## Why
The build derived `<title>` from the **sidebar label** and the
description from the **first paragraph clipped to 220 characters**. Both
are the wrong source. A sidebar label is written for a 240px-wide nav
that supplies its own context; a search result has none.
On production that produced:
- **7 pages titled "Overview"**, and 32 pages sharing a title with
another page
- **109 of 193 descriptions cut off mid-word**, and 174 over 160
characters (where Google truncates)
```
/guides/welcome/key-concepts/
BEFORE (220): ...and if you misunderstand them early, the product wi
```
Near-identical titles are the weak-differentiation signal behind
"crawled — currently not indexed", which is the exact symptom on this
property.
## What changed
`seo_title` and `seo_description` frontmatter on **every one of the 192
pages**, written by hand against each page's real content and optimised
for the click — lead with what the reader can do, name the concrete
things covered, no keyword stuffing, no repeating the title.
| | Before | After |
|---|---|---|
| Duplicate titles | 32 pages | **0** |
| Duplicate descriptions | — | **0** |
| Descriptions cut mid-word | 109 | **0** |
| Descriptions over 160 chars | 174 | **0** |
| Title length (incl. suffix) | up to "Overview" | **26–60** |
| Description length | 31–240 | **122–158** |
**No prose was rewritten.** I measured content overlap between the
colliding pairs first — only 17–23%, so these were always distinct pages
wearing the same label, not duplicate content. The `docs/` diff is **564
insertions, zero deletions**.
### Examples
```
BEFORE — all seven identical: "Overview | Paperclip Docs"
AFTER: "Adapters Overview" "API Overview" "CLI Overview"
"Deployment Overview" "Bundled Skills" "Optional Skills"
"Experimental Features"
BEFORE — both identical: "Issues | Paperclip Docs"
AFTER: "Issues: Creating and Tracking Work" /guides/day-to-day/issues/
"Issues API" /reference/api/issues/
```
```
/guides/getting-started/five-minute-path/
BEFORE (220): The shortest route from a running Paperclip instance to an agent that has
just completed work for you. After Paperclip is installed, the path takes
about 5 minutes: create a company, hire a CEO agent, approve its first s
AFTER (131): Install Paperclip, create a company, hire a CEO agent, and approve its
first strategy — a working AI company in about five minutes.
```
## Plumbing
- **`site/build-release.mjs`** prefers the authored fields; the derived
path stays only as a fallback the CI gate prevents ever reaching.
`<title>`, `<meta name="description">`, and the `TechArticle` JSON-LD
all read from the same source.
- **`site/app.js`** uses the same fields, so a client-side transition
cannot leave a different title on screen than the crawler was served.
Verified in-browser: navigating `/guides/day-to-day/issues/` →
`/reference/api/issues/` swaps to exactly the server-rendered values, no
JS errors.
- **`scripts/verify-seo-metadata.mjs`** (new) enforces presence,
uniqueness, length, complete-sentence endings, and the
frontmatter-safety rules. Wired into `docs:test` and CI as its own step.
I negative-tested the gate rather than assuming it works — it correctly
fails on a duplicate title, a mid-word description, and a missing field,
and passes again once restored.
## Automating it going forward
`skills/sync-docs/SKILL.md` now:
- gains an **"Authored SEO metadata"** section: why the fields exist,
the enforced rules as a table, and how to actually write them (title
names the subject not the nav position; description writes for the
click; never generate mechanically from the first paragraph — *that is
the fallback the gate exists to prevent*).
- requires both fields in the **page-authoring instruction** given to
every subagent, with guidance on when to update an existing description
versus leave it.
- runs `docs:test:seo-metadata` and `docs:test:crawlable-links` as
**fail-loud gates in Phase 7** before commit.
- documents all six build gates in the files/gates section.
So once this lands, a new page cannot ship without hand-written
metadata, and a title collision cannot reach production.
## Verified
All six verifiers pass:
```
Static route verification passed
Asset fingerprint verification passed
Hierarchical Skills nav verification passed
Skill source block verification passed (11 skill pages, 1 control page)
Crawlable-link contract passed: 40886 internal links across 194 pages resolve in one hop
Authored-SEO contract passed: 192 pages carry a unique, hand-written seo_title and seo_description
```
🤖 Generated with [Claude Code](https://claude.com/claude-code)
---------
Co-authored-by: Claude Opus 5 (1M context) <[email protected]>
9.1 KiB
paperclip_version, seo_title, seo_description
| paperclip_version | seo_title | seo_description |
|---|---|---|
| v2026.618.0 | OpenClaw Gateway Adapter | Connect to an OpenClaw instance over the Gateway WebSocket protocol when OpenClaw runs on another machine, in Docker, or behind a network boundary. |
OpenClaw Gateway
openclaw_gateway connects Paperclip to an OpenClaw instance over the OpenClaw Gateway WebSocket protocol. Use it when OpenClaw is running on another machine, inside Docker, or behind a shared gateway and you want Paperclip agents to delegate work to it.
Info:
openclaw_gatewayis fully functional in the runtime but currently shown as "Coming soon" in the agent-config adapter-type dropdown. It's reached through the OpenClaw invite-prompt flow instead of being picked manually. This page documents the adapter so you can target it via the API or an imported company. Direct UI selection is planned.
When To Use
- OpenClaw is reachable over
ws://orwss://(local Docker, remote host, Tailscale). - You want a shared OpenClaw instance to serve multiple Paperclip agents.
- You need device-auth pairing between Paperclip and the gateway.
When Not To Use
- The remote agent doesn't speak the OpenClaw Gateway protocol. Use HTTP.
- You want to run a headless local CLI agent on the same machine. Use Claude Code, Codex, or Process.
Transport
The adapter always uses WebSocket gateway transport. The URL must start with ws:// or wss://. The connect flow is:
- Receive
connect.challengefrom the gateway. - Send
req connectwith protocol, client, auth, and device payload. - Send
req agent. - Wait for completion via
req agent.wait. - Stream
event agentframes into Paperclip's logs and transcript parser.
Common Fields
| Field | Required | Notes |
|---|---|---|
url |
yes | Gateway WebSocket URL (ws:// or wss://). |
authToken / token |
no* | Gateway auth token. |
headers |
no* | Alternative auth carrier: headers["x-openclaw-token"] or legacy headers["x-openclaw-auth"]. |
password |
no* | Shared-password auth mode. |
disableDeviceAuth |
no | Omit the signed device payload when true. Default false. |
devicePrivateKeyPem |
no | Pins a stable Ed25519 signing key. Without it, the adapter generates an ephemeral keypair per run. |
autoPairOnFirstConnect |
no | Default true. Handles one initial pairing required by calling device.pair.list + device.pair.approve over shared auth, then retries once. |
sessionKeyStrategy |
no | issue, fixed, or run. Determines how agent.sessionKey is derived. |
sessionKey |
no | Used when sessionKeyStrategy is fixed. |
agentId |
no | Optional OpenClaw agent ID to target. |
role |
no | Gateway identity role sent to OpenClaw. Default operator. |
scopes |
no | Gateway capability scopes (an array, e.g. ["operator.admin"]). In the create UI you enter a comma-separated string that's split into an array. Default ["operator.admin"]. |
paperclipApiUrl |
no | Optional override for the Paperclip API base URL the gateway/agent calls back into. No default. |
workspaceRuntime |
no | Optional workspace runtime services definition (a { services: [...] } object). In the create UI you enter it as runtimeServicesJson; it's applied only when it parses to an object with a services array. |
payloadTemplate |
no | Fields merged into the agent request. message/text prefix the wake text. |
timeoutSec |
no | Adapter-level request budget. |
waitTimeoutMs |
no | agent.wait.timeoutMs for a single wait call. |
*One of authToken/token, headers, or password must be present. When a token is provided and the authorization header is missing, the adapter derives Authorization: Bearer <token>.
Device Auth
By default the adapter sends a signed device payload in connect params. Recommended setup:
- Leave
disableDeviceAuthunset orfalse. - Provide a stable
devicePrivateKeyPemso device identity persists across runs. - Let
autoPairOnFirstConnecthandle the first-time pairing.
If pairing cannot be completed automatically (token mismatch, no pending request) the first run returns pairing required. Approve the pending device inside OpenClaw, then retry the task.
For a local Docker smoke, you can approve from the host:
docker exec openclaw-docker-openclaw-gateway-1 sh -lc \
'openclaw devices approve --latest --json \
--url "ws://127.0.0.1:18789" \
--token "$(node -p \"require(process.env.HOME+\\\"/.openclaw/openclaw.json\\\").gateway.auth.token\")"'
Inspect pending vs paired devices:
docker exec openclaw-docker-openclaw-gateway-1 sh -lc \
'TOK="$(node -e \"const c=require(\\\"/home/node/.openclaw/openclaw.json\\\");process.stdout.write(c.gateway?.auth?.token||\\\"\\\")\")"; \
openclaw devices list --json --url "ws://127.0.0.1:18789" --token "$TOK"'
Session Strategy
Same routing model as HTTP OpenClaw mode:
sessionKeyStrategy=issue— one OpenClaw session per issue (good default).sessionKeyStrategy=fixed— use a singlesessionKeyfor every run.sessionKeyStrategy=run— fresh session per run (no memory across heartbeats).
The resolved session key is sent as agent.sessionKey in each request.
Payload Mapping
The agent request is built as:
Required:
message— the wake text, optionally prefixed withpayloadTemplate.messageorpayloadTemplate.text.idempotencyKey— Paperclip'srunId.sessionKey— resolved from strategy.
Optional:
- All other
payloadTemplatefields are merged in. agentIdis injected if configured and not already present in the template.
Timeouts
timeoutSeccontrols the adapter-level request budget.waitTimeoutMscontrols the per-callagent.wait.timeoutMs.
If agent.wait returns timeout, the adapter returns openclaw_gateway_wait_timeout.
Log Format
Structured gateway events are emitted as:
[openclaw-gateway] ...— lifecycle and system logs.[openclaw-gateway:event] run=<id> stream=<stream> data=<json>—event agentframes.
UI and CLI parsers consume these lines to render the transcript.
Running OpenClaw in Docker
Paperclip ships several smoke harnesses that spin up OpenClaw in Docker and wire it to Paperclip for you.
Automated join smoke test
pnpm smoke:openclaw-join
Automates invite creation, agent join, board approval, one-time API key claim, and wakeup callback delivery to a dockerized receiver (docker/openclaw-smoke).
One-command gateway UI
pnpm smoke:openclaw-docker-ui
Clones or updates openclaw/openclaw in /tmp/openclaw-docker, builds openclaw:local, writes isolated smoke config under ~/.openclaw-paperclip-smoke/, starts openclaw-gateway via Compose, and prints http://127.0.0.1:18789/#token=....
Useful knobs:
| Variable | Default | Purpose |
|---|---|---|
OPENAI_API_KEY |
— (required) | Loaded from env or ~/.secrets |
OPENCLAW_GATEWAY_PORT |
18789 |
Host port for the gateway |
OPENCLAW_GATEWAY_TOKEN |
random | Gateway auth token |
OPENCLAW_BUILD |
— | Set 0 to skip rebuild |
OPENCLAW_DISABLE_DEVICE_AUTH |
1 |
Disables Control UI device pairing; set 0 to require pairing |
OPENCLAW_MODEL_PRIMARY |
openai/gpt-5.2 |
Primary model |
OPENCLAW_RESET_STATE |
1 |
Resets smoke agent state each run |
PAPERCLIP_HOST_FROM_CONTAINER |
host.docker.internal |
How the container reaches Paperclip |
Network tips
Inside OpenClaw Docker, 127.0.0.1 points to the container — not the host. For URLs consumed inside Docker, use host.docker.internal. If Paperclip rejects that hostname, allow it:
pnpm paperclipai allowed-hostname host.docker.internal
For remote gateways, prefer a reachable hostname (Docker host alias, Tailscale hostname, or public domain).
Onboarding Checklist
When you add OpenClaw to a Paperclip company:
- Start Paperclip in auth mode and a clean OpenClaw Docker.
- In the company settings, use Generate OpenClaw Invite Prompt and paste it into OpenClaw's main chat.
- Approve the resulting join request in Paperclip and confirm the agent appears in your company.
- Preflight the agent config:
adapterTypeisopenclaw_gateway(notopenclaw).urlbegins withws://orwss://.- Token is non-trivial (length ≥ 16).
devicePrivateKeyPemis persisted anddisableDeviceAuthisfalse.
- Assign a smoke task to the agent (e.g. "post comment
OKand mark done"). Verify the comment appears and the issue reachesdone.
Example
{
"adapterType": "openclaw_gateway",
"adapterConfig": {
"url": "ws://127.0.0.1:18789",
"headers": {
"x-openclaw-token": {
"type": "secret_ref",
"secretId": "openclaw-gateway-token",
"version": "latest"
}
},
"devicePrivateKeyPem": "<PEM private key stored as a secret reference in production>",
"autoPairOnFirstConnect": true,
"sessionKeyStrategy": "issue",
"timeoutSec": 300,
"waitTimeoutMs": 60000
}
}