name-discovery

name	name-discovery
description	Use when checking if a software/tool name is available for use — searches GitHub, web, and registries for conflicts, evaluates severity, and generates alternatives.
version	1.0.0
author	Hermes Cortex
license	MIT
metadata	{"hermes":{"tags":["naming","brand","discovery","research","naming-conventions"],"related_skills":["spike"]}}

Name Discovery

Systematically check if a software name is available for a new project, CLI tool, or brand — and find alternatives if it's not.

When to Use

• User asks "check if [name] is available"
• User asks "find a good name for [project/tool]"
• You need to name a new component, repo, or tool and want to avoid namespace collisions
• A proposed name has obvious conflicts and you need to generate alternatives

Don't use for: internal variable/function naming, repo slug naming for private repos, or codenames that will never be public-facing.

Methodology

Phase 1: Check the primary name

Run these checks in parallel (batch independent searches):

1. GitHub search — web_search(query="<name> github")

   • Look for: orgs (github.com/<name>), major repos (github.com/X/<name>), topics
   • Assess stars, activity, and recency

2. Web search — web_search(query="<name> software tool OR CLI OR framework")

   • Look for: actual projects, products, or companies using the name
   • Ignore: dictionary-word uses unrelated to software (e.g. "motif" in biology, "ballast" in shipping)

3. Package registries — web_search(query="<name> pypi"), web_search(query="<name> npm")

   • Check if the name is published on PyPI, npm, or other relevant registries

4. Domain check (if relevant) — web_search(query="<name>.ai OR <name>.dev OR <name>.io")

   • For public-facing tools, domain availability matters

Phase 2: Evaluate conflict severity

Severity	Criteria	Verdict
Major	Same space (AI agent infra, orchestration, developer tools) + active project	❌ Unusable
Moderate	Different space but well-known name (e.g. Hyundai Sonata) or adjacent space	⚠️ High confusion risk
Minor	Different space, obscure/small project, or antiquated (e.g. 90s Unix toolkit)	⚡ Usable with caveat
Clean	No significant conflicts found in any registry	✅ Fully available

Phase 3: Generate alternatives (when primary is taken)

When the primary name fails, generate alternatives anchored to the concept space:

1. Identify the core concept: what does the tool do? (orchestrate, know, guide, connect, stabilize)
2. Generate 5-10 candidates from these angles:
   • Root language — Latin/Greek roots of the concept (nexum, cantus, sensus, ballast)
   • Metaphor — analogous real-world objects (lighthouse, keystone, anchor, compass)
   • Composite — blend two concept words (cortex+text = cortext, source+sorcerer = sorcer)
   • Musical — if the concept has rhythmic/timing elements (motif, cadence, etude, fugue)
   • Shortened — truncate/extend the original (cortex → cortx, orchestrator → orchex)
3. Run Phase 1 on each candidate
4. Present top candidates ranked by availability + conceptual fit

Phase 4: Deliver the recommendation

Format the output as a ranked table:

Name	Availability	Fit	Notes
✅ ballast	Clean	Strong (stability)	Best option — no conflicts
⚠️ motif	Minor conflict	Strong (pattern)	Old Unix GUI toolkit, different space
❌ cortx	Major conflict	Strong	Seagate + AI assistant at cortx.me

Include your top pick with reasoning.

Common Pitfalls

1. Confusing "exists" with "conflict." A biology paper called "Motif" is not a software conflict. A 3.3k-star GitHub project in your exact space (agent orchestration) is. Always assess what space the existing project is in.

2. Stopping after one search. A name may be clean on GitHub but taken on PyPI, or vice versa. Check multiple registries.

3. Overlooking adjacent spaces. "cortx" conflicts aren't just Seagate's object store — there's also a self-hosted AI assistant at cortx.me doing something eerily similar. Check the user's actual niche.

4. Generating weak alternatives. "Just add a suffix" (cortx-cli, cortx-tool) doesn't solve the problem — the base name still collides. Generate conceptually distinct options.

5. Missing the obvious domain check. If the .ai or .dev domain is parked by a squatter or used by a different project, flag it.

6. Recommending a name without conceptual fit. A clean name is worthless if it doesn't evoke what the tool does. Assess fit alongside availability.

7. Spending too many rounds. If the first 3-4 candidates all fail, you're in a crowded naming space. Present the best of what you found and recommend the user pick rather than searching 20 more.