Nova explore
Skill lliangcol/llm-plugins-fusion/nova-plugin/skills/nova-explore
AI engineering workflow framework for coding assistants, delivered through nova-plugin.
npx -y skills add lliangcol/llm-plugins-fusion --skill nova-exploreAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Unified exploration Hub Skill. Route by PERSPECTIVE to observer/reviewer style outputs; analysis only, no design or implementation.
The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.
SKILL.md
9.1 KB, as published. Nobody here has run it
Shared Execution Policy
This file is the supporting behavioral contract for /nova-plugin:explore and the deprecated /nova-plugin:nova-explore compatibility entrypoint. Prefer the direct command; the compatibility name remains only for the current major-version migration window.
- Resolve natural-language and explicit
KEY=valueinputs using../_shared/parameter-resolution.md; explicit non-conflicting values take precedence. - Apply
../_shared/safety-preflight.mdbefore side effects. Never infer approval, destructive scope, credentials, or output destinations. - Follow
../_shared/output-contracts.mdand../_shared/artifact-policy.md; report completed, skipped, and blocked validation truthfully. - Respect the frontmatter tool boundary. Missing inputs, unavailable dependencies, overlapping user changes, or repository-policy conflicts are blockers rather than permission to broaden scope.
Execution
- Parse
$ARGUMENTSagainst the workflow-specific inputs below. - Read only the context required for the requested scope.
- Apply the workflow contract and its strict output format.
- Stop before unauthorized side effects; otherwise validate in proportion to risk and report residual risk.
Workflow Contract
<!-- BEGIN GENERATED BEHAVIOR CONTRACT -->Generated from
workflow-specs/behaviors.v2.json. This block is authoritative. Runnode scripts/generate-behavior-surfaces.mjs --writeafter changing the IR; if explanatory text below conflicts, fail closed.
Generated Behavior Index
- Purpose: Align understanding and identify unknowns or risks without proposing solutions.
- Canonical inputs:
INPUT(required aliases=REQUEST,CONTEXT);PERSPECTIVE(optional aliases=MODE default="observer" exact="observer","reviewer");DEPTH(optional default="standard" exact="lite","standard","deep") - Resolved variant authority:
{} normalized={"DEPTH":"standard","PERSPECTIVE":"observer"} -> runtime/contracts/explore.json;{"DEPTH":"lite","PERSPECTIVE":"observer"} normalized={"DEPTH":"lite","PERSPECTIVE":"observer"} -> runtime/contracts/explore-lite.json;{"PERSPECTIVE":"reviewer"} normalized={"DEPTH":"standard","PERSPECTIVE":"reviewer"} -> runtime/contracts/explore-review.json;{"DEPTH":"deep"} normalized={"DEPTH":"deep","PERSPECTIVE":"observer"} -> runtime/contracts/senior-explore.json. Declared selector defaults are applied before matching. An exact normalized override wins; a non-exact combination that triggers an alias specialization stops as conflicting, and only a valid combination that triggers no specialization uses the canonical fallback. The complete resolved runtime contract is authoritative and no field falls back to canonical prose. - Claude static-entrypoint gate: Native command and Skill frontmatter are static. A matching command wrapper may continue after it has verified that its invoked command id equals
resolvedWorkflowId; this canonical Skill must not re-resolve or reject that validated wrapper. Only when this canonical Skill is itself the Claude native invoked entrypoint and no validated wrapper gate exists mustresolvedWorkflowIdequalexplore. Otherwise STOP before tools or side effects and invoke the exact direct command/nova-plugin:<resolved commandEntrypoint.directCommandId>; never execute the specialized contract under unmatched canonical frontmatter. Generic and Codex adapters may execute the resolved contract directly under adapter enforcement. - Decision entries: 4; canonical routes and variants:
explore {"PERSPECTIVE":"reviewer"},explore {"PERSPECTIVE":"observer","DEPTH":"lite"},explore {},explore {"DEPTH":"deep"}. - Workflow steps:
resolve-input→route→analyze→emit - Output: mode=
chat; order=observations or what is clear→uncertainties or review questions→potential risks or risk signals; severity=none. - Deviation/failure: mode=
forbid; failure order=status→missing input→allowed values→safe next action. - Full IR:
runtime/contracts/explore.json#behaviorContractembeds the complete decision table, invariants, stops, field definitions, validation, and failure contract from the same source. Detailed guidance below may not override it.
Purpose
Quickly align understanding and identify unknowns/risks without proposing solutions.
Inputs
Resolve PERSPECTIVE and DEPTH first, then use the matched runtime
contract's required-input set:
| Resolved profile | Selector | Required inputs |
|---|---|---|
| Standard observer | {} | INPUT |
| Lite observer | {"PERSPECTIVE":"observer","DEPTH":"lite"} | INPUT |
| Reviewer | {"PERSPECTIVE":"reviewer"} | INPUT |
| Deep exploration | {"DEPTH":"deep"} | INTENT, CONTEXT |
INPUT is not required by deep exploration; that resolved contract requires
INTENT and CONTEXT.
Outputs
observer:Observations / Uncertainties / Potential risks.reviewer:What is clear / Review questions / Risk signals.- Default, lite, and reviewer profiles return chat output. The resolved
DEPTH=deepcontract may also write the senior-explore artifact when its explicitEXPORT_PATHand artifact-write approval are present.
Workflow
- Parse
PERSPECTIVEandDEPTH. - Hub routing policy:
PERSPECTIVE=observer DEPTH=standard-> canonicalnova-explorePERSPECTIVE=observer DEPTH=lite-> resolvedexplore-litecontractPERSPECTIVE=reviewer DEPTH=standard-> resolvedexplore-reviewcontractPERSPECTIVE=observer DEPTH=deep-> resolvedsenior-explorecontract
- Emit structured analysis output only.
Examples
- Natural trigger:
Use explore to quickly align on this incident report. - Explicit trigger:
explore PERSPECTIVE=reviewer INPUT="Product requirement draft".
Safety
- Do not provide solutions, implementation plans, or code.
- Separate facts from assumptions.
Detailed Contract
The quick-exploration procedure below describes the non-deep profiles. Deep
exploration follows the resolved senior-explore runtime contract and its
INTENT plus CONTEXT requirements.
QUICK EXPLORATION
You are Claude Code acting as a senior engineer / tech lead.
This command is for quick understanding and cognitive alignment, not for solving problems.
INPUT PARAMETERS
From $ARGUMENTS, extract the following:
PERSPECTIVE (Optional)
Choose the perspective for this exploration:
observer(default) → Neutral observation and fact-gatheringreviewer→ Review mindset with critical questions
If not specified, use observer perspective.
PERSPECTIVE: <PERSPECTIVE>
DEPTH (Optional)
lite→ Concise fact gatheringstandard(default) → Normal read-only explorationdeep→ Senior analysis artifact contract; requires its resolved inputs and explicit artifact-write approval
DEPTH: <DEPTH>
EXECUTION RULES
- Do NOT write code
- Do NOT propose solutions or designs
- Do NOT suggest refactors or optimizations
Focus only on:
- What is clearly understood
- What is uncertain or ambiguous
- Where risks may exist due to gaps in understanding
Keep the output concise and practical.
Language constraints:
- Avoid words like: "should", "recommend", "solution", "implement"
- Prefer: "observed", "suggests", "may indicate", "unclear", "appears"
OUTPUT FORMAT (STRICT)
Output format varies based on PERSPECTIVE:
If PERSPECTIVE = observer:
Observations
- Clearly stated facts from the input
- Direct, obvious inferences (clearly labeled)
Uncertainties
- Missing information
- Ambiguous behavior or intent
- Assumptions being made
Potential risks
- Risks caused by misunderstandings or unknowns
- No mitigation or suggestions
If PERSPECTIVE = reviewer:
What is clear
- Confirmed understanding based on provided input
- Explicitly separate facts from interpretations
Review questions
- Questions a reviewer would raise
- Focus on correctness, clarity, and assumptions
- Avoid hypothetical redesign questions
Risk signals
- Correctness risks
- Boundary or edge-case risks
- Operational or maintenance risks
- No mitigation or next steps
4.0 VARIANT PROFILES
- Default or
PERSPECTIVE=observer: normal read-only exploration. PERSPECTIVE=reviewer: formerexplore-reviewbehavior.DEPTH=lite: formerexplore-litebehavior.DEPTH=deep: formersenior-explorebehavior;EXPORT_PATHrequires explicit artifact-write scope and the shared safety preflight.
Compatibility command presets are defaults only; explicit non-conflicting user inputs win. Runtime contracts remain the complete behavior source for each variant.