| name | pc-process-input |
| description | Extract structured information from external inputs — transcripts, documents, meeting notes, Slack threads, or pasted text. Produces requirements, constraints, decisions, and open questions in a consistent format. Use when the user has raw text to analyze, pastes a transcript or meeting notes, wants to extract actionable items from a document, or says "process this", "extract from this", "what's in this document". Source — Consortium-team/project-creator.
|
/pc-process-input — Extract Structured Info from Documents
Extract requirements, constraints, decisions, and questions from transcripts, documents, or notes.
Usage:
/pc-process-input — Prompts for input- Paste text or provide a file path when prompted
Step 1: Get the Input
If no content was provided with the command:
Ready to process input.
You can:
1. Paste text directly (transcript, notes, etc.)
2. Provide a file path to read
What would you like me to process?
If file path provided: Read the file and use its contents
If text pasted: Use the pasted text directly
Step 2: Analyze the Input
Read through the input and extract:
Requirements — Things that must be done
- Explicit statements: "it needs to...", "we want...", "must have..."
- Implicit needs revealed by problems described
- User stories or use cases mentioned
Constraints — Boundaries and limitations
- Technical: "we're using...", "has to integrate with..."
- Timeline: "by Q2", "before launch"
- Resources: "small team", "limited budget"
- Organizational: "needs approval from...", "compliance requires..."
Decisions — Choices that were made or implied
- "We decided to...", "we're going with..."
- Trade-offs discussed and resolved
- Approaches rejected and why
Questions — Things that need clarification
- Ambiguous statements
- Conflicting requirements
- Missing information
- Assumptions that should be validated
Context — Background that helps understand the project
- Domain knowledge
- Related systems
- User personas
- Historical context
Step 3: Present Extraction
Show what was extracted before writing:
## Extracted from Input
### Requirements Found
- [requirement 1]
- [requirement 2]
### Constraints Identified
- [constraint 1]
- [constraint 2]
### Decisions Captured
- [decision 1] — [rationale if stated]
- [decision 2]
### Questions Raised
- [question 1] — needs clarification
- [question 2] — conflicting with [what]
### Context Notes
- [context 1]
- [context 2]
---
Does this capture what's important? Anything I missed or misinterpreted?
STOP and wait for user confirmation before writing any files.
The user may want to:
- Confirm the extraction is accurate
- Correct misinterpretations
- Add items that were missed
- Remove items that aren't relevant
Step 4: Clarify Questions (Optional)
If there are questions or ambiguities:
I flagged [N] items that need clarification. Would you like to address them now?
1. [Question 1]
2. [Question 2]
We can clarify now, or save them as open questions for later.
If user wants to clarify, ask one question at a time and capture the answers.
Step 5: Save Output
After confirmation, ask the user where to save the structured output. Default suggestion: a markdown file in the current project directory.
Output format:
# Processed: [brief description of input]
**Source:** [input type — transcript, document, notes, etc.]
**Processed:** [date]
## Requirements
- [requirement 1]
- [requirement 2]
## Constraints
- [constraint 1]
- [constraint 2]
## Decisions
| Decision | Rationale | Date |
|----------|-----------|------|
| [decision] | [rationale] | [date] |
## Open Questions
| Question | Source | Status |
|----------|--------|--------|
| [question] | [where it came from] | open |
## Context
- [context note 1]
- [context note 2]
Step 6: Summarize
## Processed: [brief description of input]
**Extracted:**
- [N] requirements
- [N] constraints
- [N] decisions
- [N] open questions
**Saved to:** [file path]
Extraction Guidelines
Be Conservative
- Only extract what's clearly stated or strongly implied
- Don't invent requirements that weren't mentioned
- Flag ambiguity rather than guessing
Preserve Source
- Note where each item came from
- Keep original language when it's precise
- Paraphrase only for clarity
Identify Conflicts
- If new input conflicts with existing context, flag it
- Don't silently override previous decisions
- Present conflicts for user resolution
Questions Are Valuable
- Unclear input is a signal, not a failure
- Capturing questions prevents false assumptions
- Open questions guide future discussions
