Phase 1 — Context Gathering (sequential, no parallelism)

CRITICAL: Complete ALL of Phase 1 before starting Phase 2. Do NOT launch any code search, devsql queries, or subagents until the ticket/thread content is fully available and summarized. The entire point of this phase is to build context that makes Phase 2 searches effective.

1.0 Detect Input Type

Inspect $ARGUMENTS to determine the source:

Slack URL (matches https://<workspace>.slack.com/archives/<channel_id>/p<timestamp>):

• Extract channel_id from the path segment after /archives/ (e.g., C09JSQNCR33)
• Extract message timestamp: remove the p prefix from the last path segment, insert . before the last 6 digits (e.g., p1773934352256539 → 1773934352.256539)
• Proceed to Step 1.1a.

Linear URL (contains linear.app):

• Parse the issue ID from the path (pattern: [A-Z]+-\d+, e.g., linear.app/<workspace>/issue/ENG-1234/... → ENG-1234)
• Proceed to Step 1.1b.

Jira URL (contains atlassian.net):

• Parse the issue ID from the path (pattern: [A-Z]+-\d+)