@orderful/droid 0.31.2 → 0.33.0

package/src/tools/wrapup/skills/wrapup/SKILL.md

@@ -0,0 +1,86 @@
+---
+name: wrapup
+description: "Session wrap-up that captures decisions, learnings, and open items to persistent docs. Use when ending a session to persist context. User prompts like 'let's wrap up', 'end of session', 'capture what we did'."
+allowed-tools: [Read, Edit, Write, Glob, Grep, Bash, Task]
+---
+
+# Wrapup Skill
+
+Session wrap-up that captures decisions, learnings, and open items before closing.
+
+## Context Lifeboat Pattern
+
+This skill runs at the END of sessions when context pressure is highest.
+To avoid triggering compaction mid-wrapup, offload heavy lifting to fresh-context subagents.
+
+## Phase 1: Quick Context Extraction
+
+You have the conversation history. Quickly extract and write to `/tmp/wrapup-session-brief.md`:
+
+```markdown
+# Session Brief
+
+## User Goals
+- {what the user was trying to accomplish}
+
+## Decisions Made
+- {explicit choices, design decisions, why}
+
+## Key Learnings
+- {discoveries, aha moments, findings}
+
+## Open Items
+- {unresolved questions, blockers, WIP state}
+
+## Files Touched
+- {files discussed/modified and why}
+
+## Mentioned But Not Done
+- {bugs to file, docs to update, future work}
+```
+
+Do this IMMEDIATELY. Don't analyse deeply—just dump what you know.
+
+## Phase 2: Artifact Analysis
+
+Spawn subagents **in parallel** to analyse artifacts. Each reads your session brief and analyses one area.
+
+| Subagent | Type | Analyses |
+|----------|------|----------|
+| Git state | Bash | Uncommitted changes, branch status |
+| Project file | Explore | ~/.droid/project.md staleness |
+| Brain docs | Explore | Recently modified docs needing updates |
+| Codex gaps | Explore | Knowledge worth codifying |
+
+See `references/subagent-prompts.md` for detailed prompts.
+
+## Phase 3: Synthesise and Present
+
+Combine subagent findings into a summary:
+
+```markdown
+## Wrapup Summary
+
+**Session:** {brief description}
+
+**Uncommitted Work:**
+{from git subagent}
+
+**Proposed Updates:**
+1. **{destination}** ({type})
+   - {item}
+
+**Suggested Actions:**
+- {action}: {description}
+```
+
+See `references/output-schema.md` for structured output format.
+
+## Phase 4: Confirm and Write
+
+Present options using AskUserQuestion:
+- **"Yes, update all"** → Write to all destinations
+- **"Let me review first"** → Show detailed diff for each
+- **"Skip"** → Close without updating
+
+On approval, write updates using appropriate format for each destination.

package/src/tools/wrapup/skills/wrapup/references/output-schema.md

@@ -0,0 +1,93 @@
+# Output Schema
+
+Structured format for subagent findings. Main agent uses this to synthesise the wrapup summary.
+
+## Schema
+
+```json
+{
+  "active_contexts": [
+    {
+      "type": "brain | project | codex | git",
+      "name": "human-readable name",
+      "path": "file path (if applicable)",
+      "status": "active | stale | clean"
+    }
+  ],
+  "proposed_updates": [
+    {
+      "destination": "filename or identifier",
+      "destination_type": "brain | project | codex",
+      "items": [
+        {
+          "category": "decision | learning | open_item | next_step",
+          "content": "text to add"
+        }
+      ]
+    }
+  ],
+  "suggested_actions": [
+    {
+      "type": "commit | codex_topic | file_bug | update_docs",
+      "content": "description of suggested action",
+      "priority": "high | medium | low"
+    }
+  ]
+}
+```
+
+## Field Descriptions
+
+### active_contexts
+
+What's currently "open" or relevant:
+- **brain**: Active planning/research docs
+- **project**: ~/.droid/project.md if it exists
+- **codex**: Relevant codex entries touched
+- **git**: Current branch and uncommitted state
+
+### proposed_updates
+
+Concrete changes to write on approval:
+- **destination**: Where to write (filename)
+- **destination_type**: Which system (brain, project, codex)
+- **items**: List of things to add, categorised
+
+Item categories:
+- `decision`: Explicit choice made during session
+- `learning`: Discovery, insight, or finding
+- `open_item`: Unresolved question or blocker
+- `next_step`: Identified future work
+
+### suggested_actions
+
+Actions that need user judgment:
+- `commit`: Git changes ready to commit
+- `codex_topic`: Knowledge worth codifying
+- `file_bug`: Bug discovered but not filed
+- `update_docs`: Documentation that needs updating
+
+## Example
+
+```json
+{
+  "active_contexts": [
+    {"type": "brain", "name": "plan-wrapup-tool", "path": "1-Projects/droid/plans/plan-wrapup-tool.md", "status": "active"},
+    {"type": "git", "name": "tf/wrapup-tool", "path": null, "status": "uncommitted"}
+  ],
+  "proposed_updates": [
+    {
+      "destination": "plan-wrapup-tool.md",
+      "destination_type": "brain",
+      "items": [
+        {"category": "decision", "content": "Using context lifeboat pattern - main agent extracts, subagents analyse"},
+        {"category": "next_step", "content": "Test with real session, iterate on prompts"}
+      ]
+    }
+  ],
+  "suggested_actions": [
+    {"type": "commit", "content": "4 files ready: TOOL.yaml, SKILL.md, wrapup.md, changeset", "priority": "high"},
+    {"type": "codex_topic", "content": "Context lifeboat pattern for end-of-session tooling", "priority": "medium"}
+  ]
+}
+```

package/src/tools/wrapup/skills/wrapup/references/subagent-prompts.md

@@ -0,0 +1,93 @@
+# Subagent Prompts
+
+Detailed prompts for Phase 2 artifact analysis. Spawn these in parallel using the Task tool.
+
+## Git State Analysis
+
+**Subagent type:** Bash
+
+```
+Analyse git status and git diff in the current working directory. Categorise changes:
+
+1. **Ready to commit** - Complete, coherent changes that could be committed now
+2. **WIP** - Partial changes, incomplete features, or messy state
+3. **Untracked files** - New files that might need to be added or ignored
+
+For each category, list the files and briefly describe what changed.
+
+Output format:
+- Status: clean | uncommitted | wip
+- Branch: {current branch name}
+- Summary of uncommitted work (if any)
+- Recommendation: commit now, stash, or leave as-is
+```
+
+## Project File Check
+
+**Subagent type:** Bash
+
+```
+Read /tmp/wrapup-session-brief.md for session context.
+
+First, run `droid config project` to get the project skill configuration.
+Parse the JSON output to find the `projects_dir` value.
+
+If `projects_dir` is not set, skip project file analysis.
+
+If configured:
+1. Search the projects directory for recently modified PROJECT.md files (last 7 days)
+2. Read the most recently modified project file
+3. Compare against session brief
+4. Identify:
+   - Stale entries that should be updated or removed
+   - New items from the session to add (decisions, blockers, next steps)
+   - Status changes (completed items, new blockers)
+
+Output specific proposed updates with exact text to add/change.
+If no active project found, note that and skip.
+```
+
+## Brain Doc Check
+
+**Subagent type:** Explore
+
+```
+Read /tmp/wrapup-session-brief.md for session context.
+
+First, run `droid config brain` and parse the JSON output for the `brain_dir` setting.
+If not configured, skip brain doc analysis.
+
+If configured:
+1. Search the brain directory for recently modified .md files (last 24 hours)
+2. Read any that seem related to the session topics
+3. Check if they need updates based on session decisions/learnings
+4. Look for @droid comments that were addressed
+
+Output:
+- Active brain docs found (if any)
+- Proposed additions for each relevant doc
+- Any resolved @droid comments to mark done
+```
+
+## Codex Gap Analysis
+
+**Subagent type:** Explore
+
+```
+Read /tmp/wrapup-session-brief.md for session context.
+
+Run `droid config codex` and check if `codex_repo` is set.
+If not configured, skip codex analysis.
+
+If the session involved:
+- Deep research on a topic
+- Discovering patterns or best practices
+- Solving a non-trivial problem with reusable insights
+
+Then suggest potential codex topics:
+- Topic name
+- Brief description of what to capture
+- Why it's worth codifying (reusable, non-obvious, hard-won knowledge)
+
+Output: List of suggested codex topics, or "No codex suggestions" if session was routine.
+```