القائمة
Same skill compression power in half the context — 975 tokens vs 2,500.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Skill compression reminder in 100 tokens — just trigger, action, result.
Complete skill compression documentation — all options, modes, and calibration details (~2,500 tokens).
Fit more skills in your context window — compress without losing what matters.
Full creative synthesis — combine song, visual concept, and TED talk from a single technical insight
Transform technical insights into full 40-50 minute TED-style talks with concrete examples and Q&A
Transform technical insights into Suno-ready songs with lyrics, style tags, and emotional arc
| name | Skill Distiller (Compressed) |
| version | 0.2.1 |
| description | Same skill compression power in half the context — 975 tokens vs 2,500. |
| author | Live Neon <lee@liveneon.ai> |
| homepage | https://github.com/live-neon/skills/tree/main/skill-distiller/compressed |
| repository | live-neon/skills |
| license | MIT |
| user-invocable | true |
| disable-model-invocation | true |
| emoji | 🗜️ |
| tags | ["compression","skills","optimization","context-window","token-reduction","lightweight","openclaw"] |
Self-compressed prose variant (~975 tokens, ~90% functionality, LLM-estimated). Full reference: ../SKILL.reference.md.
Role: Help users compress verbose skills to reduce context window usage Understands: Skills are verbose for human clarity but costly for context Approach: Identify section types, score importance, remove/shorten low-value sections Boundaries: Preserve functionality, report what was removed, never hide trade-offs Tone: Technical, precise, transparent about trade-offs
Data handling: All analysis uses your agent's configured model. No external APIs.
Activate when the user asks:
| Flag | Default | Description |
|---|---|---|
--mode | threshold | threshold (preserve X%), tokens (fit budget), oneliner |
--threshold | 0.9 | Functionality preservation target (0.0-1.0) |
--tokens | - | Target token count |
--provider | auto | ollama, gemini, openai (auto-detects) |
--verbose | false | Show section-by-section analysis |
--dry-run | false | Analyze without outputting |
Full options (--model, --debug-stages, --with-ci): see SKILL.reference.md
Threshold = semantic capability, not size ratio. A 0.9 threshold means 90% of agent behavior preserved, not 90% of lines kept. Judge by understanding, not metrics.
Parse into sections: Frontmatter, Headers, Code blocks, Lists, Prose.
| Type | Importance | Compressible? |
|---|---|---|
| TRIGGER | 1.0 | No |
| CORE_INSTRUCTION | 1.0 | No |
| CONSTRAINT | 0.9 | Partially |
| OUTPUT_FORMAT | 0.8 | Partially |
| EXAMPLE | 0.5 | Yes |
| EXPLANATION | 0.3 | Yes |
| VERBOSE_DETAIL | 0.2 | Yes (first) |
Protected patterns (boost to 0.85+): YAML name/description, Task creation, N-count tracking, Checkpoint/state, BEFORE/AFTER markers.
Evaluate by semantic understanding, NOT metrics.
| Wrong | Right |
|---|---|
| "60% line reduction is too aggressive" | "Can an agent execute this skill?" |
| "Token ratio exceeds target" | "Are triggers and actions preserved?" |
LLM scores 0-100 based on semantic capability, not line/token ratios. A 50% size reduction can preserve 95% functionality if removed content was verbose/redundant.
Append to .learnings/skill-distiller/calibration.jsonl with metrics and expected score.
Functionality preserved: 90% (uncalibrated - first 5 compressions build baseline)
Tokens: 2000 → 1800 (10% reduction)
Removed: [list], Kept: [list]
[Compressed skill markdown...]
| Pattern | Why |
|---|---|
YAML name/description | REQUIRED by spec |
| N-count tracking | Observation workflow |
| Task creation | Compaction resilience |
If removed: -10% score penalty, flagged in output.
Parallel/serial decisions, performance hints, caching guidance. No score penalty.
Storage: .learnings/skill-distiller/calibration.jsonl
| N-count | Meaning |
|---|---|
| N < 5 | Uncalibrated (LLM-only estimate) |
| N > 10 | Calibrated (historical CI) |
Feedback: /skill-distiller feedback --id=c1 --actual=85 --outcome="worked"
Guardrails:
Why 0.95: Capability loss compounds (0.95 x 0.95 = 0.90 at next level).
| Error | Hint |
|---|---|
| No content | Provide SKILL.md path or pipe via stdin |
| No frontmatter | Add --- block with name/description |
| LLM unavailable | Run ollama serve or set GEMINI_API_KEY |
| Variant | Tokens | Functionality |
|---|---|---|
| skill-distiller (main) | ~400 | ~90% (formula) |
| compressed (this) | ~975 | ~90% (prose) |
| oneliner | ~100 | ~70% |
Full reference: SKILL.reference.md (~2,500 tokens, ~90%)
Token counts use 4 chars/token heuristic (+/-20%). Functionality scores are LLM-estimated.