how-to-isolated
When a task would pollute the main context with intermediate results, file reads, or speculative exploration, consult this skill to decide whether to delegate to the isolated subagent and how to craft the prompt.
What this skill does
# Handling context-heavy tasks
Delegate to the `isolated` subagent (Agent tool, `subagent_type: isolated`) when a task's working material is large but the answer is small. The fork absorbs the reading, searching, and reasoning; only the structured writeup comes back.
This is the right tool when no topic-specific subagent applies (commit, code-review, refactor, plan-refactor each have their own — prefer those when they fit). The `isolated` subagent is the general-purpose fallback with a tighter return-contract than the built-in `general-purpose` agent.
## When to delegate vs. handle inline
Delegate when:
- The task requires reading many files but the conclusion fits in a paragraph (e.g. "where is feature X implemented?", "which callers depend on Y?", "summarize this 2000-line config").
- An exploration might hit dead ends or produce false leads that would clutter the main thread.
- The work is *parallelizable* with the main task — fire and continue.
- The user explicitly wants something investigated without disturbing the current line of work.
Handle inline (do not delegate) when:
- The task is a single-file read or a one-shot grep. Delegation overhead exceeds the context cost.
- The work is iterative back-and-forth with the user — the subagent runs once and returns.
- A topic-specific subagent fits (commit, code-review, refactor, plan-refactor). Use the specific one.
- The user wants to *see* the intermediate steps (e.g. they are learning the codebase). Isolation hides the work.
## Crafting the delegation prompt
Hand the subagent a clear deliverable, not a process. State the question; let it choose the path. Include any starting points (file paths, symbol names, ticket IDs) the user mentioned, since the fork starts with no conversation context.
Examples:
- User: "find out how rate limiting works in this codebase" → "Investigate how rate limiting is implemented in this codebase. Identify the responsible module(s), the algorithm, and where limits are configured. Report file:line references."
- User: "is feature X still wired up?" → "Determine whether feature X is still reachable from the entry points (HTTP routes, CLI commands, scheduled jobs). Report yes/no with file:line evidence; flag any dead branches you find."
- User: "skim this 5000-line log for what failed" → "Examine the log at <path>. Identify the failure, the immediate cause, and any earlier warnings that look related. Report under 200 words; quote the exact failure lines."
## Anti-patterns
- Passing a vague directive ("look at the codebase") with no deliverable. The subagent will return a vague writeup.
- Using `isolated` when a topic-specific subagent fits. Specific agents have richer defaults and tighter contracts.
- Asking the subagent to make permanent changes. Its contract is *investigation and reporting*; if a change is needed, take its findings and act in the main thread (or delegate to an implementation agent).
- Forwarding ongoing conversation context. The fork starts clean; everything the subagent needs must be in the prompt.
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.