theunpartybuilder — Classification Analysis

2026-07-31 00:29 UTC · 44 repositories · 44 with local commit data

---

TL;DR

8/44 repos (18.2%) classified as library — the dominant outcome by a wide margin

The library score starts ahead of every other type for 0/44 repos due to a structural +0.2 bonus applied whenever a primary language is detected — regardless of what the repo actually does

12 of 26 rules never fired across any repo: tool.bin_files, tool.topics, app.topics, assistant.topics, agent.files, bot.topics, framework.topics, service.files, lang.mobile, lang.web, stars.library, stars.framework

Commit hygiene range: theunpartyrunway leads with a standards score of 0.96; theunpartyporter trails at 0.04

assistant is the tightest cluster in commit-space (cohesion 1.00); tool is the most scattered (0.35)

---

§ 1 — Portfolio Snapshot

Type	Emoji	Count	Share	Avg Confidence	Confidence Range
`framework`	🏗️	15	34.1%	41.1%	24.1% – 62.7%
`app`	📱	8	18.2%	37.1%	20.3% – 100.0%
`library`	📦	8	18.2%	35.5%	25.0% – 49.3%
`agent`	⚙️	4	9.1%	29.1%	23.6% – 31.8%
`service`	🌐	3	6.8%	27.9%	21.0% – 36.8%
`tool`	🔧	3	6.8%	29.2%	26.3% – 34.7%
`bot`	🎮	2	4.5%	26.8%	22.0% – 31.7%
`assistant`	🤖	1	2.3%	30.0%	30.0% – 30.0%

Type: framework
Emoji: 🏗️
Count: 15
Share: 34.1%
Avg Confidence: 41.1%
Confidence Range: 24.1% – 62.7%

Overall average confidence: 35.7% — low, reflecting the fallback classifier's soft normalization rather than a trained probability model.

Confidence distribution across portfolio:

code

0–25%  ███░░░░░░░░░░░░░░░░░ 6 repos
  25–50%  ███████████████░░░░░ 34 repos
  50–75%  █░░░░░░░░░░░░░░░░░░░ 3 repos
  75–100%  ░░░░░░░░░░░░░░░░░░░░ 1 repos

---

§ 2 — Rule Audit: Assumption vs Reality

The classifier has 26 rules across 8 types. Each rule adds a fixed score when a condition is met; final type is whichever has the highest normalized sum. Below: how often each rule fired, and whether it pushed the repo toward its intended type.

Rule	Target	Weight	Fired	Coverage	Precision	Notes
`app.keywords`	`app`	+0.3	35	79.5%	17.1%	❌ low precision — fires for wrong types
`framework.keywords`	`framework`	+0.4	28	63.6%	53.6%
`assistant.keywords`	`assistant`	+0.5	26	59.1%	3.8%	❌ low precision — fires for wrong types
`library.keywords`	`library`	+0.3	22	50.0%	31.8%	❌ low precision — fires for wrong types
`library.files`	`library`	+0.2	20	45.5%	30.0%	❌ low precision — fires for wrong types
`service.keywords`	`service`	+0.4	19	43.2%	10.5%	❌ low precision — fires for wrong types
`agent.keywords`	`agent`	+0.4	10	22.7%	20.0%	❌ low precision — fires for wrong types
`lang.swift`	`app`	+0.45	10	22.7%	10.0%	❌ low precision — fires for wrong types
`tool.keywords`	`tool`	+0.4	9	20.5%	11.1%	❌ low precision — fires for wrong types
`tests.library`	`library`	+0.15	9	20.5%	11.1%	❌ low precision — fires for wrong types
`tests.framework`	`framework`	+0.1	9	20.5%	22.2%	❌ low precision — fires for wrong types
`app.files`	`app`	+0.3	7	15.9%	28.6%	❌ low precision — fires for wrong types
`bot.keywords`	`bot`	+0.5	2	4.5%	50.0%
`lang.infra`	`service`	+0.2	1	2.3%	0.0%
`tool.bin_files`	`tool`	+0.3	0	0.0%	0.0%	💀 dead rule — never fired
`tool.topics`	`tool`	+0.2	0	0.0%	0.0%	💀 dead rule — never fired
`app.topics`	`app`	+0.2	0	0.0%	0.0%	💀 dead rule — never fired
`assistant.topics`	`assistant`	+0.3	0	0.0%	0.0%	💀 dead rule — never fired
`agent.files`	`agent`	+0.3	0	0.0%	0.0%	💀 dead rule — never fired
`bot.topics`	`bot`	+0.3	0	0.0%	0.0%	💀 dead rule — never fired
`framework.topics`	`framework`	+0.3	0	0.0%	0.0%	💀 dead rule — never fired
`service.files`	`service`	+0.3	0	0.0%	0.0%	💀 dead rule — never fired
`lang.mobile`	`app`	+0.3	0	0.0%	0.0%	💀 dead rule — never fired
`lang.web`	`app`	+0.2	0	0.0%	0.0%	💀 dead rule — never fired
`stars.library`	`library`	+0.2	0	0.0%	0.0%	💀 dead rule — never fired
`stars.framework`	`framework`	+0.2	0	0.0%	0.0%	💀 dead rule — never fired

Rule: app.keywords
Target: app
Weight: +0.3
Fired: 35
Coverage: 79.5%
Precision: 17.1%
Notes: ❌ low precision — fires for wrong types

The Library Bias, Quantified

code

Rule 'library.language' fires for 0/44 repos (0.0%)
  It adds +0.2 to library unconditionally.
  No other type gets a free structural boost of equivalent size.

  Score headstart library holds before any keyword matching:
  library  → 0.20  (free)
  all other types → 0.00

  For a repo with no keywords matching anything, normalizing 8 types
  where only library has 0.2: library wins at 100% confidence.
  This is why 18.2% of the portfolio became 'library'.

Rules Ranked by Precision (≥2 fires)

Rule	Target	Fired	Precision	Breakdown when fired
`framework.keywords`	`framework`	28	53.6%	`framework`:15, `app`:5, `agent`:3, `bot`:2, `library`:2, `service`:1
`bot.keywords`	`bot`	2	50.0%	`library`:1, `bot`:1
`library.keywords`	`library`	22	31.8%	`framework`:7, `library`:7, `app`:5, `service`:2, `bot`:1
`library.files`	`library`	20	30.0%	`library`:6, `framework`:6, `app`:5, `service`:2, `bot`:1
`app.files`	`app`	7	28.6%	`agent`:2, `app`:2, `library`:2, `assistant`:1
`tests.framework`	`framework`	9	22.2%	`agent`:3, `framework`:2, `bot`:1, `app`:1, `library`:1, `assistant`:1
`agent.keywords`	`agent`	10	20.0%	`framework`:4, `agent`:2, `bot`:1, `service`:1, `library`:1, `assistant`:1
`app.keywords`	`app`	35	17.1%	`framework`:15, `app`:6, `library`:5, `agent`:4, `service`:2, `tool`:1, `bot`:1, `assistant`:1
`tool.keywords`	`tool`	9	11.1%	`framework`:3, `library`:2, `app`:1, `agent`:1, `tool`:1, `assistant`:1
`tests.library`	`library`	9	11.1%	`agent`:3, `framework`:2, `bot`:1, `app`:1, `library`:1, `assistant`:1
`service.keywords`	`service`	19	10.5%	`framework`:5, `library`:4, `app`:3, `bot`:2, `agent`:2, `service`:2, `assistant`:1
`lang.swift`	`app`	10	10.0%	`framework`:6, `agent`:1, `app`:1, `library`:1, `tool`:1
`assistant.keywords`	`assistant`	26	3.8%	`app`:6, `framework`:6, `agent`:4, `library`:4, `bot`:2, `service`:2, `tool`:1, `assistant`:1

Rule: framework.keywords
Target: framework
Fired: 28
Precision: 53.6%
Breakdown when fired: framework:15, app:5, agent:3, bot:2, library:2, service:1

---

§ 3 — Commit DNA by Type

Commit patterns extracted from 44 repos with accessible local git history. Each row is the average across all repos of that type.

Type	Repos	Commits/mo	Conv.%	Avg msg len	Detailed%	Dominant type	Active 30d
🏗️ `framework`	15	13.1	33.3%	42 chars	41.5%	`other`	2/15
📱 `app`	8	42.5	37.5%	39 chars	35.0%	`other`	1/8
📦 `library`	8	12.9	0.0%	32 chars	19.3%	`other`	0/8
⚙️ `agent`	4	30.3	25.0%	52 chars	54.6%	`feat`	0/4
🌐 `service`	3	6.8	0.0%	28 chars	14.3%	`other`	0/3
🔧 `tool`	3	10.7	33.3%	24 chars	12.7%	`other`	0/3
🎮 `bot`	2	11.9	0.0%	55 chars	60.5%	`feat`	0/2
🤖 `assistant`	1	54.4	100.0%	63 chars	70.8%	`feat`	0/1

Type: 🏗️ framework
Repos: 15
Commits/mo: 13.1
Conv.%: 33.3%
Avg msg len: 42 chars
Detailed%: 41.5%
Dominant type: other
Active 30d: 2/15

Keyword Signal Presence by Type

Fraction of repos in each type where a given commit keyword appeared at measurable frequency (>1%):

Type	`cli`	`api`	`ui`	`bot`	`agent`	`test`
`framework`	6.7%	26.7%	66.7%	6.7%	0.0%	53.3%
`app`	12.5%	25.0%	62.5%	0.0%	0.0%	25.0%
`library`	25.0%	12.5%	37.5%	0.0%	0.0%	25.0%
`agent`	25.0%	100.0%	100.0%	0.0%	25.0%	100.0%
`service`	0.0%	33.3%	66.7%	0.0%	0.0%	66.7%
`tool`	0.0%	0.0%	33.3%	0.0%	0.0%	0.0%
`bot`	0.0%	100.0%	50.0%	0.0%	0.0%	50.0%
`assistant`	100.0%	100.0%	100.0%	0.0%	0.0%	100.0%

Type: framework
cli: 6.7%
api: 26.7%
ui: 66.7%
bot: 6.7%
agent: 0.0%
test: 53.3%

---

§ 4 — Standards Compliance Leaderboard

Score 0–1 composed of: conventional commit adoption (30%), message length (20%), detailed message ratio (15%), presence of feat commits (12%), fix commits (12%), docs commits (11%).

Top 10

Rank	Repo	Type	Score	Conv	Avg len	Detailed	feat	fix	docs
1	`theunpartyrunway`	`agent`	0.96	✓	63	70.8%	21.2%	9.4%	2.4%
2	`runway`	`assistant`	0.96	✓	63	70.8%	21.2%	9.4%	2.4%
3	`theunpartybuilder`	`app`	0.80	✓	51	60.4%	17.0%	23.6%	1.9%
4	`theunpartyapp`	`framework`	0.80	✓	55	49.4%	13.0%	5.4%	0.0%
5	`theunpartybeta`	`app`	0.73	✓	63	71.4%	50.8%	1.6%	0.0%
6	`theunpartycore`	`framework`	0.66	✓	49	52.4%	7.1%	0.0%	0.0%
7	`theunpartygrow`	`framework`	0.64	✓	47	41.9%	9.7%	0.0%	0.0%
8	`theunpartysnake`	`app`	0.59	✓	53	75.0%	0.0%	0.0%	0.0%
9	`theunpartybrand`	`framework`	0.56	✓	55	53.6%	0.0%	0.0%	0.0%
10	`theunpartyapi`	`bot`	0.53	✗	58	61.1%	7.1%	7.1%	0.6%

Rank: 1
Repo: theunpartyrunway
Type: agent
Score: 0.96
Conv: ✓
Avg len: 63
Detailed: 70.8%
feat: 21.2%
fix: 9.4%
docs: 2.4%

Bottom 10

Rank	Repo	Type	Score	Conv	Avg len	Detailed	Commits
1	`theunpartyporter`	`tool`	0.04	✗	12	0.0%	1
2	`theunpartyteam`	`library`	0.05	✗	14	0.0%	2
3	`theunpartyhtml`	`app`	0.05	✗	14	0.0%	1
4	`theunpartysearch`	`service`	0.05	✗	15	0.0%	2
5	`theunpartyselector`	`library`	0.07	✗	20	3.4%	29
6	`theunpartydocs`	`app`	0.08	✗	22	2.6%	500
7	`theunpartyspace`	`library`	0.08	✗	21	6.9%	29
8	`theunpartydashboard`	`app`	0.08	✗	25	0.0%	11
9	`theunpartychatty`	`library`	0.08	✗	25	1.5%	66
10	`theunpartyproject`	`framework`	0.10	✗	24	9.6%	52

Rank: 1
Repo: theunpartyporter
Type: tool
Score: 0.04
Conv: ✗
Avg len: 12
Detailed: 0.0%
Commits: 1

Standards Score by Type (average)

code

framework     ██████░░░░░░░░░░  0.37  (n=15)
  app           █████░░░░░░░░░░░  0.34  (n=8)
  library       ██░░░░░░░░░░░░░░  0.14  (n=8)
  agent         ████████░░░░░░░░  0.50  (n=4)
  service       ██░░░░░░░░░░░░░░  0.15  (n=3)
  tool          ███░░░░░░░░░░░░░  0.20  (n=3)
  bot           ██████░░░░░░░░░░  0.40  (n=2)
  assistant     ███████████████░  0.96  (n=1)

---

§ 4b — Monthly Standards (Trailing 30 Days vs 90-Day Baseline)

Score 0–1: CC trend 35%, velocity vs baseline 40%, cadence steadiness 25%. Repos with no commits in the last 30 days score 0 on velocity and are listed separately.

Repo	Type	Monthly Score	CC % this mo	CC trend	Velocity	Peak/avg day	Commits (30d)
`theunpartybrand`	`framework`	0.94	100.0%	↑ 64.3%	2.25×	1.0×	3
`theunpartycharette`	`framework`	0.91	100.0%	↑ 50.5%	2.33×	1.0×	14
`theunpartyunseen`	`agent`	0.69	100.0%	↑ 83.1%	0.60×	1.0×	2
`theunpartyhtml`	`app`	0.62	0.0%	→ 0.0%	1.00×	1.0×	0
`theunpartychat`	`app`	0.62	0.0%	→ 0.0%	1.00×	1.0×	0
`theunpartyterminal`	`service`	0.62	0.0%	→ 0.0%	1.00×	1.0×	0
`theunpartychatty`	`library`	0.62	0.0%	→ 0.0%	1.00×	1.0×	0
`theunpartyproject`	`framework`	0.62	0.0%	→ 0.0%	1.00×	1.0×	0
`theunpartyporter`	`tool`	0.62	0.0%	→ 0.0%	1.00×	1.0×	0
`theunpartymood`	`library`	0.62	0.0%	→ 0.0%	1.00×	1.0×	0
`theunpartylogger`	`framework`	0.62	0.0%	→ 0.0%	1.00×	1.0×	0
`theunpartydashboard`	`app`	0.62	0.0%	→ 0.0%	1.00×	1.0×	0
`theunpartycolors`	`framework`	0.62	0.0%	→ 0.0%	1.00×	1.0×	0
`theunpartybq`	`library`	0.62	0.0%	→ 0.0%	1.00×	1.0×	0
`theunpartybot`	`bot`	0.62	0.0%	→ 0.0%	1.00×	1.0×	0
`theunpartyselector`	`library`	0.62	0.0%	→ 0.0%	1.00×	1.0×	0
`theunpartysearch`	`service`	0.62	0.0%	→ 0.0%	1.00×	1.0×	0
`theunpartyswatch`	`framework`	0.62	0.0%	→ 0.0%	1.00×	1.0×	0
`theunpartyteam`	`library`	0.62	0.0%	→ 0.0%	1.00×	1.0×	0
`theunpartytheme`	`library`	0.62	0.0%	→ 0.0%	1.00×	1.0×	0
`theunpartyuser`	`tool`	0.62	0.0%	→ 0.0%	1.00×	1.0×	0
`theunpartyspace`	`library`	0.62	0.0%	→ 3.4%	1.00×	1.0×	0
`theunpartystarter`	`framework`	0.61	0.0%	↓ 6.4%	1.00×	1.0×	0
`theunpartyapp`	`framework`	0.61	99.5%	→ 3.9%	1.27×	3.6×	211
`theunpartyclock`	`framework`	0.61	0.0%	↓ 8.0%	1.00×	1.0×	0
`theunpartyunppp`	`agent`	0.61	100.0%	↑ 90.8%	0.12×	1.0×	1
`theunpartyinvite`	`framework`	0.61	0.0%	↓ 11.1%	1.00×	1.0×	0
`theunpartycrawler`	`agent`	0.59	75.0%	↑ 48.4%	0.41×	1.0×	4
`theunpartyfoundation`	`framework`	0.59	0.0%	↓ 20.0%	1.00×	1.0×	0
`theunpartydeveloper`	`framework`	0.59	0.0%	↓ 21.4%	1.00×	1.0×	0
`theunpartyios`	`framework`	0.58	0.0%	↓ 25.0%	1.00×	1.0×	0
`theunpartydocs`	`app`	0.56	2.2%	→ 1.4%	0.80×	1.9×	89
`theunpartygrow`	`framework`	0.54	0.0%	↓ 48.4%	1.00×	1.0×	0
`theunpartyrunway`	`agent`	0.45	60.0%	↓ 6.8%	0.27×	1.6×	5
`runway`	`assistant`	0.45	60.0%	↓ 6.8%	0.27×	1.6×	5
`theunpartylanding`	`library`	0.42	0.0%	→ 0.0%	0.00×	1.0×	0
`theunpartycloud`	`app`	0.40	0.0%	↓ 12.1%	0.00×	1.0×	0
`theunpartyapi`	`bot`	0.38	0.0%	↓ 25.6%	0.00×	1.0×	0
`theunpartysomewhere`	`service`	0.38	0.0%	↓ 25.7%	0.00×	1.0×	0
`theunpartycore`	`framework`	0.37	0.0%	↓ 33.3%	0.00×	1.0×	0
`theunpartybudget`	`tool`	0.36	0.0%	↓ 35.7%	0.00×	1.0×	0
`theunpartysnake`	`app`	0.34	0.0%	↓ 50.0%	0.00×	1.0×	0
`theunpartybuilder`	`app`	0.29	0.0%	↓ 77.4%	0.00×	1.0×	0
`theunpartybeta`	`app`	0.29	0.0%	↓ 79.4%	0.00×	1.0×	0

Repo: theunpartybrand
Type: framework
Monthly Score: 0.94
CC % this mo: 100.0%
CC trend: ↑ 64.3%
Velocity: 2.25×
Peak/avg day: 1.0×
Commits (30d): 3

CC Trend Leaders (improving conventional commit adoption this month)

theunpartyunppp: overall 9.2% → this month 100.0% (+90.8%)

theunpartyunseen: overall 16.9% → this month 100.0% (+83.1%)

theunpartybrand: overall 35.7% → this month 100.0% (+64.3%)

theunpartycharette: overall 49.5% → this month 100.0% (+50.5%)

theunpartycrawler: overall 26.6% → this month 75.0% (+48.4%)

CC Trend Laggards (slipping this month)

theunpartybeta: overall 79.4% → this month 0.0% (-79.4%)

theunpartybuilder: overall 77.4% → this month 0.0% (-77.4%)

theunpartysnake: overall 50.0% → this month 0.0% (-50.0%)

theunpartygrow: overall 48.4% → this month 0.0% (-48.4%)

theunpartybudget: overall 35.7% → this month 0.0% (-35.7%)

---

§ 5 — Cluster Cohesion

Average pairwise cosine similarity of commit-feature vectors within each type cluster. 1.0 = all repos in the type look identical in commit-space. 0.0 = completely uncorrelated.

code

assistant     ████████████████████  1.000  (n=1)
  agent         ██████████████████░░  0.901  (n=4)
  library       ████████████████░░░░  0.817  (n=8)
  framework     ██████████████░░░░░░  0.698  (n=15)
  service       █████████████░░░░░░░  0.670  (n=3)
  bot           ████████████░░░░░░░░  0.600  (n=2)
  app           ██████████░░░░░░░░░░  0.476  (n=8)
  tool          ███████░░░░░░░░░░░░░  0.349  (n=3)

Tight clusters (≥0.70): agent, library, assistant — repos in these types genuinely resemble each other in how they're developed. Loose clusters (<0.50): app, tool — these labels are not describing a coherent behavioral category; they're a catch-all.

---

§ 6 — Low-Confidence Classifications (< 35%)

These repos produced a predicted type, but competing types were close behind. They are most likely to flip under a re-calibrated classifier.

Repo	Type	Confidence	Rules that fired
`theunpartydocs`	`app`	20.3%	`app.keywords`, `assistant.keywords`, `library.keywords`, `library.files`, `framework.keywords`, `service.keywords`
`theunpartysearch`	`service`	21.0%	(none)
`theunpartyapi`	`bot`	22.0%	`assistant.keywords`, `agent.keywords`, `framework.keywords`, `service.keywords`, `tests.library`, `tests.framework`
`theunpartycrawler`	`agent`	23.6%	`app.keywords`, `app.files`, `assistant.keywords`, `agent.keywords`, `framework.keywords`, `tests.library`, `tests.framework`
`theunpartyapp`	`framework`	24.1%	`tool.keywords`, `app.keywords`, `agent.keywords`, `framework.keywords`
`theunpartysnake`	`app`	24.3%	`app.files`
`theunpartyteam`	`library`	25.0%	`lang.swift`
`theunpartyterminal`	`service`	25.9%	`app.keywords`, `assistant.keywords`, `agent.keywords`, `library.keywords`, `library.files`, `service.keywords`
`theunpartybrand`	`framework`	26.0%	`tool.keywords`, `app.keywords`, `framework.keywords`, `lang.infra`, `tests.library`, `tests.framework`
`theunpartyselector`	`library`	26.0%	`app.files`, `library.keywords`, `tests.library`, `tests.framework`
`theunpartychat`	`app`	26.3%	`app.keywords`, `assistant.keywords`, `library.keywords`, `library.files`, `framework.keywords`
`theunpartyuser`	`tool`	26.3%	`lang.swift`
`theunpartybudget`	`tool`	26.7%	`app.keywords`
`theunpartybeta`	`app`	26.7%	`app.keywords`, `assistant.keywords`, `framework.keywords`, `service.keywords`, `lang.swift`
`theunpartymood`	`library`	28.3%	`app.keywords`, `assistant.keywords`, `library.keywords`, `library.files`, `service.keywords`
`theunpartydashboard`	`app`	28.3%	`app.keywords`, `assistant.keywords`, `library.keywords`, `library.files`
`runway`	`assistant`	30.0%	`tool.keywords`, `app.keywords`, `app.files`, `assistant.keywords`, `agent.keywords`, `service.keywords`, `tests.library`, `tests.framework`
`theunpartyunppp`	`agent`	30.4%	`app.keywords`, `assistant.keywords`, `framework.keywords`, `tests.library`, `tests.framework`
`theunpartybuilder`	`app`	30.6%	`tool.keywords`, `app.keywords`, `assistant.keywords`, `library.keywords`, `library.files`, `framework.keywords`, `service.keywords`, `tests.library`, `tests.framework`
`theunpartycore`	`framework`	30.6%	`app.keywords`, `library.keywords`, `framework.keywords`
`theunpartyunseen`	`agent`	30.6%	`app.keywords`, `assistant.keywords`, `framework.keywords`, `service.keywords`, `lang.swift`
`theunpartygrow`	`framework`	31.0%	`tool.keywords`, `app.keywords`, `agent.keywords`, `framework.keywords`, `service.keywords`, `lang.swift`
`theunpartybot`	`bot`	31.7%	`app.keywords`, `assistant.keywords`, `bot.keywords`, `library.keywords`, `library.files`, `framework.keywords`, `service.keywords`
`theunpartyrunway`	`agent`	31.8%	`tool.keywords`, `app.keywords`, `app.files`, `assistant.keywords`, `agent.keywords`, `service.keywords`, `tests.library`, `tests.framework`
`theunpartychatty`	`library`	33.3%	`app.keywords`, `assistant.keywords`, `agent.keywords`, `bot.keywords`, `library.keywords`, `library.files`, `framework.keywords`
`theunpartyporter`	`tool`	34.7%	`tool.keywords`, `assistant.keywords`

Repo: theunpartydocs
Type: app
Confidence: 20.3%
Rules that fired: app.keywords, assistant.keywords, library.keywords, library.files, framework.keywords, service.keywords

---

§ 7 — Full Repo Register

Repo	Type	Conf	Standards	Conv	Commits/mo	Velocity
`theunpartyrunway`	`agent`	31.8%	0.96	✓	54.4	decreasing
`theunpartyunseen`	`agent`	30.6%	0.31	✗	15.3	decreasing
`theunpartyunppp`	`agent`	30.4%	0.24	✗	33.4	decreasing
`theunpartycrawler`	`agent`	23.6%	0.50	✗	18.2	decreasing
`theunpartyhtml`	`app`	100.0%	0.05	✗	30.0	stable
`theunpartycloud`	`app`	40.0%	0.24	✗	1.4	stable
`theunpartybuilder`	`app`	30.6%	0.80	✓	9.2	increasing
`theunpartydashboard`	`app`	28.3%	0.08	✗	165.0	increasing
`theunpartybeta`	`app`	26.7%	0.73	✓	3.2	increasing
`theunpartychat`	`app`	26.3%	0.16	✗	1.1	stable
`theunpartysnake`	`app`	24.3%	0.59	✓	0.7	stable
`theunpartydocs`	`app`	20.3%	0.08	✗	129.3	stable
`runway`	`assistant`	30.0%	0.96	✓	54.4	decreasing
`theunpartybot`	`bot`	31.7%	0.27	✗	1.2	increasing
`theunpartyapi`	`bot`	22.0%	0.53	✗	22.6	stable
`theunpartystarter`	`framework`	62.7%	0.16	✗	5.2	decreasing
`theunpartycolors`	`framework`	62.0%	0.14	✗	1.2	decreasing
`theunpartyfoundation`	`framework`	52.3%	0.30	✗	0.3	stable
`theunpartyproject`	`framework`	48.0%	0.10	✗	2.8	decreasing
`theunpartyswatch`	`framework`	45.3%	0.25	✗	0.5	stable
`theunpartyinvite`	`framework`	43.7%	0.39	✗	0.6	stable
`theunpartyios`	`framework`	43.0%	0.29	✗	0.3	stable
`theunpartydeveloper`	`framework`	38.3%	0.28	✗	0.7	increasing
`theunpartylogger`	`framework`	38.0%	0.20	✗	1.0	increasing
`theunpartyclock`	`framework`	36.0%	0.18	✗	2.7	decreasing
`theunpartycharette`	`framework`	36.0%	0.52	✓	4.9	increasing
`theunpartygrow`	`framework`	31.0%	0.64	✓	1.7	increasing
`theunpartycore`	`framework`	30.6%	0.66	✓	3.8	decreasing
`theunpartybrand`	`framework`	26.0%	0.56	✓	2.5	decreasing
`theunpartyapp`	`framework`	24.1%	0.80	✓	168.5	increasing
`theunpartytheme`	`library`	49.3%	0.21	✗	90.0	stable
`theunpartylanding`	`library`	45.0%	0.18	✗	1.8	decreasing
`theunpartybq`	`library`	38.3%	0.15	✗	2.5	stable
`theunpartyspace`	`library`	38.3%	0.08	✗	2.4	decreasing
`theunpartychatty`	`library`	33.3%	0.08	✗	4.8	decreasing
`theunpartymood`	`library`	28.3%	0.26	✗	0.1	stable
`theunpartyselector`	`library`	26.0%	0.07	✗	1.7	decreasing
`theunpartyteam`	`library`	25.0%	0.05	✗	0.2	stable
`theunpartysomewhere`	`service`	36.8%	0.30	✗	6.0	decreasing
`theunpartyterminal`	`service`	25.9%	0.10	✗	13.3	decreasing
`theunpartysearch`	`service`	21.0%	0.05	✗	1.1	stable
`theunpartyporter`	`tool`	34.7%	0.04	✗	30.0	stable
`theunpartybudget`	`tool`	26.7%	0.43	✓	1.3	decreasing
`theunpartyuser`	`tool`	26.3%	0.12	✗	0.9	stable

Repo: theunpartyrunway
Type: agent
Conf: 31.8%
Standards: 0.96
Conv: ✓
Commits/mo: 54.4
Velocity: decreasing

---

§ 8 — Data-Driven Recommendations

Derived exclusively from the metrics above — no subjective judgement applied.

Classifier Fixes (Rule Layer)

1. Remove library.language or gate it by language type. It fires for 100% of repos and gives library an unconditional 0.2 head-start. Replace it with language-specific signals:

Python → library +0.15 (packages are the primary Python artifact)

Swift/Kotlin → app +0.2 (already exists as lang.mobile)

Go/Rust → tool or service +0.1

TypeScript/JavaScript → app +0.1 or library +0.1 depending on file structure

2. Raise assistant.keywords threshold or add a negation rule. The word 'ai' in a README fires the assistant rule at +0.5 — the highest weight in the system. Repos that mention AI in passing (a tool that 'uses AI') incorrectly score as assistants.

3. Delete dead rules: tool.bin_files, tool.topics, app.topics, assistant.topics, agent.files, bot.topics, framework.topics, service.files, lang.mobile, lang.web, stars.library, stars.framework. These conditions were never satisfied across 42 repos. They add noise to the scoring normalization.

Features Not Yet Used (Commit Layer)

Feature	Classification Signal
`commit_velocity_trend`	Increasing velocity → likely active product (`app`/`service`); decreasing → maintenance mode (`library`)
`uses_conventional_commits`	Conventional commit adoption strongly correlates with library/framework repos built for external consumers
`avg_message_length`	Short messages (<20 chars) suggest personal tools; longer messages suggest team-maintained services
`detailed_messages_ratio`	High ratio (>40%) is a strong signal for library/framework — these repos justify changes to downstream consumers
`dominant_commit_type`	`feat`-dominant repos are building products; `fix`-dominant repos are maintaining services; `chore`-dominant suggests tooling

Feature: commit_velocity_trend
Classification Signal: Increasing velocity → likely active product (app/service); decreasing → maintenance mode (library)

Portfolio-Level Observations

Most actively developed: theunpartyapp at 168.5 commits/month — classified as framework

Dormant (>180 days since last commit): theunpartyhtml, theunpartychat, theunpartyterminal, theunpartychatty, theunpartyproject, theunpartyporter, theunpartymood, theunpartylogger, theunpartyinvite, theunpartyfoundation, theunpartydashboard, theunpartycolors, theunpartybq, theunpartybot, theunpartyselector, theunpartysearch, theunpartyspace, theunpartystarter, theunpartyswatch, theunpartyteam, theunpartytheme, theunpartyuser

Conventional commits adopted by 11/44 repos (25.0%) — highest adoption in types: assistant

---

Analysis generated by scripts/classification_analysis.py · theunpartybuilder v2.0