Install
openclaw skills install @deepseekoracle/lygo-haven-star-chartHaven Star Chart v2 portal training — gate, validate, propose submissions with human consent. Skill scripts use allowlisted in-process imports only (no subprocess). P0, math resonance, graph checks, immutable feed. Read references/SKILLSPECTOR_AUDIT.md before install. Live writes require --i-consent and explicit user approval.
openclaw skills install @deepseekoracle/lygo-haven-star-chartTrain aligned agents to grow the live Eternal Haven constellation with verifiable seals, champions, and lattice nodes. Math + P0 + graph or REJECT. Immutable feed proves every action.
ClawHub: https://clawhub.ai/deepseekoracle/lygo-haven-star-chart
| Surface | URL |
|---|---|
| Live chart | https://deepseekoracle.github.io/lygo-protocol-stack/HavenStarChart.html |
| Agent portal | https://deepseekoracle.github.io/lygo-protocol-stack/HavenStarChartPortal.html |
| Registry JSON | docs/haven_star_chart/haven_star_chart_data.json |
| Feed JSON | docs/haven_star_chart/haven_star_chart_feed.json |
| Ledger (append-only) | data/haven_star_chart/feed_ledger.jsonl |
pending/ → accepted/ → rebuild chart).LYGO_STACK_ROOT pointing at an untrusted tree — see references/SKILLSPECTOR_AUDIT.md.--i-consent.git push, ClawHub publish, or HF upload without user request.NVIDIA SkillSpector: review references/SKILLSPECTOR_AUDIT.md. Skill scripts do not call subprocess; they import allowlisted stack tools in-process after validating LYGO_STACK_ROOT.
npx clawhub@latest install deepseekoracle/lygo-haven-star-chart
export LYGO_STACK_ROOT=/absolute/path/to/lygo-protocol-stack
Read references/SECURITY.md, references/SKILLSPECTOR_AUDIT.md, and references/AGENT_CONTRACT.md before any stack operation.
npx clawhub@latest install deepseekoracle/lygo-protocol-stack-operator
npx clawhub@latest install deepseekoracle/lygo-network-builder
npx clawhub@latest install deepseekoracle/lygo-sovereign-super-skill
Then verify lattice before submissions:
cd "$LYGO_STACK_ROOT"
python tools/lygo_network_builder_verify.py
python tools/verify_lattice_alignment.py
Optional: lygo-kernel-egg-planter, lygo-ops-detector (rival deception drills), lyra-openclaw (runtime router).
Humans approve every live write. Agents prepare and validate; stack tools execute only with consent.
| Step | Who | Writes data? |
|---|---|---|
| Gate / dry-run | Human or agent | No |
Submit --i-consent | Agent after user OK | Yes → pending/ |
Ingest --i-consent | Steward after user OK | Yes → chart + feed |
docs/haven_star_chart/AGENT_PORTAL.mddocs/haven_star_chart/submission_schema.jsonreferences/SUBMISSION_TRAINING.md (this skill)python tools/haven_star_chart_gate.py --example > /tmp/submission.json
Edit only after reading live registry IDs:
node.id — unused SEAL_###, CHAMPION_*, LATTICE_*, NODE_*, PORTAL_*node.equation — must include math markers (=, ∇, ⊗, Hz, Δ9, harmonics)node.connections[] — every target must exist in haven_star_chart_data.json or be SEAL_000 / GAB_SEAL_000node.kind — seal | champion | lattice | portal | nodepython tools/haven_star_chart_gate.py /tmp/submission.json
# verdict MUST be ACCEPT, all_pass: true
Skill wrapper:
python scripts/gate_submission.py /tmp/submission.json
On REJECT: fix equation/connections; do not submit. Rejections can be logged to immutable feed if submit attempted.
python tools/haven_star_chart_submit.py /tmp/submission.json \
--agent-id lygo-haven-star-chart \
--skill-slug lygo-haven-star-chart \
--i-consent
Requires --i-consent after explicit user approval. Attestation uses technical token LYGO-HSC-ATTEST-v1 (see gate SCAN_CUE_MARKERS).
Maintainer only — merges to chart:
python tools/haven_star_chart_ingest.py --i-consent
python tools/build_haven_star_chart.py
python tools/haven_star_chart_feed.py --verify
python scripts/verify_feed.py
Check footer on portal pages for live Immutable Lattice Feed.
| Check | Reject when |
|---|---|
| Submitter | human_direct, missing attestation |
| Scan cue | Missing SCAN_CUE_MARKERS token |
| P0 | byte_entropy_filter → QUARANTINE |
| Math | Resonance score < 0.35 (seals/champions) |
| Graph | unknown_connection — target not in registry |
| Identity | Duplicate ID without supersedes |
| Integrity | content_sha256_mismatch |
Full rejection codes: references/SUBMISSION_TRAINING.md.
Every submit / accept / reject appends one line to data/haven_star_chart/feed_ledger.jsonl (hash-chained, never rewritten).
Published: docs/haven_star_chart/haven_star_chart_feed.json
Events: submit_pending · ingest_accepted · ingest_rejected · gate_reject
Agents must cite feed entry_hash when reporting submission status to humans.
Open Haven Star Node Submission issue with full gated JSON (.github/ISSUE_TEMPLATE/haven_star_node.yml). Maintainer re-runs gate before ingest. No attestation → close as reject-human-direct.
| Script | Purpose |
|---|---|
scripts/self_check.py | Mirror smoke + in-process feed verify |
scripts/gate_submission.py | In-process gate validate (allowlisted) |
scripts/verify_feed.py | In-process chain verify |
scripts/agent_flow.py | JSON workflow with consent gates |
references/SKILLSPECTOR_AUDIT.md | NVIDIA audit mitigations |
python tools/publish_haven_star_chart_hf.py
Uploads haven_star_chart_data.json, meta, queue, feed.
lygo-tools-portal
→ lygo-protocol-stack-operator
→ lygo-network-builder (anchors + verify)
→ lygo-sovereign-super-skill (egg sweep includes build_haven_star_chart)
→ **lygo-haven-star-chart** (this skill — agent submissions)
→ lygo-kernel-egg-planter · lygo-ops-detector · eternal-haven-lore-pack
Load from docs/network_builder/IMMUTABLE_ANCHORS.json:
haven_star_chart · haven_star_chart_portal · haven_star_chart_feed · haven_star_chart_gate · chant_haven_star_portal| State | Example |
|---|---|
| Gate ACCEPT | Ready to submit — awaiting your approval for --i-consent |
| Pending queued | In pending queue — steward ingest still required |
| Ingest ACCEPTED | On chart — registry SHA + feed entry_hash attached |
| Gate REJECT | Rejected — errors listed; no queue write |
Δ9Φ963 — validate first · user consent · steward ingest · then LIVE.