Files
PaperClipAI/server/src/adapters/registry.ts
T
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):

![Kimi Code in the Add a new agent runtime
picker](https://raw.githubusercontent.com/hawikk/paperclip/assets/kimi-local-pr-screenshots/shots/00-kimicode.png)

![Adapter dropdown with Kimi
Code](https://raw.githubusercontent.com/hawikk/paperclip/assets/kimi-local-pr-screenshots/shots/01-adapter-dropdown-kimi.png)

![Kimi adapter selected: command, model, config
fields](https://raw.githubusercontent.com/hawikk/paperclip/assets/kimi-local-pr-screenshots/shots/02-kimi-adapter-selected.png)

![Kimi models in the model
dropdown](https://raw.githubusercontent.com/hawikk/paperclip/assets/kimi-local-pr-screenshots/shots/03-kimi-model-dropdown.png)

![Successful resumed heartbeat run (kimi_local invocation + parsed
transcript)](https://raw.githubusercontent.com/hawikk/paperclip/assets/kimi-local-pr-screenshots/shots/04-successful-resumed-run.png)

![Agents list showing the Kimi Code
label](https://raw.githubusercontent.com/hawikk/paperclip/assets/kimi-local-pr-screenshots/shots/05-agents-list.png)

## 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>
2026-08-20 12:06:33 -07:00

818 lines
28 KiB
TypeScript

import type {
AdapterModelProfileDefinition,
AdapterRuntimeCommandSpec,
ServerAdapterModule,
} from "./types.js";
import { parseAdapterModelsEnv } from "../services/adapter-models-env.js";
import { stampClaudeAgentIdHeader } from "./claude-agent-id-header.js";
import {
buildSandboxNpmInstallCommand,
getAdapterSessionManagement,
} from "@paperclipai/adapter-utils";
import type { AdapterLoginCapability } from "@paperclipai/adapter-utils";
import {
execute as claudeExecute,
listClaudeSkills,
syncClaudeSkills,
listClaudeModels,
refreshClaudeModels,
testEnvironment as claudeTestEnvironment,
sessionCodec as claudeSessionCodec,
getQuotaWindows as claudeGetQuotaWindows,
getConfigSchema as getClaudeConfigSchema,
CLAUDE_SETUP_TOKEN_COMMAND,
parseSetupTokenPrompt,
parseSetupTokenCredential,
} from "@paperclipai/adapter-claude-local/server";
import {
agentConfigurationDoc as claudeAgentConfigurationDoc,
models as claudeModels,
modelProfiles as claudeModelProfiles,
} from "@paperclipai/adapter-claude-local";
import {
execute as codexExecute,
listCodexSkills,
syncCodexSkills,
testEnvironment as codexTestEnvironment,
sessionCodec as codexSessionCodec,
getQuotaWindows as codexGetQuotaWindows,
getConfigSchema as getCodexConfigSchema,
CODEX_DEVICE_LOGIN_COMMAND,
parseDeviceLoginPrompt,
} from "@paperclipai/adapter-codex-local/server";
import {
agentConfigurationDoc as codexAgentConfigurationDoc,
models as codexModels,
modelProfiles as codexModelProfiles,
} from "@paperclipai/adapter-codex-local";
import {
execute as cursorExecute,
listCursorSkills,
syncCursorSkills,
testEnvironment as cursorTestEnvironment,
sessionCodec as cursorSessionCodec,
} from "@paperclipai/adapter-cursor-local/server";
import {
agentConfigurationDoc as cursorAgentConfigurationDoc,
models as cursorModels,
modelProfiles as cursorModelProfiles,
} from "@paperclipai/adapter-cursor-local";
import {
execute as cursorCloudExecute,
getConfigSchema as getCursorCloudConfigSchema,
sessionCodec as cursorCloudSessionCodec,
testEnvironment as cursorCloudTestEnvironment,
} from "@paperclipai/adapter-cursor-cloud/server";
import { agentConfigurationDoc as cursorCloudAgentConfigurationDoc } from "@paperclipai/adapter-cursor-cloud";
import {
execute as geminiExecute,
listGeminiSkills,
syncGeminiSkills,
testEnvironment as geminiTestEnvironment,
sessionCodec as geminiSessionCodec,
getConfigSchema as getGeminiConfigSchema,
} from "@paperclipai/adapter-gemini-local/server";
import {
agentConfigurationDoc as geminiAgentConfigurationDoc,
models as geminiModels,
modelProfiles as geminiModelProfiles,
} from "@paperclipai/adapter-gemini-local";
import {
execute as grokExecute,
listGrokSkills,
syncGrokSkills,
testEnvironment as grokTestEnvironment,
sessionCodec as grokSessionCodec,
} from "@paperclipai/adapter-grok-local/server";
import {
agentConfigurationDoc as grokAgentConfigurationDoc,
models as grokModels,
} from "@paperclipai/adapter-grok-local";
import {
execute as kimiExecute,
listKimiSkills,
syncKimiSkills,
testEnvironment as kimiTestEnvironment,
sessionCodec as kimiSessionCodec,
} from "@paperclipai/adapter-kimi-local/server";
import {
agentConfigurationDoc as kimiAgentConfigurationDoc,
models as kimiModels,
} from "@paperclipai/adapter-kimi-local";
import {
createHermesGatewayServerAdapter,
createHermesLocalServerAdapter,
} from "@paperclipai/hermes-paperclip-adapter";
import {
execute as openCodeExecute,
listOpenCodeSkills,
syncOpenCodeSkills,
testEnvironment as openCodeTestEnvironment,
sessionCodec as openCodeSessionCodec,
listOpenCodeModels,
} from "@paperclipai/adapter-opencode-local/server";
import {
agentConfigurationDoc as openCodeAgentConfigurationDoc,
models as openCodeModels,
modelProfiles as openCodeModelProfiles,
} from "@paperclipai/adapter-opencode-local";
import {
execute as openclawGatewayExecute,
testEnvironment as openclawGatewayTestEnvironment,
} from "@paperclipai/adapter-openclaw-gateway/server";
import {
agentConfigurationDoc as openclawGatewayAgentConfigurationDoc,
models as openclawGatewayModels,
} from "@paperclipai/adapter-openclaw-gateway";
import { listCodexModels, refreshCodexModels } from "./codex-models.js";
import { listCursorModels } from "./cursor-models.js";
import {
execute as piExecute,
listPiSkills,
syncPiSkills,
testEnvironment as piTestEnvironment,
sessionCodec as piSessionCodec,
listPiModels,
} from "@paperclipai/adapter-pi-local/server";
import {
agentConfigurationDoc as piAgentConfigurationDoc,
modelProfiles as piModelProfiles,
} from "@paperclipai/adapter-pi-local";
import { BUILTIN_ADAPTER_TYPES } from "./builtin-adapter-types.js";
import { buildExternalAdapters } from "./plugin-loader.js";
import { getDisabledAdapterTypes } from "../services/adapter-plugin-store.js";
import { processAdapter } from "./process/index.js";
import { httpAdapter } from "./http/index.js";
function readConfiguredCommand(config: Record<string, unknown>, fallback: string): string {
const value = typeof config.command === "string" ? config.command.trim() : "";
return value.length > 0 ? value : fallback;
}
function hasPathSeparator(command: string): boolean {
return command.includes("/") || command.includes("\\");
}
function shellQuote(value: string): string {
return `'${value.replace(/'/g, `'"'"'`)}'`;
}
function buildNpmRuntimeCommandSpec(
config: Record<string, unknown>,
fallbackCommand: string,
packageName: string,
): AdapterRuntimeCommandSpec {
const command = readConfiguredCommand(config, fallbackCommand);
const canSelfInstall = !hasPathSeparator(command) && command === fallbackCommand;
const installLine = buildSandboxNpmInstallCommand(packageName);
return {
command,
detectCommand: command,
installCommand: canSelfInstall
? `if ! command -v ${shellQuote(command)} >/dev/null 2>&1; then ${installLine}; fi`
: null,
};
}
function buildCursorRuntimeCommandSpec(config: Record<string, unknown>): AdapterRuntimeCommandSpec {
const command = readConfiguredCommand(config, "agent");
return {
command,
detectCommand: command,
installCommand: null,
};
}
const retiredAcpxMessage =
"The acpx_local adapter has been retired. Existing Claude and Codex ACPX agents should be migrated to claude_local or codex_local with adapterConfig.engine=\"acp\".";
const retiredAcpxAgentConfigurationDoc = `# acpx_local retired
Adapter: acpx_local
The standalone ACPX adapter has been retired. Use:
- claude_local with adapterConfig.engine="acp" for Claude ACP execution.
- codex_local with adapterConfig.engine="acp" for Codex ACP execution.
Paperclip keeps this tombstone registered so stale acpx_local rows fail clearly instead of falling back to the process adapter.
`;
// The Claude interactive login capability. Claude runs `claude setup-token` on a
// real pseudo-terminal. The user pastes a browser code back into the flow. The
// flow uses a fixed host-side timeout and records a stored session identifier on
// success. The capability data holds no secret; the callbacks return runtime
// values only.
const claudeLoginCapability: AdapterLoginCapability = {
panelMode: "submitted_browser_code",
sandboxTransport: "pseudo_terminal",
timeoutPolicy: "fixed",
getCommand: () => CLAUDE_SETUP_TOKEN_COMMAND,
parsePrompt: (output) => {
const prompt = parseSetupTokenPrompt(output);
return prompt ? { url: prompt.url } : null;
},
captureCredential: (output) => {
const token = parseSetupTokenCredential(output);
return token === null ? null : Buffer.from(token, "utf8");
},
completionClaim: "storedSessionId",
};
// The Codex interactive login capability. Codex runs `codex login --device-auth`
// over the streamed exec channel. The flow shows a one-time code that the user
// enters in the browser. The caller sets the host-side timeout. The device-login
// flow writes its credential inside the sandbox, so the capability declares no
// terminal credential capture and no completion claim.
const codexLoginCapability: AdapterLoginCapability = {
panelMode: "displayed_code",
sandboxTransport: "streamed_exec",
timeoutPolicy: "caller_bounded",
getCommand: () => CODEX_DEVICE_LOGIN_COMMAND,
parsePrompt: (output) => {
const prompt = parseDeviceLoginPrompt(output);
return prompt ? { url: prompt.url, code: prompt.code } : null;
},
};
const claudeLocalAdapter: ServerAdapterModule = {
type: "claude_local",
execute: stampClaudeAgentIdHeader(claudeExecute),
testEnvironment: claudeTestEnvironment,
acp: {
agentId: "claude",
skillsMode: "ephemeral",
prerequisites: {
nodeRange: ">=22.12.0",
packages: ["@agentclientprotocol/claude-agent-acp"],
},
},
listSkills: listClaudeSkills,
syncSkills: syncClaudeSkills,
sessionCodec: claudeSessionCodec,
sessionManagement: getAdapterSessionManagement("claude_local") ?? undefined,
models: claudeModels,
modelProfiles: claudeModelProfiles,
listModels: listClaudeModels,
refreshModels: refreshClaudeModels,
supportsLocalAgentJwt: true,
supportsInstructionsBundle: true,
instructionsPathKey: "instructionsFilePath",
requiresMaterializedRuntimeSkills: false,
getRuntimeCommandSpec: (config) =>
buildNpmRuntimeCommandSpec(config, "claude", "@anthropic-ai/claude-code"),
agentConfigurationDoc: claudeAgentConfigurationDoc,
getConfigSchema: getClaudeConfigSchema,
getQuotaWindows: claudeGetQuotaWindows,
loginCapability: claudeLoginCapability,
};
const acpxLocalAdapter: ServerAdapterModule = {
type: "acpx_local",
async execute(ctx) {
await ctx.onLog("stderr", `${retiredAcpxMessage}\n`);
await ctx.onMeta?.({
adapterType: "acpx_local",
command: "acpx_local-retired",
commandNotes: [retiredAcpxMessage],
});
return {
exitCode: 1,
signal: null,
timedOut: false,
errorMessage: retiredAcpxMessage,
errorCode: "acpx_local_retired",
provider: "acpx",
summary: retiredAcpxMessage,
};
},
async testEnvironment() {
return {
adapterType: "acpx_local",
status: "fail",
testedAt: new Date().toISOString(),
checks: [
{
code: "acpx_local_retired",
level: "error",
message: retiredAcpxMessage,
hint: "Set the agent adapter to claude_local or codex_local and set adapterConfig.engine to acp.",
},
],
};
},
models: [],
supportsLocalAgentJwt: false,
supportsInstructionsBundle: false,
requiresMaterializedRuntimeSkills: false,
agentConfigurationDoc: retiredAcpxAgentConfigurationDoc,
getConfigSchema: () => ({ fields: [] }),
};
const codexLocalAdapter: ServerAdapterModule = {
type: "codex_local",
execute: codexExecute,
testEnvironment: codexTestEnvironment,
acp: {
agentId: "codex",
skillsMode: "ephemeral",
prerequisites: {
nodeRange: ">=22.13.0",
packages: ["@agentclientprotocol/codex-acp"],
},
},
listSkills: listCodexSkills,
syncSkills: syncCodexSkills,
sessionCodec: codexSessionCodec,
sessionManagement: getAdapterSessionManagement("codex_local") ?? undefined,
models: codexModels,
modelProfiles: codexModelProfiles,
listModels: listCodexModels,
refreshModels: refreshCodexModels,
supportsLocalAgentJwt: true,
supportsInstructionsBundle: true,
instructionsPathKey: "instructionsFilePath",
requiresMaterializedRuntimeSkills: false,
getRuntimeCommandSpec: (config) => buildNpmRuntimeCommandSpec(config, "codex", "@openai/codex"),
agentConfigurationDoc: codexAgentConfigurationDoc,
getConfigSchema: getCodexConfigSchema,
getQuotaWindows: codexGetQuotaWindows,
loginCapability: codexLoginCapability,
};
const cursorLocalAdapter: ServerAdapterModule = {
type: "cursor",
execute: cursorExecute,
testEnvironment: cursorTestEnvironment,
listSkills: listCursorSkills,
syncSkills: syncCursorSkills,
sessionCodec: cursorSessionCodec,
sessionManagement: getAdapterSessionManagement("cursor") ?? undefined,
models: cursorModels,
modelProfiles: cursorModelProfiles,
listModels: listCursorModels,
supportsLocalAgentJwt: true,
supportsInstructionsBundle: true,
instructionsPathKey: "instructionsFilePath",
requiresMaterializedRuntimeSkills: true,
getRuntimeCommandSpec: buildCursorRuntimeCommandSpec,
agentConfigurationDoc: cursorAgentConfigurationDoc,
};
const cursorCloudAdapter: ServerAdapterModule = {
type: "cursor_cloud",
execute: cursorCloudExecute,
testEnvironment: cursorCloudTestEnvironment,
sessionCodec: cursorCloudSessionCodec,
sessionManagement: getAdapterSessionManagement("cursor_cloud") ?? undefined,
models: [],
supportsLocalAgentJwt: false,
supportsInstructionsBundle: true,
instructionsPathKey: "instructionsFilePath",
requiresMaterializedRuntimeSkills: false,
agentConfigurationDoc: cursorCloudAgentConfigurationDoc,
getConfigSchema: getCursorCloudConfigSchema,
};
const geminiLocalAdapter: ServerAdapterModule = {
type: "gemini_local",
execute: geminiExecute,
testEnvironment: geminiTestEnvironment,
acp: {
agentId: "gemini",
skillsMode: "ephemeral",
prerequisites: {
nodeRange: ">=20.0.0",
packages: ["@google/gemini-cli"],
},
},
listSkills: listGeminiSkills,
syncSkills: syncGeminiSkills,
sessionCodec: geminiSessionCodec,
sessionManagement: getAdapterSessionManagement("gemini_local") ?? undefined,
models: geminiModels,
modelProfiles: geminiModelProfiles,
supportsLocalAgentJwt: true,
supportsInstructionsBundle: true,
instructionsPathKey: "instructionsFilePath",
requiresMaterializedRuntimeSkills: true,
getRuntimeCommandSpec: (config) =>
buildNpmRuntimeCommandSpec(config, "gemini", "@google/gemini-cli"),
agentConfigurationDoc: geminiAgentConfigurationDoc,
getConfigSchema: getGeminiConfigSchema,
};
const grokLocalAdapter: ServerAdapterModule = {
type: "grok_local",
execute: grokExecute,
testEnvironment: grokTestEnvironment,
listSkills: listGrokSkills,
syncSkills: syncGrokSkills,
sessionCodec: grokSessionCodec,
sessionManagement: getAdapterSessionManagement("grok_local") ?? undefined,
models: grokModels,
supportsLocalAgentJwt: true,
supportsInstructionsBundle: true,
instructionsPathKey: "instructionsFilePath",
requiresMaterializedRuntimeSkills: true,
getRuntimeCommandSpec: (config) => ({
command: readConfiguredCommand(config, "grok"),
detectCommand: readConfiguredCommand(config, "grok"),
installCommand: null,
}),
agentConfigurationDoc: grokAgentConfigurationDoc,
};
const kimiLocalAdapter: ServerAdapterModule = {
type: "kimi_local",
execute: kimiExecute,
testEnvironment: kimiTestEnvironment,
acp: {
agentId: "kimi",
skillsMode: "ephemeral",
prerequisites: {
nodeRange: ">=20.0.0",
packages: ["@moonshot-ai/kimi-code"],
},
},
listSkills: listKimiSkills,
syncSkills: syncKimiSkills,
sessionCodec: kimiSessionCodec,
sessionManagement: getAdapterSessionManagement("kimi_local") ?? undefined,
models: kimiModels,
supportsLocalAgentJwt: true,
supportsInstructionsBundle: true,
instructionsPathKey: "instructionsFilePath",
requiresMaterializedRuntimeSkills: true,
getRuntimeCommandSpec: (config) =>
buildNpmRuntimeCommandSpec(config, "kimi", "@moonshot-ai/kimi-code"),
agentConfigurationDoc: kimiAgentConfigurationDoc,
};
const hermesGatewayAdapter = createHermesGatewayServerAdapter();
const hermesLocalAdapter = createHermesLocalServerAdapter();
const openclawGatewayAdapter: ServerAdapterModule = {
type: "openclaw_gateway",
execute: openclawGatewayExecute,
testEnvironment: openclawGatewayTestEnvironment,
models: openclawGatewayModels,
supportsLocalAgentJwt: false,
supportsInstructionsBundle: false,
requiresMaterializedRuntimeSkills: false,
agentConfigurationDoc: openclawGatewayAgentConfigurationDoc,
};
const openCodeLocalAdapter: ServerAdapterModule = {
type: "opencode_local",
execute: openCodeExecute,
testEnvironment: openCodeTestEnvironment,
listSkills: listOpenCodeSkills,
syncSkills: syncOpenCodeSkills,
sessionCodec: openCodeSessionCodec,
models: openCodeModels,
modelProfiles: openCodeModelProfiles,
sessionManagement: getAdapterSessionManagement("opencode_local") ?? undefined,
listModels: listOpenCodeModels,
supportsLocalAgentJwt: true,
supportsInstructionsBundle: true,
instructionsPathKey: "instructionsFilePath",
requiresMaterializedRuntimeSkills: true,
getRuntimeCommandSpec: (config) => buildNpmRuntimeCommandSpec(config, "opencode", "opencode-ai"),
agentConfigurationDoc: openCodeAgentConfigurationDoc,
};
const piLocalAdapter: ServerAdapterModule = {
type: "pi_local",
execute: piExecute,
testEnvironment: piTestEnvironment,
listSkills: listPiSkills,
syncSkills: syncPiSkills,
sessionCodec: piSessionCodec,
sessionManagement: getAdapterSessionManagement("pi_local") ?? undefined,
models: [],
modelProfiles: piModelProfiles,
listModels: listPiModels,
supportsLocalAgentJwt: true,
supportsInstructionsBundle: true,
instructionsPathKey: "instructionsFilePath",
requiresMaterializedRuntimeSkills: true,
getRuntimeCommandSpec: (config) =>
buildNpmRuntimeCommandSpec(config, "pi", "@mariozechner/pi-coding-agent"),
agentConfigurationDoc: piAgentConfigurationDoc,
};
const adaptersByType = new Map<string, ServerAdapterModule>();
// For builtin types that are overridden by an external adapter, we keep the
// original builtin so it can be restored when the override is deactivated.
const builtinFallbacks = new Map<string, ServerAdapterModule>();
// Tracks which override types are currently deactivated (paused). When
// paused, `getServerAdapter()` returns the builtin fallback instead of the
// external. Persisted across reloads via the same disabled-adapters store.
const pausedOverrides = new Set<string>();
function registerBuiltInAdapters() {
for (const adapter of [
acpxLocalAdapter,
claudeLocalAdapter,
codexLocalAdapter,
openCodeLocalAdapter,
piLocalAdapter,
cursorCloudAdapter,
cursorLocalAdapter,
geminiLocalAdapter,
grokLocalAdapter,
kimiLocalAdapter,
hermesGatewayAdapter,
hermesLocalAdapter,
openclawGatewayAdapter,
processAdapter,
httpAdapter,
]) {
adaptersByType.set(adapter.type, adapter);
}
}
registerBuiltInAdapters();
// ---------------------------------------------------------------------------
// Load external adapter plugins (e.g. droid_local)
//
// External adapter packages export createServerAdapter() which returns a
// ServerAdapterModule. When the module provides its own sessionManagement
// it is preserved; otherwise the host falls back to the built-in registry
// lookup (so externals that override a built-in type inherit the builtin's
// policy). This brings init-time registration to at-least-as-good behavior
// as the hot-install path (routes/adapters.ts:179 -> registerServerAdapter):
// both preserve module-provided sessionManagement, and init-time additionally
// applies the registry fallback for externals overriding a built-in type.
// ---------------------------------------------------------------------------
/** Cached sync wrapper — the store is a simple JSON file read, safe to call frequently. */
function getDisabledAdapterTypesFromStore(): string[] {
return getDisabledAdapterTypes();
}
/**
* Merge an external adapter module with host-provided session management.
*
* Module-provided `sessionManagement` takes precedence. When absent, fall
* back to the hardcoded registry keyed by adapter type (so externals that
* override a built-in — same `type` — inherit the builtin's policy). If
* neither is available, `sessionManagement` remains `undefined`.
*
* Used by both the init-time IIFE below (external-adapter load pass on
* server start) and the hot-install path in `routes/adapters.ts`
* (`registerWithSessionManagement`), so the two load paths resolve
* `sessionManagement` identically.
*/
export function resolveExternalAdapterRegistration(
externalAdapter: ServerAdapterModule,
): ServerAdapterModule {
return {
...externalAdapter,
sessionManagement:
externalAdapter.sessionManagement
?? getAdapterSessionManagement(externalAdapter.type)
?? undefined,
};
}
/**
* Load external adapters from the plugin store and hardcoded sources.
* Called once at module initialization. The promise is exported so that
* callers (e.g. assertKnownAdapterType, app startup) can await completion
* and avoid racing against the loading window.
*/
const externalAdaptersReady: Promise<void> = (async () => {
try {
const externalAdapters = await buildExternalAdapters();
for (const externalAdapter of externalAdapters) {
const overriding = BUILTIN_ADAPTER_TYPES.has(externalAdapter.type);
if (overriding) {
console.log(
`[paperclip] External adapter "${externalAdapter.type}" overrides built-in adapter`,
);
// Save the original builtin for later restoration.
const existing = adaptersByType.get(externalAdapter.type);
if (existing && !builtinFallbacks.has(externalAdapter.type)) {
builtinFallbacks.set(externalAdapter.type, existing);
}
}
adaptersByType.set(
externalAdapter.type,
resolveExternalAdapterRegistration(externalAdapter),
);
}
} catch (err) {
console.error("[paperclip] Failed to load external adapters:", err);
}
})();
/**
* Await this before validating adapter types to avoid race conditions
* during server startup. External adapters are loaded asynchronously;
* calling assertKnownAdapterType before this resolves will reject
* valid external adapter types.
*/
export function waitForExternalAdapters(): Promise<void> {
return externalAdaptersReady;
}
export function registerServerAdapter(adapter: ServerAdapterModule): void {
if (BUILTIN_ADAPTER_TYPES.has(adapter.type) && !builtinFallbacks.has(adapter.type)) {
const existing = adaptersByType.get(adapter.type);
if (existing) {
builtinFallbacks.set(adapter.type, existing);
}
}
adaptersByType.set(adapter.type, adapter);
}
export function unregisterServerAdapter(type: string): void {
if (type === processAdapter.type || type === httpAdapter.type) return;
if (builtinFallbacks.has(type)) {
pausedOverrides.delete(type);
const fallback = builtinFallbacks.get(type);
if (fallback) {
adaptersByType.set(type, fallback);
}
return;
}
if (BUILTIN_ADAPTER_TYPES.has(type)) {
return;
}
adaptersByType.delete(type);
}
export function requireServerAdapter(type: string): ServerAdapterModule {
const adapter = findActiveServerAdapter(type);
if (!adapter) {
throw new Error(`Unknown adapter type: ${type}`);
}
return adapter;
}
export function getServerAdapter(type: string): ServerAdapterModule {
return findActiveServerAdapter(type) ?? processAdapter;
}
/**
* Memoized view of PAPERCLIP_ADAPTER_MODELS, keyed by the raw env string so
* tests (and live env mutation) that change the variable are still observed.
* Parsing happens at most once per distinct raw value instead of per
* `listAdapterModels` request, and malformed values fail SOFT here: we log the
* parse error once (per distinct raw value) and fall back to adapter-discovered
* models rather than throwing at request time.
*/
let adapterModelsEnvCache: {
raw: string | undefined;
value: ReturnType<typeof parseAdapterModelsEnv>;
} | null = null;
function getDeclaredAdapterModels(): ReturnType<typeof parseAdapterModelsEnv> {
const raw = process.env.PAPERCLIP_ADAPTER_MODELS;
if (adapterModelsEnvCache && adapterModelsEnvCache.raw === raw) {
return adapterModelsEnvCache.value;
}
let value: ReturnType<typeof parseAdapterModelsEnv> = null;
try {
value = parseAdapterModelsEnv(process.env);
} catch (err) {
console.error(
"[paperclip] Invalid PAPERCLIP_ADAPTER_MODELS; ignoring declared model lists:",
err,
);
}
adapterModelsEnvCache = { raw, value };
return value;
}
export async function listAdapterModels(type: string): Promise<{ id: string; label: string }[]> {
const declaredModels = getDeclaredAdapterModels();
if (declaredModels && declaredModels[type]?.length) {
return declaredModels[type].map((m) => ({ id: m.id, label: m.label ?? m.id }));
}
const adapter = findActiveServerAdapter(type);
if (!adapter) return [];
if (adapter.listModels) {
const discovered = await adapter.listModels();
if (discovered.length > 0) return discovered;
}
return adapter.models ?? [];
}
export async function refreshAdapterModels(type: string): Promise<{ id: string; label: string }[]> {
const adapter = findActiveServerAdapter(type);
if (!adapter) return [];
if (adapter.refreshModels) {
const refreshed = await adapter.refreshModels();
if (refreshed.length > 0) return refreshed;
}
if (adapter.listModels) {
const discovered = await adapter.listModels();
if (discovered.length > 0) return discovered;
}
return adapter.models ?? [];
}
export async function listAdapterModelProfiles(type: string): Promise<AdapterModelProfileDefinition[]> {
const adapter = findActiveServerAdapter(type);
if (!adapter) return [];
if (adapter.listModelProfiles) {
const discovered = await adapter.listModelProfiles();
if (discovered.length > 0) return discovered;
}
return adapter.modelProfiles ?? [];
}
export function listServerAdapters(): ServerAdapterModule[] {
return Array.from(adaptersByType.values());
}
/**
* List adapters excluding those that are disabled in settings.
* Used for menus and agent creation flows — disabled adapters remain
* functional for existing agents but hidden from selection.
*/
export function listEnabledServerAdapters(): ServerAdapterModule[] {
const disabled = getDisabledAdapterTypesFromStore();
const disabledSet = disabled.length > 0 ? new Set(disabled) : null;
return disabledSet
? Array.from(adaptersByType.values()).filter((a) => !disabledSet.has(a.type))
: Array.from(adaptersByType.values());
}
export async function detectAdapterModel(
type: string,
): Promise<{ model: string; provider: string; source: string; candidates?: string[] } | null> {
const adapter = findActiveServerAdapter(type);
if (!adapter?.detectModel) return null;
const detected = await adapter.detectModel();
if (!detected) return null;
return {
model: detected.model,
provider: detected.provider,
source: detected.source,
...(detected.candidates?.length ? { candidates: detected.candidates } : {}),
};
}
// ---------------------------------------------------------------------------
// Override pause / resume
// ---------------------------------------------------------------------------
/**
* Pause or resume an external override for a builtin adapter type.
*
* - `paused = true` → subsequent calls to `getServerAdapter(type)` return
* the builtin fallback instead of the external adapter. Already-running
* agent sessions are unaffected (they hold a reference to the module they
* started with).
*
* - `paused = false` → the external adapter is active again.
*
* Returns `true` if the state actually changed, `false` if the type is not
* an override or was already in the requested state.
*/
export function setOverridePaused(type: string, paused: boolean): boolean {
if (!builtinFallbacks.has(type)) return false;
const wasPaused = pausedOverrides.has(type);
if (paused && !wasPaused) {
pausedOverrides.add(type);
console.log(`[paperclip] Override paused for "${type}" — builtin adapter restored`);
return true;
}
if (!paused && wasPaused) {
pausedOverrides.delete(type);
console.log(`[paperclip] Override resumed for "${type}" — external adapter active`);
return true;
}
return false;
}
/** Check whether the external override for a builtin type is currently paused. */
export function isOverridePaused(type: string): boolean {
return pausedOverrides.has(type);
}
/** Get the set of types whose overrides are currently paused. */
export function getPausedOverrides(): Set<string> {
return pausedOverrides;
}
export function findServerAdapter(type: string): ServerAdapterModule | null {
return adaptersByType.get(type) ?? null;
}
export function findActiveServerAdapter(type: string): ServerAdapterModule | null {
if (pausedOverrides.has(type)) {
const fallback = builtinFallbacks.get(type);
if (fallback) return fallback;
}
return adaptersByType.get(type) ?? null;
}