Input Tolerance

"Method over hacks. Evidence over assumption."

TL;DR

Normalize imperfect input: typos, voice, dyslexia, multilingual, multimodal. Extract intent from noise. This is an orchestration-layer skill used internally by Pristino and the adk-orchestrator. Protocol details in PRISTINO.md. [EXPLICIT]

Procedure

Step 1: Discover

• Gather input and context
• Load relevant indexes and configuration

Step 2: Analyze

• Evaluate options per Constitution XIII/XIV
• Score candidates by relevance and confidence

Step 3: Execute

• Apply the selected approach
• Tag all outputs with evidence markers

Step 4: Validate

• Verify quality criteria met
• Confirm confidence >= 0.95

Quality Criteria

• Evidence tags applied
• Constitution-compliant
• Confidence >= 0.95
• Actionable output

Related Skills

• See PRISTINO.md for full orchestration protocol

Usage

Example invocations:

• "/input-tolerance" — Run the full input tolerance workflow
• "input tolerance on this project" — Apply to current context

Assumptions & Limits

• Assumes access to project artifacts (code, docs, configs) [EXPLICIT]
• Requires English-language output unless otherwise specified [EXPLICIT]
• Does not replace domain expert judgment for final decisions [EXPLICIT]

Edge Cases

Scenario	Handling
Empty or minimal input	Request clarification before proceeding
Conflicting requirements	Flag conflicts explicitly, propose resolution
Out-of-scope request	Redirect to appropriate skill or escalate