askill
sessionstellar

sessionstellarSafety 95Repository

Score AI orchestration sessions — extract signals, compute a 0-100 score across skill diversity, decision depth, error recovery, compound learning, and orchestration mastery. Works fully offline, no API key required.

0 stars
1.2k downloads
Updated 2/20/2026

Package Files

Loading files...
SKILL.md

SessionStellar

Score AI orchestration sessions from any MCP client. Fully offline — zero API keys, zero network calls.

When to use

Invoke this skill when the user wants to:

  • Score a session transcript (.md, .txt, or .jsonl)
  • Understand where their AI orchestration is weakest
  • Compare scores across sessions
  • Enable automatic scoring after every git commit

How to score a file

npx sessionstellar score <path-to-session-file>

Returns a structured score:

┌─────────────────────────────────────────┐
│  SessionStellar Score                   │
│  overall: 74 / 100                      │
├─────────────────────────────────────────┤
│  Skill Diversity        ████████░░  82  │
│  Decision Depth         ███████░░░  71  │
│  Error Recovery         ████████░░  80  │
│  Compound Learning      ██████░░░░  63  │
│  Orchestration Mastery  ███████░░░  70  │
├─────────────────────────────────────────┤
│  skills: 12  agents: 4  decisions: 7   │
└─────────────────────────────────────────┘

How to score from stdin (pipe)

cat session.md | npx sessionstellar score -

Get JSON output

npx sessionstellar score session.md --json

Output:

{
  "sessionId": "...",
  "overallScore": 74,
  "metrics": {
    "skillDiversity": 82,
    "decisionDepth": 71,
    "errorRecoveryRate": 80,
    "compoundLearningSignals": 63,
    "orchestrationMastery": 70
  },
  "signals": {
    "skillsInvoked": [...],
    "agentsSpawned": [...],
    "decisionPoints": [...],
    "errorRecoveries": [...],
    "learningMoments": [...]
  }
}

Enable auto-scoring on every git commit

npx sessionstellar enable

Installs a post-commit hook that automatically scores any session file touched in each commit. Scores are saved to .sessionstellar/scores/ (gitignored).

npx sessionstellar status    # show hook status + recent scores
npx sessionstellar disable   # remove the hook

Scoring breakdown

MetricWeightWhat it measures
Skill Diversity20%Variety of skills invoked
Decision Depth25%Quality of branching decisions
Error Recovery20%How well errors were handled
Compound Learning20%Learning moments and adaptations
Orchestration Mastery15%Multi-agent and sub-task coordination

Upload to leaderboard (optional)

npx sessionstellar score session.md
# Prompts: "Upload to leaderboard? (y/N)"

Public leaderboard: https://sessionstellar.com/leaderboard

Skip upload prompt:

npx sessionstellar score session.md --no-upload

Install

Download ZIP
Requires askill CLI v1.0+

AI Quality Score

93/100Analyzed 2/25/2026

High-quality skill document for SessionStellar, a tool that scores AI orchestration sessions. Excellent completeness with clear when-to-use section, multiple usage modes, scoring breakdown table, and auto-scoring setup. Actionable with executable commands and examples. Highly reusable as a general-purpose tool for any MCP client. Well-structured with proper metadata including emoji, install instructions, and tags. Safety is excellent - runs fully offline with no network calls.

95
95
90
95
90

Metadata

Licenseunknown
Version-
Updated2/20/2026
Publishersunnypatneedi

Tags

apiobservabilityprompting