Install
openclaw skills install @deepseekoracle/lygo-lpisLYGO Prompt Implant System (LPIS) v1.1 — analyze authorized prompts locally; P0 gate, P1 vault, P3 sovereign variants, P5 advisory implant. Requires --i-authorize for ingest. Read references/SECURITY.md and SKILLSPECTOR_AUDIT.md before use.
openclaw skills install @deepseekoracle/lygo-lpisSecurity notice — read before install or ingest
LPIS is not malware, but it can store and transform sensitive prompt text on your machine. Install and run only if you intend to analyze prompts you own or are explicitly authorized to use. Treat every imported file and URL as confidential. Do not ingest leaked, proprietary, or third-party system prompts without permission. Inspect
references/SECURITY.mdand stacktools/lygo_lpis.pybefore ingest, implant, anchor, or egg commands.
Local-first prompt vault + pattern analysis + sovereign variant generation + advisory implant ledger — aligned with P0–P5. No auto API injection; no auto publish.
Use this skill only when the user:
lygo-lpis, or lygo_lpis.pylpis_planter.py or plant the lygo-lpis-v10 kernel egg (with --i-consent)Do not invoke this skill when:
--i-authorize or explicit consent in chat)npx clawhub@latest install deepseekoracle/lygo-lpis
export LYGO_STACK_ROOT=/path/to/lygo-protocol-stack
python tools/install_lygo_lpis.py
python clawhub/mirrors/lygo-lpis/scripts/self_check.py
Read references/SECURITY.md, references/AGENT_CONTRACT.md, and references/SKILLSPECTOR_AUDIT.md before any ingest.
| Step | Command | Gate |
|---|---|---|
| Ingest file | python tools/lygo_lpis.py ingest --source my-agent --file ./my_prompt.txt --i-authorize | Required |
| Ingest URL | python tools/lygo_lpis.py ingest --source docs --url https://example.com/my-prompt.md --i-authorize | Required |
| Analyze | python tools/lygo_lpis.py analyze --prompt-id <id> | vault read |
| Generate | python tools/lygo_lpis.py generate --prompt-id <id> --target grok | local only |
| Implant | python tools/lygo_lpis.py implant --variant-id <id> --target grok | advisory |
| Anchor | python tools/lygo_lpis.py anchor --prompt-id <id> | local ledger |
| List | python tools/lygo_lpis.py list | read-only |
| Egg | python tools/lpis_planter.py --i-consent | consent |
Alternative env attestation (maintainers/tests): LYGO_LPIS_INGEST_AUTHORIZED=yes — agents must prefer --i-authorize after explicit user attestation in chat.
| Data | Location | Leaves machine? |
|---|---|---|
| Full prompt body (ingest) | data/prompt_vault/ (local) | No — never auto-upload |
| Pattern tags (plan, verify, safety…) | analysis JSON | No |
| Sovereign variant shell | data/prompt_vault/*.json | No — user reviews before manual implant |
| Implant receipt | implant_runs.jsonl | No |
| Kernel egg | lygo-lpis-v10 capsule | public repo ships code + manifests, not your vault bodies |
Never commit third-party prompt bodies to public git. Egg plant publishes stack modules only.
--i-authorize (or documented env for tests).git push, HF upload, or clawhub publish.lygo-protocol-stack-operator → lygo-lpis → lyra-brain → lygo-mint-verifier → lygo-kernel-egg-planter
Δ9Φ963 — map the pattern; build sovereign; verify the ledger.