so

name	so
description	Standardized output format for GOSM projects.
output	{"format":"prose"}

Structured Output Format

Input: $ARGUMENTS

---

Overview

Standardized output format for GOSM projects. Problem: Current outputs are markdown files — good for reading, hard to analyze. Solution: Structured format that enables cross-project synthesis.

Like ARAW uses databases for synthesis, GOSM needs structured project records.

Steps

Step 1: Identify Output Type

What kind of output is being produced?

Output Type	Key Fields	Example
Analysis	claim, evidence, confidence, caveats	ARAW result
Decision	options, criteria, scores, selection	DCP result
Plan	goals, steps, dependencies, timeline	Project plan
Diagnosis	symptoms, causes, evidence, fix	RCA result
Evaluation	subject, dimensions, scores, verdict	Evaluation result
Review	findings, severity, recommendations	Audit result

Step 2: Apply Standard Structure

Every output includes these sections:

HEADER:
  type: [analysis|decision|plan|diagnosis|evaluation|review]
  skill: [which procedure produced this]
  input: [what was analyzed]
  date: [when]
  confidence: [overall confidence: high|medium|low]

SUMMARY:
  one_line: [single sentence result]
  verdict: [the bottom line]
  key_finding: [most important thing discovered]

BODY:
  [type-specific fields — see Step 3]

METADATA:
  assumptions: [what was assumed]
  limitations: [what this analysis can't address]
  follow_ups: [what should be investigated next]
  related: [links to related analyses]

Step 3: Type-Specific Body Fields

Analysis body:

claims:
  - claim: [statement]
    evidence_for: [supporting evidence]
    evidence_against: [contradicting evidence]
    confidence: [H/M/L]
    status: [supported|refuted|uncertain]

Decision body:

options:
  - name: [option]
    pros: [advantages]
    cons: [disadvantages]
    score: [weighted score]
selected: [chosen option]
rationale: [why this one]

Plan body:

goal: [what we're trying to achieve]
steps:
  - step: [description]
    depends_on: [prerequisites]
    output: [what this step produces]
    risk: [what could go wrong]

Diagnosis body:

symptoms: [what was observed]
hypotheses:
  - cause: [possible cause]
    evidence: [supporting/contradicting]
    probability: [H/M/L]
root_cause: [most likely cause]
fix: [recommended action]

Evaluation body:

subject: [what was evaluated]
dimensions:
  - dimension: [evaluation criterion]
    score: [rating]
    evidence: [basis for score]
overall: [aggregate assessment]

Review body:

findings:
  - finding: [what was found]
    severity: [critical|major|minor|info]
    recommendation: [what to do]
    status: [new|known|resolved]

Step 4: Cross-Output Synthesis

Structured outputs enable:

1. Trend analysis: Compare confidence levels across analyses
2. Decision tracking: Review past decisions and outcomes
3. Assumption auditing: Find shared assumptions across projects
4. Gap identification: What hasn't been analyzed?
5. Contradiction detection: Where do analyses disagree?

Step 5: Convert Current Output

If applying to existing output:

1. Read the unstructured output
2. Extract fields for the appropriate type
3. Identify anything that doesn't fit standard fields → add to metadata
4. Fill in any missing standard fields (especially limitations and assumptions)
5. Output in structured format

Step 6: Report

Format the input according to the appropriate structured output template. If the input is a request to format something, apply the template. If it's a question about the format, explain the relevant section.

When to Use

• When producing output from any GOSM procedure
• When standardizing existing outputs for synthesis
• When designing new output formats
• When comparing results across multiple analyses

Verification

• Output type correctly identified
• All standard header fields present
• Summary is genuinely one line (not a paragraph)
• Confidence level justified
• Assumptions and limitations explicitly stated
• Follow-ups identified