overseer
Manage tasks via Overseer codemode MCP. Use when tracking multi-session work, breaking down implementation, or persisting context for handoffs.
What this skill does
# Agent Coordination with Overseer
## Core Principle: Tickets, Not Todos
Overseer tasks are **tickets** - structured artifacts with comprehensive context:
- **Description**: One-line summary (issue title)
- **Context**: Full background, requirements, approach (issue body)
- **Result**: Implementation details, decisions, outcomes (PR description)
Think: "Would someone understand the what, why, and how from this task alone AND what success looks like?"
## Task IDs are Ephemeral
**Never reference task IDs in external artifacts** (commits, PRs, docs). Task IDs like `task_01JQAZ...` become meaningless once tasks complete. Describe the work itself, not the task that tracked it.
## Overseer vs OpenCode's TodoWrite
| | Overseer | TodoWrite |
| --------------- | ------------------------------------- | ---------------------- |
| **Persistence** | SQLite database | Session-only |
| **Context** | Rich (description + context + result) | Basic |
| **Hierarchy** | 3-level (milestone -> task -> subtask)| Flat |
Use **Overseer** for persistent work. Use **TodoWrite** for ephemeral in-session tracking only.
## When to Use Overseer
**Use Overseer when:**
- Breaking down complexity into subtasks
- Work spans multiple sessions
- Context needs to persist for handoffs
- Recording decisions for future reference
**Skip Overseer when:**
- Work is a single atomic action
- Everything fits in one message exchange
- Overhead exceeds value
- TodoWrite is sufficient
## Finding Work
```javascript
// Get next ready task with full context (recommended for work sessions)
const task = await tasks.nextReady(milestoneId); // TaskWithContext | null
if (!task) {
console.log("No ready tasks");
return;
}
// Get all ready tasks (for progress overview)
const readyTasks = await tasks.list({ ready: true }); // Task[]
```
**Use `nextReady()`** when starting work - returns `TaskWithContext | null` (deepest ready leaf with full context chain + inherited learnings).
**Use `list({ ready: true })`** for status/progress checks - returns `Task[]` without context chain.
## Basic Workflow
```javascript
// 1. Get next ready task (returns TaskWithContext | null)
const task = await tasks.nextReady();
if (!task) return "No ready tasks";
// 2. Review context (available on TaskWithContext)
console.log(task.context.own); // This task's context
console.log(task.context.parent); // Parent's context (if depth > 0)
console.log(task.context.milestone); // Root milestone context (if depth > 1)
console.log(task.learnings.own); // Learnings attached to this task (bubbled from children)
// 3. Start work (VCS required - creates bookmark, records start commit)
await tasks.start(task.id);
// 4. Implement...
// 5. Complete with learnings (VCS required - commits changes, bubbles learnings to parent)
await tasks.complete(task.id, {
result: "Implemented login endpoint with JWT tokens",
learnings: ["bcrypt rounds should be 12 for production"]
});
// Alternative: Cancel if abandoning (does NOT satisfy blockers)
await tasks.cancel(task.id);
// 6. Archive finished tasks to hide from default list
await tasks.archive(task.id);
```
See @file references/workflow.md for detailed workflow guidance.
## Understanding Task Context
Tasks have **progressive context** - inherited from ancestors:
```javascript
const task = await tasks.get(taskId); // Returns TaskWithContext
// task.context.own - this task's context (always present)
// task.context.parent - parent task's context (if depth > 0)
// task.context.milestone - root milestone's context (if depth > 1)
// Task's own learnings (bubbled from completed children)
// task.learnings.own - learnings attached to this task
```
## Return Type Summary
| Method | Returns | Notes |
|--------|---------|-------|
| `tasks.get(id)` | `TaskWithContext` | Full context chain + inherited learnings |
| `tasks.nextReady()` | `TaskWithContext \| null` | Deepest ready leaf with full context |
| `tasks.list()` | `Task[]` | Basic task fields only |
| `tasks.create()` | `Task` | No context chain |
| `tasks.start/complete()` | `Task` | No context chain |
## Blockers
Blockers prevent a task from being ready until the blocker completes.
**Constraints:**
- Blockers cannot be self
- Blockers cannot be ancestors (parent, grandparent, etc.)
- Blockers cannot be descendants
- Creating/reparenting with invalid blockers is rejected
```javascript
// Add blocker - taskA waits for taskB
await tasks.block(taskA.id, taskB.id);
// Remove blocker
await tasks.unblock(taskA.id, taskB.id);
```
## Task Hierarchies
Three levels: **Milestone** (depth 0) -> **Task** (depth 1) -> **Subtask** (depth 2).
| Level | Name | Purpose | Example |
|-------|------|---------|---------|
| 0 | **Milestone** | Large initiative | "Add user authentication system" |
| 1 | **Task** | Significant work item | "Implement JWT middleware" |
| 2 | **Subtask** | Atomic step | "Add token verification function" |
**Choosing the right level:**
- Small feature (1-2 files) -> Single task
- Medium feature (3-7 steps) -> Task with subtasks
- Large initiative (5+ tasks) -> Milestone with tasks
See @file references/hierarchies.md for detailed guidance.
## Recording Results
Complete tasks **immediately after implementing AND verifying**:
- Capture decisions while fresh
- Note deviations from plan
- Document verification performed
- Create follow-up tasks for tech debt
Your result must include explicit verification evidence. See @file references/verification.md.
## Best Practices
1. **Right-size tasks**: Completable in one focused session
2. **Clear completion criteria**: Context should define "done"
3. **Don't over-decompose**: 3-7 children per parent
4. **Action-oriented descriptions**: Start with verbs ("Add", "Fix", "Update")
5. **Verify before completing**: Tests passing, manual testing done
---
## Reading Order
| Task | File |
|------|------|
| Understanding API | @file references/api.md |
| Implementation workflow | @file references/workflow.md |
| Task decomposition | @file references/hierarchies.md |
| Good/bad examples | @file references/examples.md |
| Verification checklist | @file references/verification.md |
## In This Reference
| File | Purpose |
|------|---------|
| `references/api.md` | Overseer MCP codemode API types/methods |
| `references/workflow.md` | Start->implement->complete workflow |
| `references/hierarchies.md` | Milestone/task/subtask organization |
| `references/examples.md` | Good/bad context and result examples |
| `references/verification.md` | Verification checklist and process |
Related in AI Agents
skill-development
IncludedComprehensive meta-skill for creating, managing, validating, auditing, and distributing Claude Code skills and slash commands (unified in v2.1.3+). Provides skill templates, creation workflows, validation patterns, audit checklists, naming conventions, YAML frontmatter guidance, progressive disclosure examples, and best practices lookup. Use when creating new skills, validating existing skills, auditing skill quality, understanding skill architecture, needing skill templates, learning about YAML frontmatter requirements, progressive disclosure patterns, tool restrictions (allowed-tools), skill composition, skill naming conventions, troubleshooting skill activation issues, creating custom slash commands, configuring command frontmatter, using command arguments ($ARGUMENTS, $1, $2), bash execution in commands, file references in commands, command namespacing, plugin commands, MCP slash commands, Skill tool configuration, or deciding between skills vs slash commands. Delegates to docs-management skill for official documentation.
reprompter
IncludedTransform messy prompts into well-structured, effective prompts — single or multi-agent. Use when: "reprompt", "reprompt this", "clean up this prompt", "structure my prompt", rough text needing XML tags and best practices, "reprompter teams", "repromptception", "run with quality", "smart run", "smart agents", multi-agent tasks, audits, parallel work, anything going to agent teams. Don't use when: simple Q&A, pure chat, immediate execution-only tasks. See "Don't Use When" section for details. Outputs: Structured XML/Markdown prompt, quality score (before/after), optional team brief + per-agent sub-prompts, agent team output files. Success criteria: Single mode quality score ≥ 7/10; Repromptception per-agent prompt quality score 8+/10; all required sections present, actionable and specific.
adaptive-compaction
IncludedAdaptive add-on policy and recovery layer that decides WHEN to compact, prune, snapshot, or fork -- replacing fixed-percent auto-compaction across Claude Code, Codex, and MCP-capable hosts. Trigger on auto-compact timing or damage: "when should I compact", "is it safe to compact now or start a fresh session", "auto-compact fires too early/mid-task", "switching to an unrelated task but the window still has space", "context rot", "answers get worse the longer the session runs", "the agent forgot the plan or my decisions after it summarized", "add a layer on top that manages context without changing the agent", raising autoCompactWindow to give the policy room, or installing/tuning a cross-tool compaction policy or PreCompact hook -- even when "compaction" is never said but the problem is context-window pressure or post-summarization memory loss. Do NOT use to summarize a conversation, build RAG, write a summarization prompt (decides WHEN not HOW), or answer max-context-length trivia.
agent-skill-creator
IncludedCreate cross-platform agent skills from workflow descriptions. Activates when users ask to create an agent, automate a repetitive workflow, create a custom skill, or need advanced agent creation. Triggers on phrases like create agent for, automate workflow, create skill for, every day I have to, daily I need to, turn process into agent, need to automate, create a cross-platform skill, validate this skill, export this skill, migrate this skill. Supports single skills, multi-agent suites, transcript processing, template-based creation, interactive configuration, cross-platform export, and spec validation.
llm-wiki
IncludedUse when building or maintaining a persistent personal knowledge base (second brain) in Obsidian where an LLM incrementally ingests sources, updates entity/concept pages, maintains cross-references, and keeps a synthesis current. Triggers include "second brain", "Obsidian wiki", "personal knowledge management", "ingest this paper/article/book", "build a research wiki", "compound knowledge", "Memex", or whenever the user wants knowledge to accumulate across sessions instead of being re-derived by RAG on every query.
skill-master
IncludedAgent Skills authoring, evaluation, and optimization. Create, edit, validate, benchmark, and improve skills following the agentskills.io specification. Use when designing SKILL.md files, structuring skill folders (references, scripts, assets), ingesting external documentation into skills, running trigger evals, benchmarking skill quality, optimizing descriptions, or performing blind A/B comparisons. Keywords: agentskills.io, SKILL.md, skill authoring, eval, benchmark, trigger optimization.