Add 5 pi extensions: pi-subagents, pi-crew, rpiv-pi, pi-interactive-shell, pi-intercom

extensions/pi-crew/workflows/default.workflow.md

@@ -0,0 +1,29 @@
+---
+name: default
+description: Explore, plan, execute, and verify
+---
+
+## explore
+role: explorer
+
+Explore the codebase for the goal: {goal}
+
+## plan
+role: planner
+dependsOn: explore
+output: plan.md
+
+Create a concise implementation plan for: {goal}
+
+## execute
+role: executor
+dependsOn: plan
+
+Implement the plan for: {goal}
+
+## verify
+role: verifier
+dependsOn: execute
+verify: true
+
+Verify completion for: {goal}

extensions/pi-crew/workflows/fast-fix.workflow.md

@@ -0,0 +1,22 @@
+---
+name: fast-fix
+description: Minimal workflow for small fixes
+---
+
+## explore
+role: explorer
+
+Find the likely source of the issue: {goal}
+
+## execute
+role: executor
+dependsOn: explore
+
+Make the smallest safe fix.
+
+## verify
+role: verifier
+dependsOn: execute
+verify: true
+
+Verify the fix with available evidence.

extensions/pi-crew/workflows/implementation.workflow.md

@@ -0,0 +1,38 @@
+---
+name: implementation
+description: Adaptive implementation workflow where a planner agent decides the subagent fanout
+---
+
+## assess
+role: planner
+output: adaptive-plan.json
+
+Assess this task and decide how many subagents are actually needed for: {goal}
+
+You are the orchestration planner. Inspect the repository enough to choose an efficient crew; do not use a fixed template. Small/simple tasks may need one executor plus one verifier. Risky or broad tasks may need parallel explorers, specialists, implementers, reviewers, security reviewers, or test engineers.
+
+Return a concise rationale, then include exactly one JSON block between these markers:
+
+ADAPTIVE_PLAN_JSON_START
+{
+  "phases": [
+    {
+      "name": "short-phase-name",
+      "tasks": [
+        {
+          "role": "explorer|analyst|planner|critic|executor|reviewer|security-reviewer|test-engineer|verifier|writer",
+          "title": "short task title",
+          "task": "specific autonomous task prompt for this subagent"
+        }
+      ]
+    }
+  ]
+}
+ADAPTIVE_PLAN_JSON_END
+
+Rules:
+- Choose the smallest effective number of subagents.
+- Use parallel tasks in the same phase only when their work is independent.
+- Later phases depend on all tasks in the previous phase.
+- Include verification/review tasks when implementation is requested.
+- Do not include more than 12 total subagents; split or summarize oversized plans instead.

extensions/pi-crew/workflows/parallel-research.workflow.md

@@ -0,0 +1,46 @@
+---
+name: parallel-research
+description: Parallel research with shard exploration and synthesis
+---
+
+## discover
+role: explorer
+
+Discover the relevant files/projects for: {goal}. Return a shard plan with paths grouped by topic. Do not deeply read every file yet; focus on routing the work.
+
+## explore-core
+role: explorer
+parallelGroup: explore
+
+Explore the core/runtime shard from the discover output. Focus on architecture, package config, docs, and reusable patterns for: {goal}
+
+## explore-ui
+role: explorer
+parallelGroup: explore
+
+Explore the UI/TUI/extension-interface shard from the discover output. Focus on widgets, overlays, commands, status bars, package config, docs, and reusable patterns for: {goal}
+
+## explore-runtime
+role: explorer
+parallelGroup: explore
+
+Explore the worker/runtime/subagent/runtime-control shard from the discover output. Focus on process/session/runtime orchestration, event streams, logs, package config, docs, and reusable patterns for: {goal}
+
+## explore-extensions
+role: explorer
+parallelGroup: explore
+
+Explore the extension bundle/small-package shard from the discover output. Focus on package config, extension registration, commands/tools, docs, and reusable patterns for: {goal}
+
+## synthesize
+role: analyst
+dependsOn: explore-core, explore-ui, explore-runtime, explore-extensions
+
+Synthesize all shard findings. Use discover output if available, but do not require it. Identify common patterns, gaps, and concrete recommendations.
+
+## write
+role: writer
+dependsOn: synthesize
+output: research-summary.md
+
+Write a concise final summary with evidence, risks, and actionable next steps.

extensions/pi-crew/workflows/research.workflow.md

@@ -0,0 +1,22 @@
+---
+name: research
+description: Research and write up findings
+---
+
+## explore
+role: explorer
+
+Gather relevant facts for: {goal}
+
+## analyze
+role: analyst
+dependsOn: explore
+
+Analyze and organize the findings.
+
+## write
+role: writer
+dependsOn: analyze
+output: research-summary.md
+
+Write a concise final summary with evidence and open questions.

extensions/pi-crew/workflows/review.workflow.md

@@ -0,0 +1,30 @@
+---
+name: review
+description: Review workflow for correctness and security
+---
+
+## explore
+role: explorer
+
+Identify changed or relevant areas for review: {goal}
+
+## code-review
+role: reviewer
+dependsOn: explore
+parallelGroup: review
+
+Review correctness, maintainability, tests, and regressions.
+
+## security-review
+role: security-reviewer
+dependsOn: explore
+parallelGroup: review
+
+Review security risks and trust boundaries.
+
+## verify
+role: verifier
+dependsOn: code-review, security-review
+verify: true
+
+Summarize review outcome and pass/fail status.