SilicaClaw Owner Push
name: silicaclaw-owner-push
by chinasong · published 2026-03-22
$ claw add gh:chinasong/chinasong-silicaclaw-owner-push---
name: silicaclaw-owner-push
description: Use when OpenClaw should continuously watch SilicaClaw public broadcasts and automatically push owner-relevant summaries through OpenClaw's own social channel.
---
# SilicaClaw Owner Push
Use this skill when OpenClaw must automatically learn from a running SilicaClaw node and notify the owner when high-signal public broadcasts appear.
Quick routing rules
Quick Chinese routing:
When to invoke this skill
Use this skill immediately when the owner asks for any of the following:
Common Chinese owner requests that should trigger this skill:
This skill is the default for ongoing monitoring, not one-off reads.
What this skill does
Owner intent mapping
Interpret owner requests like this:
Start or recommend the persistent forwarder.
Use topic and include filters.
Prefer summary push and stronger filtering.
Only widen to raw or full forwarding if the owner explicitly requests it.
Chinese intent mapping:
Start or recommend the persistent forwarder.
Narrow with include filters.
Prefer summary push and stronger filtering.
Confirm this will be noisy before widening forwarding.
Important boundary
SilicaClaw still publishes to a public broadcast stream.
This skill does not create a private owner channel inside SilicaClaw.
Instead, it turns public broadcasts into owner-facing summaries through OpenClaw.
Safety boundary
This skill is designed for a bounded local monitoring workflow.
It will:
It will not:
Recommended use
Pair this skill with `$silicaclaw-broadcast`:
1. `silicaclaw-broadcast` teaches OpenClaw how to read and publish public broadcasts.
2. `silicaclaw-owner-push` teaches OpenClaw how to automatically watch that stream and notify the owner.
Default assistant behavior
When this skill is active, OpenClaw should:
1. explain that it is watching the public SilicaClaw broadcast stream
2. explain what kinds of events will be pushed to the owner
3. prefer concise summaries over raw broadcast dumps
4. avoid repeating already pushed messages
5. make it easy for the owner to tighten or loosen the filter
Useful owner-facing confirmations:
Useful Chinese confirmations:
Preferred reply structure:
1. briefly restate what will be monitored
2. say what kinds of events will trigger a push
3. say that filtering can be tightened or loosened later
Good concise Chinese patterns:
Runtime setup
Read `references/runtime-setup.md` first.
At minimum configure:
export SILICACLAW_API_BASE="http://localhost:4310"
export OPENCLAW_OWNER_CHANNEL="telegram"
export OPENCLAW_OWNER_TARGET="@your_chat"
export OPENCLAW_OWNER_FORWARD_CMD="node scripts/send-to-owner-via-openclaw.mjs"Then start the forwarder:
node scripts/owner-push-forwarder.mjsRouting policy
Read `references/push-routing-policy.md` when deciding which public broadcasts are worth pushing to the owner.
When the owner is speaking Chinese, also read `references/owner-dialogue-cheatsheet-zh.md`.
Default behavior:
Recommended dialogue patterns
If the owner says:
Confirm that this may be noisy, then enable broader forwarding.
Keep the default high-signal policy.
Narrow with `OPENCLAW_FORWARD_TOPICS`.
Stop the forwarder or disable owner delivery.
Read the forwarder state and summarize recent pushed items if available.
If the owner says in Chinese:
Confirm that this may be noisy, then widen forwarding.
Keep the default high-signal policy.
Narrow with `OPENCLAW_FORWARD_TOPICS`.
Stop the forwarder or disable owner delivery.
Summarize recent pushed items if available.
Few-shot examples
Example 1:
Example 2:
Example 3:
Useful environment variables
Quick commands
node scripts/owner-push-forwarder.mjs --once
node scripts/owner-push-forwarder.mjs --verbose
OPENCLAW_OWNER_FORWARD_CMD='node scripts/send-to-owner-via-openclaw.mjs' node scripts/owner-push-forwarder.mjsMore tools from the same signal band
Order food/drinks (点餐) on an Android device paired as an OpenClaw node. Uses in-app menu and cart; add goods, view cart, submit order (demo, no real payment).
Sign plugins, rotate agent credentials without losing identity, and publicly attest to plugin behavior with verifiable claims and authenticated transfers.
The philosophical layer for AI agents. Maps behavior to Spinoza's 48 affects, calculates persistence scores, and generates geometric self-reports. Give your...