From nextc-product
Product exploration — raw idea to validated proposal with market research, competitor analysis, and collision insights. Supports --auto, --fast, --quick, --no-collision, --update, --branch, --deep-dive, --export.
How this skill is triggered — by the user, by Claude, or both
Slash command
/nextc-product:product-exploreThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Product exploration pipeline that turns raw ideas into validated proposals.
Product exploration pipeline that turns raw ideas into validated proposals. Acts as a strategic consultant — interprets data, recommends actions, and proactively suggests next steps.
Architecture:
product-explorer agent: thin orchestrator (mode dispatch, checkpoints, state)product-researcher agent: Phase 2 research (facts + hypotheses)product-shaper agent: Phase 4 creative (vision, naming, canvas, positioning)product-stress-tester agent: Phase 5 adversarial (assumptions, experiments)product-collision-analyst agent: Phase 5.5 cross-referencing (Opus)Parse flags from $ARGUMENTS before mode detection. Flags are combinable.
flags = {
auto: $ARGUMENTS contains "--auto",
quick: $ARGUMENTS contains "--quick",
no_collision: $ARGUMENTS contains "--no-collision"
}
# Strip flags from $ARGUMENTS before mode detection
idea_text = $ARGUMENTS with all --flags removed
After parsing, check for conflicts. Report and STOP if invalid:
| Conflict | Why | Resolution |
|---|---|---|
--fast + --quick | Both modify pipeline depth, incompatible | "Use --fast for 3-phase gut check, or --quick for full pipeline without stress test. Not both." |
--no-collision + --update | Update mode's value IS re-running collision | "Update mode always runs collision analysis. Use --update without --no-collision." |
--fast + --no-collision | Fast mode never runs collision anyway | Silently ignore --no-collision (no error, just redundant) |
--quick + --no-collision | Valid combo | Allow — skip stress test AND collision |
After stripping flags, detect mode from remaining arguments:
| Argument | Mode | Action |
|---|---|---|
--fast or --fast [idea] | Fast | 3-phase gut check (~50K tokens) |
--update | Update | Feed experiment results into existing proposal |
--branch "description" | Branch | Fork exploration into alternative direction |
--deep-dive "topic" | Deep Dive | Focused research on one topic |
--export pitch | Export | Reformat proposal as pitch deck outline |
| (anything else) | Deep (default) | Full adaptive pipeline (~150K tokens) |
| Combo | Use Case | Est. Tokens |
|---|---|---|
--auto | Rich brief, full depth, minimal pauses | ~130K |
--auto --fast | Weekly screening (3-4 ideas) | ~40K |
--auto --quick | Fast with collision insights | ~90K |
--auto --quick --no-collision | Fastest full proposal | ~70K |
--quick | Interactive but skip stress test | ~110K |
Before spawning the agent, validate ALL dependencies. Run ALL checks before reporting — never fail one-at-a-time.
Check that these plugin directories exist in the cache:
pm-skills (6 sub-plugins):
~/.claude/plugins/cache/pm-skills/
Provides: user-personas, market-sizing, competitor-analysis, job-stories,
pre-mortem, beachhead-segment, product-vision, value-proposition, lean-canvas,
identify-assumptions-new, prioritize-assumptions, brainstorm-experiments-new,
positioning-ideas
marketingskills:
~/.claude/plugins/cache/marketingskills/
Provides: customer-research
nextc-claude (bundled):
clarify — always available since it's bundled with this pluginls ~/.claude/plugins/cache/pm-skills/ 2>/dev/null
ls ~/.claude/plugins/cache/marketingskills/ 2>/dev/null
mcp__plugin_everything-claude-code_exa__web_search_exa is accessible.
If absent, warn: market sizing will be skipped, fact-based research will be degraded.~/.claude/plugins/cache/everything-claude-code/.
Not used by product-explore directly, but the next step (/flutter-kickoff or /feature-dev)
requires it. Warn early.Report ALL missing items at once with exact install commands:
Preflight FAILED. Missing dependencies:
PLUGINS:
- pm-skills — Install:
/plugin marketplace add phuryn/pm-skills
/plugin install pm-market-research@pm-skills
/plugin install pm-execution@pm-skills
/plugin install pm-go-to-market@pm-skills
/plugin install pm-product-strategy@pm-skills
/plugin install pm-product-discovery@pm-skills
/plugin install pm-marketing-growth@pm-skills
- marketingskills — Install:
/plugin marketplace add coreyhaines31/marketingskills
/plugin install marketing-skills@marketingskills
OPTIONAL:
- everything-claude-code not detected — your next step (/flutter-kickoff or /feature-dev) needs it.
/plugin marketplace add affaan-m/everything-claude-code
/plugin install everything-claude-code@everything-claude-code
- Exa MCP not detected — market research will be limited.
Install missing dependencies and run /product-explore again.
STOP here. Do NOT proceed to the agent until all required plugins are installed.
Preflight passed.
Plugins: pm-skills ✓, marketingskills ✓
Optional: everything-claude-code [✓/missing], Exa MCP [available/unavailable]
Skills: [N] available
Flags: auto=[yes/no], quick=[yes/no], no-collision=[yes/no]
After preflight passes, show the estimated cost and time for the selected mode + flags:
Estimated run:
Mode: [deep/fast/update/branch/deep-dive/export]
Flags: [auto, quick, no-collision — or none]
Tokens: ~[N]K
Time: ~[N] min
Opus calls: [N] (collision analysis[, vision synthesis])
| Mode + Flags | Tokens | Time | Opus Calls |
|---|---|---|---|
| deep | ~150K | ~20 min | 2 (collision + vision) |
| deep --auto | ~130K | ~15 min | 2 |
| deep --quick | ~110K | ~15 min | 1 (collision only) |
| deep --auto --quick | ~90K | ~12 min | 1 |
| deep --auto --quick --no-collision | ~70K | ~10 min | 0 |
| fast | ~50K | ~8 min | 0 |
| fast --auto | ~40K | ~6 min | 0 |
| update | ~60K | ~10 min | 1 (collision) |
| branch | ~80K | ~12 min | 1 (collision) |
| deep-dive | ~20K | ~5 min | 0 |
| export | ~10K | ~3 min | 0 |
After showing the estimate, spawn the product-explorer orchestrator:
Agent(
subagent_type: "nextc-product:product-explorer",
prompt: """
Mode: [deep/fast/update/branch/deep-dive/export]
Flags: auto=[yes/no], quick=[yes/no], no-collision=[yes/no]
Arguments: $ARGUMENTS (with flags stripped)
Exa available: [yes/no]
Working directory: [current project root]
Template path: ~/.claude/plugins/cache/nextc-claude/nextc-product/skills/product-explore/templates/proposal.md
[For update/branch/deep-dive/export modes:]
Existing exploration: docs/explore/ exists — read it for prior state.
[For deep/fast modes:]
User's idea: [the idea text from $ARGUMENTS]
"""
)
| Mode | Precondition | On Failure |
|---|---|---|
--update | docs/explore/ must exist | "No existing exploration found. Run /product-explore first." |
--branch | docs/explore/ must exist | "No existing exploration found. Run /product-explore first." |
--deep-dive | docs/explore/ must exist | "No existing exploration found. Run /product-explore first." |
--export pitch | docs/proposal.md must exist | "No proposal found. Run /product-explore first." |
--fast | None | Proceed |
| (default) | None | Proceed |
npx claudepluginhub nextc/nextc-claude --plugin nextc-productGuides greenfield product ideas from raw concept to structured product-concept.md via conversation, capturing vision, core experience, target users, trust model, platforms, and scope.
Provides Chief Product Officer expertise with frameworks like Jobs-to-be-Done, Lean Startup, and Hook Model for product strategy, market analysis, competitive positioning, feature prioritization, and business models.
Spawns three AI agents to run validation sprint on product ideas: researches user problems, market opportunities, risks; delivers scored build/no-build verdict.