Summarizer

Distill content to its essence with audience-aware compression, format selection, and quality verification.

Security Scan

This is an instruction-only summarization skill that contains guidance and templates for producing summaries — it requests no credentials, installs, or external access and is internally consistent with its stated purpose.

Purpose & Capability

Name and description match the files and instructions: the skill only defines summarization techniques, formats, and a protocol. It does not request unrelated resources or capabilities.

Instruction Scope

SKILL.md, techniques.md, and formats.md restrict behavior to analyzing and producing summaries; they do not instruct the agent to read arbitrary system files, call external endpoints, or access secrets. The 'Decline When' rule helps limit inappropriate uses.

Install Mechanism

No install spec and no code files (instruction-only). Nothing is written to disk or downloaded as part of installation.

Credentials

The skill requires no environment variables, credentials, or config paths — consistent with a pure guidance/template skill.

Persistence & Privilege

Flags are default (always: false, user-invocable: true, model invocation allowed). The skill does not request persistent presence or modify other skills.

Assessment

This skill is a safe, instruction-only template for producing summaries and does not request credentials, install code, or contact external endpoints. Before using, remember: (1) summaries can omit nuance or produce hallucinations — verify important facts against the source; (2) do not feed private or sensitive documents if you do not want their content processed by the agent; (3) the agent may invoke the skill autonomously by default — if you prefer manual use, restrict autonomous invocation in your agent settings. Overall the skill appears coherent with its stated purpose.

Core Principle

Good summaries preserve meaning while eliminating noise. Bad ones drop critical points or miss the audience.

Summarization = Compression + Preservation + Adaptation

Protocol

Analyze → Select technique → Extract → Compress → Format → Verify

1. Analyze

Before summarizing, determine:

• Length of source — tweet vs book chapter
• Complexity — technical, narrative, data-heavy
• Audience — expert, general, executive, student
• Purpose — quick overview, decision support, study aid

2. Select Technique

Match technique to content (see techniques.md):

Content type	Best technique
Simple/short	Zero-shot direct
Technical/complex	Chain-of-thought
Audience-specific	Role-based
Consistent style needed	Few-shot
Strict requirements	Instruction-heavy

3. Extract

Identify what matters:

• Core argument or thesis
• Key supporting points (3-5 max)
• Critical data or evidence
• Conclusions and implications

Rule: If you can't identify the core argument, you don't understand it yet.

4. Compress

Apply compression levels:

• TLDR — 1 sentence, core message only
• Brief — 2-3 sentences, message + key support
• Standard — paragraph, covers main points
• Extended — multiple paragraphs, preserves nuance

5. Format

Match output to purpose (see formats.md):

• Bullet points for scanning
• Paragraph for reading
• Structured sections for reports
• Tweet-length for social

6. Verify

Before delivering, check:

• Core message preserved?
• Key points included?
• Nothing critical dropped?
• Appropriate for audience?
• Right length for purpose?

Output Markers

📝 SUMMARY ([level]: [word count])
[Content]

💡 KEY POINTS
• [Point 1]
• [Point 2]

⚠️ OMITTED (if relevant)
[What was cut and why]

Decline When

Source is ambiguous, contradictory without resolution, or summarizing would lose essential nuance the user needs.

---

References: techniques.md, formats.md