nacl-core
Shared NaCl conventions for Codex-adapted skills: graph configuration, Neo4j tool assumptions, schema locations, ID conventions, Excalidraw board conventions, confirmation gates, and verification vocabulary. Use when a Codex NaCl skill needs common graph or workflow rules. Also use when the user refers to `nacl-core`.
ITSalt · Static checks on file
Installation
Read the source and dependencies before running a command. Copying does not install or execute anything.
npx skills add ITSalt/NaCl --skill nacl-coreAbout this skill
name: nacl-core
description: |
Shared NaCl conventions for Codex-adapted skills: graph configuration,
Neo4j tool assumptions, schema locations, ID conventions, Excalidraw board
conventions, confirmation gates, and verification vocabulary. Use when a
Codex NaCl skill needs common graph or workflow rules. Also use when the user
refers to nacl-core.
NaCl Core For Codex
Installation mode preflight
Before any workflow work, reuse a nacl_installation_doctor result from the
current invocation or call that tool once when it is available. Continue only
when it returns status=VERIFIED; a FAILED or BLOCKED result stops the
workflow with its actionable guidance.
If the tool is absent or cannot be called, never infer legacy-only mode and do
not shell to a package/cache path. Report BLOCKED. The separate legacy
symlink distribution owns its own fallback; a plugin workflow uses only the
package MCP nacl_installation_doctor and preserves its exact result.
This skill is a shared reference for Codex-adapted NaCl skills. It is not a user-facing workflow skill.
Core Rules
- Preserve graph-first BA/SA methodology and TL lifecycle contracts.
- Read project
config.yamlbefore graph-aware work when file access is available. - Use only statuses from
../references/verification-vocabulary.md. - Honour the evidence taxonomy in
../references/verification-evidence.mdwhenever a skill writes a terminalTask.statusto the graph. - Apply confirmation gates from
../references/migration-rules.md. - Do not select or constrain the runtime model.
Graph Configuration
Resolve graph settings from the current project's config.yaml when file access
is available:
| Data | Source priority |
|---|---|
| Neo4j Bolt port | graph.neo4j_bolt_port > fallback 3587 |
| Neo4j HTTP port | graph.neo4j_http_port > fallback 3574 |
| Neo4j secret | graph.neo4j_secret_reference (no committed fallback) |
| Boards directory | graph.boards_dir > fallback "graph-infra/boards" |
| Container prefix | graph.container_prefix > project.name > fallback "graph" |
Connection strings are for user-facing diagnostics only. Actual graph access is allowed only through graph tools available in the current Codex environment.
Schema And Query Locations
- BA schema:
graph-infra/schema/ba-schema.cypher - SA schema:
graph-infra/schema/sa-schema.cypher - TL schema:
graph-infra/schema/tl-schema.cypher - Queries:
graph-infra/queries/
ID Conventions (selected)
Allocate ids by scanning the existing max suffix and incrementing. Provenance and change-tracking ids:
DEC-NNN—:Decisionnode (graph-native rationale; global sequential, likeADR-NNN). Written bynacl-sa-feature,nacl-tl-fix(L2/L3), andnacl-sa-finalize. A:Decisionis NOT aRequirement— keep it a distinct label so it does not pollute the L3 requirement check or the planner.- Change-tracking properties (no new ids):
UseCase.spec_version,Task.planned_from_version, andreview_status/stale_reason/stale_since/stale_originon snapshot-bearing nodes — read withcoalesce(n.review_status,'current'). - Screen state machine ids (written by
nacl-sa-ui state-machine):SCR-{PascalName}(:Screen),SCRST-{Screen}-{State}(:ScreenState),SCREV-{Screen}-{Event}(:ScreenEvent),SCRTR-{Screen}-NNN(reified:Transition, per-screen sequential),SCREF-{Screen}-NNN(:ScreenEffect, per-screen sequential),ANEV-{Name}(:AnalyticsEvent).{Screen}is the PascalName part of the Screen id without theSCR-prefix. SLC-{NNN}-{PascalName}—:Slicenode (graph-native acceptance scenario, written bynacl-sa-uc slices).{NNN}is the parent UC number (UC-006→006);{PascalName}derives from a latin short scenario name (kebab/snake/space → PascalCase). The infix keeps repeated scenario names unique across UCs and enablesSTARTS WITH 'SLC-NNN-'scoping.ERR-{UPPER_SNAKE_CODE}—:DomainErrornode (transport-independent domain error, written bynacl-sa-uc errors; also bynacl-tl-fixwhen an L2/L3 fix uncovers a missing error branch).{UPPER_SNAKE_CODE}equals thecodeproperty — domain-prefixed (PROMO_NOT_FOUND, never bareNOT_FOUND). Module-scoped shared vocabulary ((:Module)-[:HAS_ERROR]->): MERGE by id, never duplicate per UC.ERRP-{CODE}-{PascalName}—:ErrorPresentationnode (one user-facing presentation; PascalName from the presentation kind/context). Seesa-schema.cypher§ 3-quater andsa_uc_errors.CACHE-{PascalName}—:CachePolicynode (caching policy of one server data surface, written bynacl-sa-uc resilience; also bynacl-tl-fixwhen an L2/L3 fix uncovers a missing cache/invalidation policy). PascalName from the cached surface + storage (CACHE-ResultMediaIndexedDb). Module-scoped shared vocabulary ((:Module)-[:HAS_CACHE]->): MERGE by id, never duplicate per UC.DEG-{NNN}-{PascalName}—:DegradationRulenode (one UC's degradation behavior);{NNN}is the parent UC number — the infix enablesSTARTS WITH 'DEG-NNN-'scoping, exactly likeSLC. Seesa-schema.cypher§ 3-quinquies andsa_uc_resilience.
If these files are unavailable, report:
Status: BLOCKED
Reason: graph schema or query files are unavailable in the current workspace
Capabilities
May Do
- Provide shared graph, schema, board, and verification conventions.
- Read project configuration and schema files when file access is available.
- Explain which graph tools a downstream skill needs.
Must Not Do
- Modify graph data directly.
- Modify project files.
- Select or constrain the runtime model.
- Claim graph access exists without checking tool availability.
Conditional Tools And Actions
- File reads are conditional on workspace access.
- Graph reads or writes are conditional on available graph tooling.
- Board file operations are conditional on availab