linear-looplisted

# Linear Loop The Loop methodology implemented inside Linear + Claude Code. Everything is an issue. The issue queue is the orchestration layer. ## The Loop ``` Idea → Research → Hypothesis → Plan → Execute → Monitor → Signal → Loop continues ``` - **Linear** is the data store (issues, projects, labels, relations) - **Claude Code** is the intelligence (triage, planning, execution, monitoring) - **DorkOS** is the infrastructure (Pulse scheduling, Relay messaging, Mesh coordination) - **The spec workflow** handles implementation depth (complex work routes through /ideate → /spec:execute) ## Configuration Read `config.json` in this skill directory for repo-specific settings: - `team` — which Linear team this repo maps to - `filter.ownership` — how `/pm` scopes projects and issues: - `"unassigned"` (default) — projects with no lead, issues with no assignee - `"all"` — everything in the team, no filtering - `pm.autoLimit` — max actions in `/pm auto` mode - `pm.approvalGates` — actions requiring human approval even in auto mode ## Accessing Linear Two interchangeable access paths reach the same workspace and DorkOS team (see `config.json` — slug `dorkos`, id `a171dbd5-3ccc-40ab-b58b-1fae7644fba8`, key `DOR`): 1. **Linear MCP tools** (primary) — `mcp__linear__*` (`list_issues`, `list_projects`, `create_issue`, …). Requires the in-session MCP server to be authenticated (OAuth); if it isn't, start the flow with `mcp__linear__authenticate`. The spec-command breadcrumbs bel