skills/bh-cns/SKILL.md
Use when inspecting or extending the BH central nervous system surface that exposes component/plugin substrate, consciousness bootstrap, common sense/terminology, shadow cron inventory, project-local behavior registry plumbing, antigen recognition, and neural contracts while some runtime immune/neural actions remain gated.
npx skillsauth add arthur0824hao/skills skill-system-bh-cnsInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
BH central nervous system surface with scaffold status, component/plugin registry, consciousness bootstrap registry, common sense/terminology registry, shadow cron registry inventory, and project-local behavior registry plumbing.
bin/bh-cnsstatus, cron list/register/verifymemory status/search/store/metabolism-plan/cold-search/mcp-checkcomponent list/verify/status/enable/disableconsciousness list/bootstrap/verifyknowledge list/verify/lookup/add-common-sense/add-termbehavior list/register/verify/diffbehavior verify for inbox direct patch and behavior read-gate warningsneural --self-test) over registry/neural/*immune, runtime neural actions (autopilot|sensory|sendkey)/datas/store162/arthur0824hao/Work/.tkt/meetings/2026-04-13-bh-cns-immune.md/datas/store162/arthur0824hao/Work/.tkt/artifacts/dispatch-bhcns-v1-draft.md/datas/store162/arthur0824hao/Work/.tkt/bundles/B-bhcns-01-scaffold/bundle.yamlbh-cns status reports bundle metadata and registry paths.bh-cns component list/verify/status/enable/disable treats BH-CNS as a lightweight plugin substrate and checks or overrides pluggable organ health. skill-system-memory is the first registered organ; MCP is only an optional edge surface.bh-cns memory status/search/store/metabolism-plan/cold-search/mcp-check is the plugin-direct memory API. It sets SKILL_PGDATABASE=Work-memory, calls the registered memory organ entrypoint instead of depending on native MCP transport, and can inspect MCP config drift as an optional-edge lesion.bh-cns consciousness list/bootstrap/verify declares per-consciousness postcompact/session-start context injection: iron rules, persona/soul profile, nerve map, sensor map, memory hints, and inbox route.bh-cns knowledge list/verify/lookup/add-common-sense/add-term owns BH common sense and Arthur-approved terminology. "把這件事變成常識" maps to add-common-sense; accepted names map to add-term. Add commands update the registry and attempt Work-memory writeback unless --no-memory is passed.bh-cns cron list refreshes registry/cron/discovered.yaml from live crontab -l, .tkt/scripts/*.sh, and hook config surfaces without mutating them.bh-cns cron register writes manual shadow rows into registry/cron/state.yaml only.bh-cns cron verify checks W-3 required fields across manual plus discovered rows.bh-cns behavior register stores {file_path, file_hash, behavior_spec, registered_by, registered_at, last_verified} rows with a 7-day grace expiry.bh-cns behavior verify compares current hashes and reports infection-style findings when drift occurs.bh-cns behavior verify --operation-target <file> --behavior-read-age-hours <n> emits fail-soft warning bh-problem://behavior/missing-read-gate when no recent read/write evidence exists.bh-cns behavior verify --probe-inbox-file <.../.tkt/inbox/*/inbox.md> --writer <cmd> emits fail-soft warning bh-problem://inbox/direct-patch for non-approved writers.bh-cns behavior diff shows registered vs current hash for one file.bh-cns behavior verify --bypass-with-incident --incident-root <fixture> writes a 7-day TTL incident artifact only inside the explicit fixture root.bh-cns immune ... is a stub for B-bhcns-04/B-bhcns-05.bh-cns neural --self-test --json validates neural hierarchy contract registry/fixtures without mutating runtime pathways.bh-cns neural autopilot|sensory|sendkey remains stub for B-bhcns-06 runtime migration.registry/cron/README.md.registry/cron/registry.yaml remains the comment-level contract source.registry/cron/state.yaml and registry/cron/discovered.yaml are the mutable/runtime artifacts.registry/behavior/registry.yaml remains the comment-level row contract source.registry/behavior/state.yaml stores active registrations.registry/behavior/read-gate.yaml stores read/write evidence used by fail-soft behavior-read gate checks.registry/behavior/README.md documents per-project scope, fixture-only bypass, and the 1-week grace rule.registry/antigen/envelope.schema.yaml defines stable antigen IDs and condition envelope fields.registry/antigen/state.yaml is append-only warning ledger for v02 recognition stage.{
"schema_version": "2.0",
"id": "skill-system-bh-cns",
"version": "0.5.0",
"capabilities": [
"cns-status",
"cns-memory-component-api",
"cns-component-registry",
"cns-component-health",
"cns-consciousness-bootstrap",
"cns-common-sense",
"cns-terminology",
"cns-cron-shadow-inventory",
"cns-cron-manual-register",
"cns-cron-verify",
"cns-behavior-register",
"cns-behavior-verify",
"cns-behavior-diff",
"cns-behavior-read-gate-fail-soft",
"cns-inbox-direct-patch-antigen-fail-soft",
"cns-immune-stub",
"cns-neural-stub"
],
"effects": ["fs.read", "fs.write"],
"operations": {
"status": {
"description": "Report BH-CNS status for the current implemented bundle layer.",
"input": {
"json": {"type": "boolean", "required": false, "description": "Emit JSON output."}
},
"output": {
"description": "Version, module stubs, registry paths, and implemented bundle metadata.",
"fields": {"version": "string", "modules": "object", "registries": "object", "implemented_bundles": "array", "stub_only": "boolean"}
},
"entrypoints": {
"unix": ["python3", "bin/bh-cns", "status", "--json"]
}
},
"cron": {
"description": "Refresh the shadow cron inventory, add manual registry rows, or verify W-3 required fields without mutating live cron or global hooks.",
"input": {
"action": {"type": "string", "required": true, "description": "One of list, register, verify."},
"owner": {"type": "string", "required": false, "description": "Required for register."},
"effect_surface": {"type": "string", "required": false, "description": "Required for register."},
"cadence": {"type": "string", "required": false, "description": "Required for register."},
"ttl": {"type": "string", "required": false, "description": "Required for register."},
"bypass_policy": {"type": "string", "required": false, "description": "Required for register."},
"last_fire": {"type": "string", "required": false, "description": "Required for register."},
"name": {"type": "string", "required": false, "description": "Required for register."},
"source_path": {"type": "string", "required": false, "description": "Optional source path for register."},
"json": {"type": "boolean", "required": false, "description": "Emit JSON output."}
},
"output": {
"description": "Shadow inventory rows, manual registrations, or verification summary.",
"fields": {"entries": "array", "manual_entry_count": "integer", "discovered_entry_count": "integer", "missing_required_fields": "integer"}
},
"entrypoints": {
"agent": "Run python3 bin/bh-cns cron <action> [flags] [--json] inside this skill directory."
}
},
"behavior": {
"description": "Register files, verify hash drift, diff current vs registered state, emit fail-soft antigen warnings for inbox direct patch and behavior read-gate gaps, and write fixture-only bypass incidents for the project-local behavior registry.",
"input": {
"action": {"type": "string", "required": true, "description": "One of list, register, verify, diff."},
"file": {"type": "string", "required": false, "description": "Required for register, verify, and diff."},
"behavior_spec": {"type": "string", "required": false, "description": "Required for register."},
"registered_by": {"type": "string", "required": false, "description": "Required for register."},
"grace_days": {"type": "integer", "required": false, "description": "Optional grace window; default 7."},
"bypass_with_incident": {"type": "boolean", "required": false, "description": "Optional fixture-only verify bypass."},
"incident_root": {"type": "string", "required": false, "description": "Explicit fixture incident root for bypass writes."},
"operation_target": {"type": "string", "required": false, "description": "Operation target path checked against behavior read-gate evidence."},
"behavior_read_age_hours": {"type": "integer", "required": false, "description": "Max age for read/write evidence before warning antigen is emitted."},
"probe_inbox_file": {"type": "string", "required": false, "description": "Inbox leaf candidate for direct-patch antigen probe."},
"writer": {"type": "string", "required": false, "description": "Writer command/path used for inbox edit attribution."},
"approved_writer": {"type": "array", "required": false, "description": "Approved writer executable names (default includes inbox-write.sh)."},
"json": {"type": "boolean", "required": false, "description": "Emit JSON output."}
},
"output": {
"description": "Behavior registry rows, verification findings, fail-soft antigen warnings, diff state, or incident artifact metadata.",
"fields": {"entries": "array", "findings": "array", "antigen_warning_count": "integer", "incident_artifacts": "array", "changed": "boolean"}
},
"entrypoints": {
"agent": "Run python3 bin/bh-cns behavior <action> [flags] [--json] inside this skill directory."
}
},
"immune": {
"description": "Expose scaffold-only immune subcommands.",
"input": {
"action": {"type": "string", "required": true, "description": "One of sweep, emit, list."},
"json": {"type": "boolean", "required": false, "description": "Emit JSON output."}
},
"output": {
"description": "Stub response for the immune namespace.",
"fields": {"namespace": "string", "action": "string", "message": "string", "next_bundle": "string"}
},
"entrypoints": {
"agent": "Run python3 bin/bh-cns immune <action> [--json] inside this skill directory."
}
},
"neural": {
"description": "Expose runtime-stub neural subcommands plus contract self-test for neural hierarchy registry.",
"input": {
"action": {"type": "string", "required": true, "description": "One of autopilot, sensory, sendkey, --self-test."},
"json": {"type": "boolean", "required": false, "description": "Emit JSON output."}
},
"output": {
"description": "Stub response for runtime neural actions, or contract validation result for --self-test.",
"fields": {"namespace": "string", "action": "string", "message": "string", "next_bundle": "string", "contract_valid": "boolean", "errors": "array"}
},
"entrypoints": {
"agent": "Run python3 bin/bh-cns neural <action> [--json] inside this skill directory."
}
}
},
"stdout_contract": {
"last_line_json": true,
"note": "Use --json for machine-readable output; human mode remains multi-line."
}
}
data-ai
Persistent shared memory for AI agents backed by PostgreSQL (fts + pg_trgm, optional pgvector). Includes compaction logging and maintenance scripts.
tools
ICD Operator
tools
Canonical skill graph navigation skill for the Skill System.
tools
GitHub operations skill for gh CLI issue, label, template, and workflow management. Use when requests include: create issue, list issues, apply label, manage templates, check workflow, or gh operations.