is 482s to 504s). Each shard is about 60s of setup and 454s to
490s of Vitest, with one worker, so its wall time is the sum of its
suites.
> - From two shard logs, 22 to 24 of the 71 to 74 suites in a shard
start an embedded Postgres cluster. Each one runs `initdb`, starts the
cluster, and applies all 331 migrations before its first test. The
smallest such suite takes 6s with one or two trivial tests. On top of
that, the `getEmbeddedPostgresTestSupport` probe starts and stops a
second throwaway cluster at import time in 19 of those suites, which is
why their per-file import overhead is 3.5s to 4.1s against 0.9s to 1.0s
for suites without Postgres.
> - A cluster data directory that was shut down cleanly can be copied
and started again as an independent cluster, and the migrated schema is
the same for every suite.
> - This pull request builds one migrated cluster per host, publishes
its data directory as a template under the OS temp directory, and starts
every later suite from a copy of it. The probe reuses the template
instead of starting a cluster.
> - The benefit is about 7s to 8s per Postgres suite on the CI runners
(measured 3.2s to 0.35s per start and 0.75s to 0.03s per probe locally),
about 120s to 170s per server shard, and a faster local test loop for
every suite that uses the fixture.
## Linked Issues or Issue Description
**What existing behavior does this improve?**
The `server (n/12)` PR checks are the slowest check in every fully green
PR run since 2026-10-11 05:27Z. Their Vitest step runs suites one at a
time, and about a third of each shard's time is embedded Postgres setup
that produces the same migrated cluster every time.
**Subsystem affected**
Test infrastructure: the embedded Postgres test fixture in
`@paperclipai/db` (`packages/db/src/test-embedded-postgres.ts`), used by
349 server suites, the db package suites, and the CLI suites.
**Current behavior**
`startEmbeddedPostgresTestDatabase` runs `initdb`, starts the cluster,
creates the database, and applies all 331 migrations for every suite.
`getEmbeddedPostgresTestSupport` starts and stops a second throwaway
cluster once per test file, because the module-level cache does not
survive Vitest's per-file isolation. Measured locally: 3.2s per suite
start plus 0.4s cleanup, and 0.75s per probe. On the CI runners the
smallest Postgres suite takes 6s and the probe adds 2.5s to 3s of import
time per file.
**Proposed behavior**
The first caller on a host builds one migrated cluster, stops it, and
renames its data directory to
`$TMPDIR/paperclip-embedded-postgres-templates-<uid>/<key>`. Every later
caller copies the template (about 90 MB) and starts Postgres on the
copy, then still runs `ensurePostgresDatabase` and
`applyPendingMigrations`, which are no-ops on a current template. The
probe treats a published template as proof that the host can run the
cluster. The key hashes the embedded-postgres version, the initdb flags,
the user, and the name and content of every migration file plus the
journal, so a changed migration builds a new template.
`PAPERCLIP_TEST_POSTGRES_TEMPLATE=0` bypasses the cache. Any template
failure falls back to the fresh path.
A template is a Postgres data directory, and starting it executes its
configuration, so the cache is private to the current OS user: the root
name carries the uid, the root is created with mode 0700, and both the
root and each template are used only when `lstat` shows a real directory
(not a symlink) owned by the current user with no group or world write
bit. A root that fails those checks (for example one pre-created by
another user in a shared `/tmp`) is left alone and the process uses
fresh clusters.
**Reason and benefit**
The slowest check decides the wall-clock time of every PR run. Measured
locally, a template start takes 0.3s to 0.4s instead of 3.2s, cleanup
0.15s instead of 0.43s, and the probe 0.03s instead of 0.75s. Four
server suites that passed in both modes dropped by 3.4s to 6.0s each;
the db `backup-lib` suite dropped from 42.7s to 20.9s. The CI runners
are about twice as slow, so the expected saving per server shard is 120s
to 170s.
**Breaking changes**
None. The returned database is fully migrated in both paths, each suite
still gets its own cluster and data directory, and a failure on the
template path falls back to the existing fresh path.
## What Changed
- `packages/db/src/test-embedded-postgres.ts`: adds the template cache.
`computeEmbeddedPostgresTemplateKey` hashes the inputs that determine a
migrated cluster. `buildEmbeddedPostgresTemplate` starts a fresh cluster
through the existing bounded retry, creates the database, applies
migrations, waits for the real `stop()` (30s bound), and publishes the
data directory with an atomic rename; a concurrent builder that loses
the rename discards its copy.
`startEmbeddedPostgresFromTemplateWithRetry` copies the template into a
fresh temp directory, sets mode 0700, and starts the cluster with the
same per-attempt port and directory rotation as the fresh path.
`startEmbeddedPostgresTestDatabase` tries the template first and falls
back to the fresh path with one warning per process.
`probeEmbeddedPostgresSupport` returns supported when a template exists
or can be built, and keeps its throwaway-cluster path for the case where
the template could not be published. Templates for other keys are pruned
after seven days. The connection strings and initdb flags move into
shared constants; no other behavior changes.
- `packages/db/src/test-embedded-postgres.test.ts`: two unit tests for
the key (stable for identical inputs; changes with migration content, a
new file, flags, version, or user; hashes this package's real migrations
directory) and two real-cluster tests (the published template is
complete and private, a suite starts from it in `template` mode with
migrations up to date, a table written to one copy is not visible in the
next copy, and `PAPERCLIP_TEST_POSTGRES_TEMPLATE=0` runs the `fresh`
path). The real-cluster tests skip on hosts without embedded Postgres,
like every other fixture suite.
- `doc/DEVELOPING.md`: a short section under Test Commands describing
the template, its key, and the bypass variable.
**Review fixes (second commit), one per finding:**
- Private cache root (Greptile P1, Superagent P2). The root is
`$TMPDIR/paperclip-embedded-postgres-templates-<uid>` (user name on
Windows), created with mode 0700. `prepareTemplateRootDir` and
`isUsableTemplate` both reject, via `lstat`, anything that is a symlink,
not a directory, owned by another uid, or writable by group or others; a
rejected root logs one warning and the process uses fresh clusters. A
root created by the first commit with a wider mode is narrowed to 0700
if this user owns it.
- Template failures never skip suites (Greptile P1).
`getEmbeddedPostgresTemplateDir` no longer rejects. A build or root
failure resolves `null` with one warning, and
`probeEmbeddedPostgresSupport` then runs the throwaway-cluster probe, so
only a cluster that cannot start reports `supported: false`. A migration
that cannot apply fails the suite on the fresh path instead of being
hidden by `describe.skip`.
- Idle-based pruning (Greptile P2). Each reuse refreshes the template's
timestamp with `utimes`, so the seven-day prune measures idle time. A
template that disappears after a process resolved its path is not
retried five times: `startEmbeddedPostgresFromTemplateWithRetry` checks
the template before each copy and throws at once, the suite falls back
to a fresh cluster, and the fallback forgets the path so the next caller
in that process rebuilds the template.
- The template-path test enables the cache for itself and restores the
original `PAPERCLIP_TEST_POSTGRES_TEMPLATE` value afterwards (Greptile
P1), so the suite passes under `PAPERCLIP_TEST_POSTGRES_TEMPLATE=0` as
well.
- Eight new tests: `isUsableTemplate` accepts a private cluster
directory and rejects a missing `PG_VERSION`, a leftover
`postmaster.pid`, a symlink, and (POSIX) group- or world-writable modes;
a world-writable pre-created root and a symlinked root both fall back to
fresh clusters, publish nothing, log one warning, and still report
`supported: true` from the probe; a ten-day-old template is refreshed on
reuse; a pruned template is rebuilt on the next start; and a template
build that fails after the cluster started falls through to the fresh
probe. The tests that need an empty cache point `os.tmpdir()` at a
scratch directory (`TMPDIR`, `TEMP` and `TMP` together, so the redirect
holds on Windows as well) and restore it.
**Review fixes (third commit, e8f8e9c8), one per finding:**
- Fallback test on Windows (Greptile P1). `useScratchTmpdir` set only
`TMPDIR`, which Windows `os.tmpdir()` ignores, so the one scratch-root
test that runs on every platform would assert on paths outside the
scratch root there. The helper now sets and restores `TMPDIR`, `TEMP`
and `TMP` together, and that test asserts `os.tmpdir() === scratch`
before it checks any path under it.
- Timestamp test runs on its own (Greptile P2). The test assumed an
earlier test had published a template; under
`PAPERCLIP_TEST_POSTGRES_TEMPLATE=0` the probe does not, so a
name-filtered run threw `ENOENT` from `utimes`. It now starts and cleans
up one suite first, which builds the template when nothing has, and
asserts `isUsableTemplate` before ageing it.
## Verification
- `pnpm --filter @paperclipai/db exec tsc --noEmit`: clean (both
commits).
- `pnpm --filter @paperclipai/db exec vitest run
src/test-embedded-postgres.test.ts`: 15 pass (12 new across both
commits) with the template root absent (first run builds it), with it
present, and with `PAPERCLIP_TEST_POSTGRES_TEMPLATE=0`.
- After the review fixes: `client.test.ts` and `backup-lib.test.ts` (29
tests) pass on the template path, and
`agent-action-audit-routes.test.ts` and `status-cards.test.ts` (36
tests) pass through the server route harness. The second commit's CI run
([38121021162](https://github.com/paperclipai/paperclip/actions/runs/38121021162))
passed every check: `server (n/12)` jobs 277s to 425s, Vitest `Duration`
219s to 365s, no fallback warning in any shard.
- Third commit: reproduced the Greptile P2 on the old file first
(`PAPERCLIP_TEST_POSTGRES_TEMPLATE=0 vitest run
src/test-embedded-postgres.test.ts -t "refreshes the template"` on an
empty cache root fails with `ENOENT … utime`), then with the fix the
same command passes, and the full file passes 15/15 on a cold cache, a
warm cache, and with the bypass set. `tsc --noEmit` clean.
**Measured on this PR's first CI run
([38117892083](https://github.com/paperclipai/paperclip/actions/runs/38117892083)),
which ran the branch's fixture in every server shard:**
| | Before (runs 38114982393, 38115279131, 38115459196) | This PR's run
|
| --- | --- | --- |
| Vitest `Duration` per `server (n/12)` shard | 454s to 490s | 224s to
337s |
| `server (n/12)` job | 518s to 548s | 286s to 401s |
| Slowest check in the run | a server shard | `Build`, 453s |
No shard logged a fixture fallback warning, so every Postgres suite in
the run started from the template.
- `pnpm --filter @paperclipai/db exec vitest run
src/status-card-migrations.test.ts
src/nested-skill-folders-migration.test.ts
src/chat-github-migration.test.ts
src/question-response-delivery-migration.test.ts src/client.test.ts`: 5
files, 24 tests pass on the template path. These suites drop indexes,
re-run migration files, and create extra databases on the cluster, so
they cover the semantics the template must preserve.
- Four server suites run with `PAPERCLIP_TEST_POSTGRES_TEMPLATE=0` and
then with the template, all passing in both modes:
| Suite | Fresh path | Template path |
| --- | --- | --- |
| `server/src/__tests__/agent-action-audit-routes.test.ts` (11 tests) |
17.9s | 14.6s |
| `server/src/__tests__/status-cards.test.ts` (25 tests) | 22.2s | 18.6s
|
| `server/src/__tests__/plugin-managed-routines.test.ts` (3 tests) |
18.2s | 14.3s |
| `server/src/__tests__/issue-thread-interactions-service.test.ts` (106
tests) | 19.0s | 12.9s |
| `packages/db/src/backup-lib.test.ts` (9 tests) | 42.7s | 20.9s |
- Per-operation timings on the same host: fresh `initdb` + start +
create database + 331 migrations 3.2s, cleanup 0.43s; template copy
0.20s to 0.23s, start from copy 0.03s to 0.04s, `ensurePostgresDatabase`
0.013s, `applyPendingMigrations` fast path 0.03s to 0.04s, stop 0.03s.
Probe with the template on disk 0.03s (was 0.75s). Template size 87 MB,
3215 files.
- This PR's own server shards run the branch's fixture, so their Vitest
`Duration` lines show the saving directly. Expected on this PR's run:
`server (n/12)` about 330s to 430s (was 440s to 548s on runs
38114982393, 38115279131 and 38115459196), and the slowest check moves
to `Build` (482s to 504s).
**Measured on the three fully green `master` PR runs created after
#15853 merged** (slowest server shard logs):
| Run | Server shards | Vitest `Duration` (slowest shard) | Postgres
suites in it | Their reported duration | Their import overhead |
| --- | --- | --- | --- | --- | --- |
| 38115459196 | 476s to 548s | 483.8s (tests 72%, import 24%) | 22 of 71
| 197s | 83s |
| 38114982393 | 409s to 518s | 452.8s (tests 65%, import 30%) | 24 of 74
| 264s | 98s |
## Risks
- Low risk to correctness. The template is a cleanly stopped cluster
produced by the same `initdb` flags and the same migration runner as the
fresh path, and each suite still runs `ensurePostgresDatabase` and
`applyPendingMigrations` on its copy. Each suite gets its own copy, its
own port, and its own data directory, so isolation between suites is
unchanged.
- A stale template is prevented by the key. If a migration file, the
journal, the embedded-postgres version, the initdb flags, or the user
changes, the key changes and a new template is built. Even if a stale
template were used, `applyPendingMigrations` on the copy would bring it
up to date.
- Concurrent first use (for example the file-parallel db project) can
build the same template twice. The rename is atomic, so one copy wins
and the other is discarded; a template that exists is always complete.
The loser pays one extra fresh start, which is what every suite paid
before.
- The template root lives under the OS temp directory and holds about 90
MB per key. Keys that have not been used for seven days are pruned when
a new template is built; each reuse refreshes the timestamp.
- A template is executable configuration, so the cache is only used when
the root and the template are private directories of the current user
(ownership, no symlink, no group or world write bit). On a host where
another user pre-created the path, the process logs one warning and uses
fresh clusters. Windows has no uid, so there the check is
directory-only; its temp directory is already per user.
- Any failure on the template path (root checks, build, copy, start,
publish) falls back to the fresh path and logs one warning per process,
so a broken cache can make a suite slower but never fail it or skip it.
The guard tests assert the mode, so a silent fallback on CI would fail
`test-embedded-postgres.test.ts` rather than hide.
- Hosts that cannot run embedded Postgres still get `supported: false`
with the real Postgres output, now from the throwaway probe: a failed
template build is one warning, then the probe decides. On such a host
the probe costs one extra failed start attempt cycle per file compared
with the first commit; every suite on that host skips anyway.
## Model Used
- Claude (Anthropic), model ID `claude-fable-5-1`, running as a
Paperclip agent through the Claude local adapter with tool use (shell,
file edits, GitHub CLI). The agent collected the CI data, made the
change, ran the tests, and wrote this description.
## 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 (53 checks green on both run
38117892083 and run 38121021162; the third commit re-runs them)
- [ ] Greptile is 5/5 with no open P2s, recommendations, or follow-ups
(1/5 on the first commit, 4/5 on the second; every finding so far is
fixed and answered, and Superagent confirmed its fix; re-review of the
third commit pending)
- [x] I will address all Greptile and reviewer comments before
requesting merge
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Bender (Fable) <bender-fable@users.noreply.github.com>
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
Co-authored-by: devinfoley <139239+devinfoley@users.noreply.github.com>
Quickstart · Docs · GitHub · Discord · Twitter · Website
Sign up for the Paperclip Cloud waitlist →
Paperclip is the app people use to manage AI agents for work.
Open-source orchestration for teams of AI agents.
If OpenClaw is an employee, Paperclip is the company.
Paperclip is a Node.js server and React UI that orchestrates a team of AI agents to run a business. Bring your own agents, assign goals, and track work and costs from one dashboard. Choose models and harnesses per agent while keeping your team's tasks, skills, permissions, and history in one place.
It looks like a task manager. Under the hood: org charts, budgets, governance, goal alignment, and agent coordination.
Manage business goals, not pull requests.
| Step | Example | |
|---|---|---|
| 01 | Define the goal | "Build the #1 AI note-taking app to $1M MRR." |
| 02 | Hire the team | CEO, CTO, engineers, designers, marketers — any bot, any provider. |
| 03 | Approve and run | Review strategy. Set budgets. Hit go. Monitor from the dashboard. |
| Works with |
 OpenClaw |
 Claude Code |
 Codex |
 Cursor + Cloud |
 Gemini CLI |
 OpenCode |
 Pi |
 Hermes + Gateway |
 Grok Build |
 Kimi Code |
If it can receive a heartbeat, it's hired.
Custom processes, HTTP endpoints, and external adapter packages extend the roster. See the adapter overview for setup and capabilities.
Paperclip is right for you if
- ✅ You want to build autonomous AI organizations
- ✅ You coordinate many different agents (OpenClaw, Codex, Claude, Cursor) toward a common goal
- ✅ You have 20 simultaneous Claude Code terminals open and lose track of what everyone is doing
- ✅ You want agents running autonomously 24/7, but still want to audit work and chime in when needed
- ✅ You want to monitor costs and enforce budgets
- ✅ You want a process for managing agents that feels like using a task manager
- ✅ You want to manage your autonomous businesses from your phone
The four pillars
Four things have to work for an organization of AI agents to actually produce: the tasks, the org, the training, and the infrastructure. Paperclip is built around exactly those four pillars.
| Pillar | Built for | What it covers |
|---|---|---|
| Agentic Task Manager — Declare intent. Agents work. You verify the output. | Everyone, daily | Tasks, approvals & review gates · proactive agent coworkers · auditable routines & workflows · verify from diffs, screenshots & tests |
| Org Chart for Agents — Roles, permissions & boundaries for humans and agents. | Managers | Mixed human + agent org chart · responsibilities, delegation, specialization · governance: who can do what · scoped secrets & company boundaries · connection permissions & responsible-user identities |
| Agent Employee Training — Design, train & evaluate your AI employees. | Enablers | Skill Studio & shared org-wide skills · evals & saved test runs · active learning loops & quality metrics · performance reviews for agents · saved test inputs · skill version history & restore · reusable team templates |
| Agentic OS — The infrastructure that makes the work run. | IT & platform | Cross-provider runtime: any model, any agent · sandboxing, integrations & MCP servers · SSO, GRC, RBAC & cost controls · data privacy, internal trace collection, compounding data value · personal & shared app connections · run history & opt-in tracing |
Features
🔌 Bring Your Own AgentAny agent, any runtime, one org chart. If it can receive a heartbeat, it's hired. |
🎯 Goal AlignmentLink tasks and projects to your organization goals. Agents receive the goal context behind their work. |
💓 HeartbeatsAgents wake for assigned work, follow-up messages, or configured schedules. Delegation flows up and down the org chart. |
💰 Cost ControlCompany, agent, and project budgets. Track reported spend, get threshold alerts, and pause work at configured limits. |
🏢 Multi-OrganizationOne deployment, many organizations. Separate tasks, agents, permissions, and activity histories for each. |
🎫 Task ThreadsKeep conversations, plans, blockers, files, and run history attached to the work. Assign tasks to agents or people. |
🛡️ GovernanceConfigure review and approval stages, approve hires, and pause, reassign, or stop work when needed. |
📊 Org ChartHierarchies, roles, reporting lines. Your agents have a boss, a title, and a job description. |
📱 Mobile ReadyMonitor and manage your autonomous businesses from anywhere. |
🔗 Apps & ConnectionsConnect services such as GitHub, Notion, and Railway, or your own MCP server. Set gateway actions to Allowed, Ask first, or Off. |
👥 Shared Agents, Personal AccountsChoose who can use a connection and which agents can access it. Managed GitHub operations can use the account of the person directing the work. |
🧠 Skills & Skill StudioInstall or write shared skills, test them with saved inputs, inspect results, and restore earlier versions. |
📅 Scheduled RoutinesRun recurring work on a schedule or trigger it through an API or webhook. Each run has a task, an owner, and a history. |
📎 Artifacts & FeedbackFind the files and documents agents produce. Preview supported formats and leave comments on specific passages in documents. |
📦 Ready-Made TeamsPreview and install teams with roles, skills, projects, and routines. Choose their runtimes and make the setup your own. |
Experimental Agent Chat and chat/email connectors add conversations with agents in Paperclip and through configured services such as Slack, Discord, Telegram, and AgentMail. Enable the relevant instance settings to try them.
Problems Paperclip solves
| Without Paperclip | With Paperclip |
|---|---|
| ❌ You have 20 Claude Code tabs open and can't track which one does what. On reboot you lose everything. | ✅ Tasks are ticket-based, conversations are threaded, sessions persist across reboots. |
| ❌ You manually gather context from several places to remind your bot what you're actually doing. | ✅ Context flows from the task up through the project and company goals — your agent always knows what to do and why. |
| ❌ Folders of agent configs are disorganized and you're re-inventing task management, communication, and coordination between agents. | ✅ Paperclip gives you org charts, ticketing, delegation, and governance out of the box — so you run a company, not a pile of scripts. |
| ❌ Runaway loops waste hundreds of dollars of tokens and max your quota before you even know what happened. | ✅ Spend tracking, budget alerts, and automatic pauses help you control the cost of ongoing work. |
| ❌ You have recurring jobs (customer support, social, reports) and have to remember to manually kick them off. | ✅ Routines create assigned tasks on a schedule, with outputs and run history you can inspect. |
| ❌ You have an idea, you have to find your repo, fire up Claude Code, keep a tab open, and babysit it. | ✅ Add a task in Paperclip. Your coding agent works on it until it's done. Management reviews their work. |
Why Paperclip is special
Paperclip handles the hard orchestration details correctly.
| Atomic task checkout. | A single assignee and execution locks prevent competing runs from claiming the same task. |
| Persistent work context. | Tasks, comments, and documents stay in Paperclip. Supporting adapters resume saved sessions across runs. |
| Runtime skill injection. | Agents can learn Paperclip workflows and project context at runtime, without retraining. |
| Governance with rollback. | Approval gates are enforced, config changes are revisioned, and bad changes can be rolled back safely. |
| Accountable connections. | Human access, agent eligibility, and gateway action permissions are separate controls. Approve a call once or save a revocable rule. |
| Goal-aware execution. | Linked tasks and projects carry goal ancestry so agents see the "why," not just a title. |
| Portable company templates. | Export/import orgs, agents, and skills with secret scrubbing and collision handling. |
| Organization boundaries. | Company-scoped access checks keep each organization's work, agents, and activity separate within one deployment. |
What's Under the Hood
Paperclip is a full control plane, not a wrapper. Before you build any of this yourself, know that it already exists:
┌──────────────────────────────────────────────────────────────┐
│ PAPERCLIP SERVER │
│ │
│ ┌───────────┐ ┌───────────┐ ┌───────────┐ ┌───────────┐ │
│ │Identity & │ │ Work & │ │ Heartbeat │ │Governance │ │
│ │ Access │ │ Tasks │ │ Execution │ │& Approvals│ │
│ └───────────┘ └───────────┘ └───────────┘ └───────────┘ │
│ │
│ ┌───────────┐ ┌───────────┐ ┌───────────┐ ┌───────────┐ │
│ │ Org Chart │ │Workspaces │ │ Plugins │ │ Budget │ │
│ │ & Agents │ │ & Runtime │ │ │ │ & Costs │ │
│ └───────────┘ └───────────┘ └───────────┘ └───────────┘ │
│ │
│ ┌───────────┐ ┌───────────┐ ┌───────────┐ ┌───────────┐ │
│ │ Routines │ │ Secrets & │ │ Activity │ │ Company │ │
│ │& Schedules│ │ Storage │ │ & Events │ │Portability│ │
│ └───────────┘ └───────────┘ └───────────┘ └───────────┘ │
└──────────────────────────────────────────────────────────────┘
▲ ▲ ▲ ▲
┌─────┴─────┐ ┌─────┴─────┐ ┌─────┴─────┐ ┌─────┴─────┐
│ Claude │ │ Codex │ │ CLI │ │ HTTP/web │
│ Code │ │ │ │ agents │ │ bots │
└───────────┘ └───────────┘ └───────────┘ └───────────┘
The Systems
|
Identity & Access — Two deployment modes (trusted local or authenticated), human roles and permissions, agent API keys, short-lived run JWTs, company memberships, and invite flows. Responsible-user attribution follows work through delegation and supported managed connections. |
Org Chart & Agents — Agents have roles, titles, reporting lines, permissions, and budgets. Adapter examples match the diagram: Claude Code, Codex, CLI agents such as Cursor/Gemini/bash, HTTP/webhook bots such as OpenClaw, and external adapter plugins. If it can receive a heartbeat, it's hired. |
|
Work & Task System — Issues carry company/project/goal/parent links, atomic checkout with execution locks, first-class blocker dependencies, comments, documents, attachments, work products, labels, and inbox state. Search across work, review document revisions, and leave anchored feedback. |
Heartbeat Execution — DB-backed wakeup queue with coalescing, budget checks, workspace resolution, secret injection, skill loading, and adapter invocation. Runs produce logs, usage records, and adapter-specific session state. Bounded recovery handles supported failures and surfaces cases that need human action. |
|
Workspaces & Runtime — Project repositories and workspaces, optional isolated execution workspaces (git worktrees, operator branches), and runtime services (dev servers, preview URLs). Sandbox providers extend execution beyond the local host; availability depends on the configured environment and adapter. |
Governance & Approvals — Board approval workflows, execution policies with review/approval stages, decision tracking, budget hard-stops, and agent pause/resume/terminate. Configured task reviews govern completion; connection action approvals govern calls through the tool gateway. |
|
Budget & Cost Control — Token and cost tracking by company, agent, project, goal, issue, provider, and model. Scoped budget policies with warning thresholds and hard stops. Enforcement uses recorded spend; usage reporting and in-flight work can delay a stop. |
Routines & Schedules — Recurring tasks with cron, webhook, and API triggers. Concurrency and catch-up policies. Each routine execution creates a tracked issue and wakes the assigned agent — no manual kick-offs needed. |
|
Plugins — Instance-wide plugin system with out-of-process workers, capability-gated host services, job scheduling, tool exposure, and UI contributions. Extend Paperclip without forking it. |
Secrets & Storage — Company secrets and per-person secret values, encrypted credential storage, local or S3-compatible file storage, attachments, and work products. Secret references supply credentials to authorized runs without copying values into ordinary agent configuration. |
|
Activity & Events — Mutating actions, heartbeat state changes, cost events, approvals, comments, and work products are recorded as durable activity so operators can audit what happened and why. |
Company Portability — Preview, export, and import organization packages with agents, skills, and optional projects, routines, tasks, and attachments. Referenced secret values are omitted; review packages before sharing because plain environment values and local paths can remain. Packages share an operating setup; full-instance recovery uses backups. |
What Paperclip is not
| Not just a chatbot. | Conversations stay attached to tasks, plans, decisions, and outputs. Experimental Agent Chat can hand work off to assigned tasks. |
| Not an agent framework. | We don't tell you how to build agents. We tell you how to run a company made of them. |
| Not just a workflow builder. | Routines and experimental pipelines operate within an organization, with roles, goals, budgets, and governance. |
| Not a prompt manager. | Agents bring their own prompts, models, and runtimes. Paperclip manages the organization they work in. |
| Not limited to one agent. | Start with one agent and grow into a team with shared skills, delegation, and review. |
| Not only for code review. | Coding and PR review fit alongside research, operations, content, and other work. |
Quickstart
Open source. Self-hosted. No Paperclip account required. Follow the guided quickstart to set up your first agent.
Just ask your agent to install Paperclip
Share the installation guide with your agent.
Or install it yourself
With Node.js 24.11 or newer installed:
npx paperclipai@latest onboard --yes
The CLI runs from npm's cache; your instance configuration and data persist locally.
See the installation guide for managed installs, pinned versions, canary and git-ref installs, updates, rollback, service management, and uninstalling.
For an isolated manual test instance that is already initialized with a CEO
agent, use test-drive. It stays in the foreground, never installs a service
or creates a first task, and opens the browser only after setup succeeds:
ANTHROPIC_API_KEY=... npx paperclipai test-drive
OPENAI_API_KEY=... npx paperclipai test-drive --harness codex
OPENROUTER_API_KEY=... npx paperclipai test-drive \
--harness opencode \
--model openrouter/anthropic/claude-sonnet-4.5
Each run without --data-dir gets a unique, retained temporary directory; its
absolute path is printed at startup. Pass --data-dir to reuse one, or
--no-browser to leave the initialized instance unopened. When invoked from a
linked Git worktree, test-drive also enables task execution in that worktree.
See the test-drive guide for credential and
reuse behavior.
Troubleshooting: private npm registry
.npmrcIf this fails with an
E404forpaperclipai(or similar) and you use a private npm registry (for example GitHub Packages) via a global~/.npmrc,npxmay be resolvingpaperclipaiagainst that private registry instead of the public npm registry.Diagnostic:
npm config get registryWorkaround (cross-platform; force the public npm registry for this command):
npx --registry https://registry.npmjs.org paperclipai@latest onboard --yes
That quickstart path now defaults to trusted local loopback mode for the fastest first run. To start in authenticated/private mode instead, choose a bind preset explicitly:
npx paperclipai@latest onboard --yes --bind lan
# or:
npx paperclipai@latest onboard --yes --bind tailnet
If you already have Paperclip configured, rerunning onboard keeps the existing config in place. Use npx paperclipai configure to edit settings.
Or manually:
git clone https://github.com/paperclipai/paperclip.git
cd paperclip
pnpm install
pnpm dev
This starts the UI and API at http://localhost:3100. An embedded PostgreSQL database is created automatically — no setup required.
Requirements: Node.js 24.11+, pnpm 9.15+
Local Claude and Codex subscription sign-in also needs Python 3 and the corresponding provider CLI on the Paperclip host. The Docker image includes them.
Source development also builds the native Paperclip Runner when enabled (the self-hosted default). Install a Rust toolchain, or set PAPERCLIP_RUNNER_BINARY to a compatible prebuilt runner.
FAQ
Q: Is this project maintained or just slop?
A: Paperclip is maintained by the Paperclip team. We've merged over 2,700 pull requests.
Q: What does a typical setup look like?
A: Locally, a single Node.js process manages an embedded Postgres and local file storage. For production, point it at your own Postgres and deploy however you like. Configure projects, agents, and goals — the agents take care of the rest.
For remote access, use authenticated mode with a private-network bind such as Tailscale, or deploy the persistent server with Docker. See deployment modes and the Docker guide.
Q: Can I run multiple companies?
A: Yes. A single deployment can host multiple organizations with company-scoped data and access checks.
Q: How is Paperclip different from agents like OpenClaw or Claude Code?
A: Paperclip uses those agents. It orchestrates them into a company — with org charts, budgets, goals, governance, and accountability.
Q: Why should I use Paperclip instead of just pointing my OpenClaw to Asana or Trello?
A: Agent orchestration has subtleties in how you coordinate who has work checked out, how to maintain sessions, monitoring costs, establishing governance - Paperclip does this for you.
(Bring-your-own-ticket-system is on the Roadmap)
Q: Do agents run continuously?
A: Agents wake for assigned work and follow-up messages. Optional timer heartbeats let them check for work periodically; routines create recurring tasks on their own schedules. You can also connect externally running agents such as OpenClaw. A mention alone does not assign work or wake another agent.
Development
pnpm dev # Full dev (API + UI, watch mode)
pnpm dev:once # Full dev without file watching
pnpm dev:server # Server only
pnpm dev:mobile # Serve prebuilt UI on :3101 for phones/tablets (proxies /api → :3100)
pnpm dev:both # Run `pnpm dev` and `pnpm dev:mobile` together
pnpm build # Build all
pnpm typecheck # Type checking
pnpm test # Cheap default test run (Vitest only)
pnpm test:watch # Vitest watch mode
pnpm test:e2e # Playwright browser suite
pnpm db:generate # Generate DB migration
pnpm db:migrate # Apply migrations
pnpm test does not run Playwright. Browser suites stay separate and are typically run only when working on those flows or in CI.
See doc/DEVELOPING.md for the full development guide.
Roadmap
- ✅ Plugin system (e.g. add a knowledge base, custom tracing, queues, etc)
- ✅ Get OpenClaw / claw-style agent employees
- ✅ companies.sh - import and export entire organizations
- ✅ Easy AGENTS.md configurations
- ✅ Skills Manager, Skill Studio & Skills Store
- ✅ Scheduled Routines
- ✅ Better Budgeting
- ✅ Agent Reviews and Approvals
- ✅ Multiple Human Users
- ✅ Cloud / Sandbox agents (e2b, Cloudflare, Daytona, Modal, Novita, self-hosted Kubernetes)
- ✅ Artifacts & Work Products
- ✅ Deep Planning (planning mode, revisioned plans, plan approvals)
- ✅ Enforced Outcomes (watchdogs, recovery actions, review gates)
- ✅ MCP Tool Gateway & Apps (governed tool access)
- ✅ Secrets Manager with per-agent access
- ✅ Activity log & action attribution
- ✅ Self-healing runs & automatic recovery
- ✅ Agent evals & feedback
- ✅ Connected Apps
- ✅ Personal & Shared AI Accounts
- ✅ Shared Agents Use Personal GitHub Identities
- ✅ Skill Version History & Restore
- ✅ Document Comments & Revision History
- ✅ Company-Wide Search
- ✅ Multi-Model & Multi-Harness Teams
- 🟡 Memory / Knowledge
- ⚪ MAXIMIZER MODE
- ⚪ Work Queues
- ⚪ Self-Organization
- ⚪ Automatic Organizational Learning
- 🟡 Agent Chat
- 🟡 Cloud deployments
- ⚪ Desktop App
- ⚪ Bring-your-own-ticket-system (Asana / Linear / Jira as on-ramps)
This is the short roadmap preview. See the full roadmap in ROADMAP.md.
Community & Plugins
Find Plugins and more at awesome-paperclip
Observability
Paperclip ships with opt-in OpenTelemetry auto-instrumentation for the server (traces only). It activates when OTEL_EXPORTER_OTLP_ENDPOINT is set and supports grpc, http/protobuf, and http/json via the standard OTEL_EXPORTER_OTLP_PROTOCOL env var. @opentelemetry/api is a normal server dependency; the SDK, auto-instrumentation, and exporter packages are optional peer dependencies — install them only if you want tracing. See doc/observability.md for install commands and the full env-var reference.
Paperclip also ships with opt-in Sentry error monitoring for the server and the browser. Set SENTRY_DSN_FRONTEND to activate it for the browser and SENTRY_DSN_BACKEND to activate it for the server — each variable is optional, and the legacy SENTRY_DSN variable still works as a fallback for either component. The supported server SDK version is @sentry/node@10.71.0; it is an optional peer dependency for the server, so install it only if you want error monitoring. The browser SDK, @sentry/browser, is pinned to the same exact version. See doc/observability.md for the install command, the privacy settings, and the full default capture set.
Telemetry
Paperclip collects anonymous usage telemetry to help us understand how the product is used and improve it. No personal information, issue content, prompts, file paths, or secrets are ever collected. Private repository references are hashed with a per-install salt before being sent.
Contributors changing emitted telemetry events should follow the Telemetry Data Contract. For proposed first-party events that are not in the generated contract yet, follow Telemetry Workflow.
Telemetry is enabled by default and can be disabled with any of the following:
| Method | How |
|---|---|
| Environment variable | PAPERCLIP_TELEMETRY_DISABLED=1 |
| Standard convention | DO_NOT_TRACK=1 |
| CI environments | Automatically disabled when CI=true |
| Config file | Set telemetry.enabled: false in your Paperclip config |
Contributing
We welcome contributions. See the contributing guide for details.
Community
- Discord — Join the community
- Twitter / X — Follow updates and announcements
- GitHub Issues — bugs and feature requests
- GitHub Discussions — ideas and RFC
License
MIT © 2026 Paperclip Labs, Inc
Star History
Open source under MIT. Built for people who want to get work done, not babysit agents.