gws-doc-action-extractor

name	gws-doc-action-extractor
description	Scan Google Docs shared with you for action patterns (e.g. "create a card", "action item", "open a ticket") in both content and comments, extract them as structured data, and create Jira issues. Use when the user asks to scan docs for actions, find TODOs in Google Docs, or automatically create Jira cards from document content.
allowed-tools	Bash(gws drive files list:*),Bash(gws drive files get:*),Bash(gws drive files export:*),Bash(gws auth:*),Bash(/output/gws-doc-action-extractor/scripts/*.sh:*),Bash(*/jira-utilities/scripts/*.sh:*)

Google Docs Action Extractor

Overview

Scan Google Docs for action patterns in content and inline comments, then create Jira cards from the extracted items.

Prerequisites:

• gws available and authenticated
• jq for JSON parsing
• Jira credentials set: JIRA_SITE, JIRA_TOKEN, JIRA_EMAIL

Detected Patterns

Pattern type	Examples
Explicit card request	"create a card", "create a ticket", "open a jira", "file a bug"
Action items section	Lines listed under "Action items:" in meeting docs
Assignee actions	"[Name] to do X", "[Name] will X"
TODO markers	"TODO:", "ACTION:", "FOLLOW UP:", "FOLLOWUP:"
Google Doc comments	Inline comment text [a]...[z] appended at end of export

Available Scripts

scan_docs.sh

Search all shared docs for action patterns using Drive full-text search.

Usage:

/output/gws-doc-action-extractor/scripts/scan_docs.sh [OPTIONS]

Options:

Option	Default	Description
--limit N	20	Max docs to return per pattern
--human	—	Human-readable output

Output: JSON list of matching docs with hit count per pattern type.

---

extract_actions.sh

Extract all action items from a specific Google Doc.

Usage:

/output/gws-doc-action-extractor/scripts/extract_actions.sh <file-id> [OPTIONS]

Options:

Option	Default	Description
--human	—	Human-readable output

Output JSON:

{
  "doc_id": "...",
  "doc_name": "...",
  "doc_link": "...",
  "actions": [
    {
      "type": "action_item",
      "text": "Alice to coordinate with Bob on the release plan",
      "assignee": "Alice",
      "line": 42,
      "context": "surrounding line"
    },
    {
      "type": "create_card",
      "text": "create a card for tracking the dependency delay",
      "assignee": "",
      "line": 110,
      "context": 
    
  

---

extract_recent_actions.sh

Extract action items only from content added in the last N days. Identifies date-labeled sections (e.g. Apr 27, 2026), restricts parsing to those sections, and builds a full Jira description including surrounding context for each action.

Usage:

/output/gws-doc-action-extractor/scripts/extract_recent_actions.sh <file-id> [OPTIONS]

Options:

Option	Default	Description
--days N	7	Look back N days
--human	—	Human-readable output

Examples:

# Last 7 days, human-readable
/output/gws-doc-action-extractor/scripts/extract_recent_actions.sh <file-id> --human

# Last 14 days, pipe to Jira card creator
/output/gws-doc-action-extractor/scripts/extract_recent_actions.sh <file-id> --days 14 \
  | /output/gws-doc-action-extractor/scripts/create_jira_cards.sh --project <PROJECT-KEY> --dry-run

Output JSON per action includes:

• section_date — the date header of the section it came from
• type — create_card, assignee_action, ai_action, spike
• text — the matched line
• assignee — extracted name if present
• context — ±3 lines surrounding the match
• jira_description — ready-to-use Jira card body with source, date, context and full section

---

create_jira_cards.sh

Create Jira issues from extracted actions (reads output of extract_actions.sh).

Usage:

/output/gws-doc-action-extractor/scripts/extract_actions.sh <file-id> \
  | /output/gws-doc-action-extractor/scripts/create_jira_cards.sh [OPTIONS]

Options:

Option	Default	Description
--project KEY	required	Jira project key (e.g. <PROJECT-KEY>)
--type TYPE	Task	Issue type (Task, Bug, Story)
--dry-run	—	Print what would be created, don't call Jira

---

Typical Workflow

Check recent changes and suggest cards (most common)

# Review last 7 days across a doc and suggest cards
/output/gws-doc-action-extractor/scripts/extract_recent_actions.sh <file-id> --days 7 --human

# Pipe directly to Jira card creator (dry-run first)
/output/gws-doc-action-extractor/scripts/extract_recent_actions.sh <file-id> --days 7 \
  | /output/gws-doc-action-extractor/scripts/create_jira_cards.sh --project <PROJECT-KEY> --dry-run

Scan all shared docs then create cards

# Step 1: Find docs with action patterns
/output/gws-doc-action-extractor/scripts/scan_docs.sh --human

# Step 2: Extract actions from a specific doc
/output/gws-doc-action-extractor/scripts/extract_actions.sh <file-id> --human

# Step 3: Review and create Jira cards (dry-run first)
/output/gws-doc-action-extractor/scripts/extract_actions.sh <file-id> \
  | /output/gws-doc-action-extractor/scripts/create_jira_cards.sh --project <PROJECT-KEY> --dry-run

# Step 4: Create for real
/output/gws-doc-action-extractor/scripts/extract_actions.sh <file-id> \
  | /output/gws-doc-action-extractor/scripts/create_jira_cards.sh --project <PROJECT-KEY>

Agent workflow (Claude handles review + creation)

1. Run scan_docs.sh to find candidate docs
2. Run extract_actions.sh on each doc
3. Present extracted actions to user for confirmation
4. Run create_jira_cards.sh for confirmed actions only

Error Handling

Scenario	Behavior
No actions found	Returns {"actions": []}
Jira auth missing	create_jira_cards.sh exits with clear error
Doc not accessible	extract_actions.sh exits with 403/404