delegating
When assigning, supervising, reusing, restarting, or verifying a sub-agent, choose delegation by context cost, brief it precisely, watch artifacts, and verify…
Skill by DearCaat
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/dearcaat-delegating
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: agent-building
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/delegating
What's inside
- Skill name
delegating- Description
- When assigning, supervising, reusing, restarting, or verifying a sub-agent, choose delegation by context cost, brief it precisely, watch artifacts, and verify what returns.
- License
- MIT
- Instructions
- 33 lines in SKILL.md · view on GitHub
Files (2)
- SKILL.md
- openai.yaml · agents/
SKILL.md (first lines)
# Delegating Place outsourced work with the right worker while keeping its process and conclusions falsifiable. ## Should you delegate? - Use the process-to-conclusion ratio: delegate when substantial scouting, entry-point discovery, parameter trials, or log tracing is needed to obtain the answer, and recover only conclusions and pointers. - Split independent heavy sub-tasks and delegate them in parallel. Do the work yourself when the answer is one command away and dispatch overhead is higher. ## Reuse or restart - Before dispatching, inventory existing workers, their problem domains, and local context. Reuse a worker when the domain and assumptions still hold; restart when the task changes domain, premises changed, old context would mislead, or an independent review is needed. ## Brief shape - Give an executor a concrete outcome and narrow freedom; give a reviewer or judgment-heavy worker the goal, evidence, and boundary without prescribing a mechanical path. - State that its user is the main agent, and tell it to ask rather than guess when only you can answer. - When creating a worker, specify model and effort explicitly; when reusing one, keep its configuration unless it conflicts with the new task. - If the user supplied a sub-agent prompt, execute it verbatim. ## Delegation is not hand-off An active status does not prove useful work or correctness. Judge progress by artifacts, counts, logs, and processes. Before dispatching, define what “done” looks like; take over if that artifact is not produced. ## Verify returns Treat every returned claim as unverified. Check numbers and facts against primary artifacts, challenge the reasons given, and resolve conflicts between results before adopting them. ## Completion criterion Every dispatch or reuse points to a problem domain and context rationale; every adopted conclusion points to primary evidence or an artifact personally checked by the main agent.
From DearCaat/research-foundation-skills, MIT.
More from DearCaat/research-foundation-skills
- 0.574score
writing-skills
DearCaat
When creating, modifying, reviewing, splitting, or merging a skill, maintain its entry point, structure, rule layering, and review discipline.
SkillBFreshness: B★ 4Claude Code +9 - 0.568score
writing-skills
DearCaat
写 skill、改 skill、review skill、拆分或合并 skill、把稳定流程沉淀为 skill 时使用;只处理 skill 本身的入口、结构、规则分层与 review,不用于执行 skill 所描述的业务任务本身。
SkillBFreshness: B★ 4Claude Code +9 - 0.566score
debugging
DearCaat
When there is a bug, error, failed test, regression, performance drop, unexpected result, or anomalous data, establish a tight feedback loop and find the root…
SkillBFreshness: B★ 4Claude Code +9 - 0.566score
grilling
DearCaat
When a research design, analysis, outline, revision strategy, or other trade-off needs user alignment, converge one decision at a time with explicit user…
SkillBFreshness: B★ 4Claude Code +9 - 0.564score
truth-seeking
DearCaat
When the user changes direction, requests a decision, restates a preference, or the agent is about to change its position, separate evidence, correction…
SkillBFreshness: B★ 4Claude Code +9 - 0.564score
red-teaming
DearCaat
Explicitly invoked only when the user asks to adversarially test a named plan, claim, draft, or decision; attack its premises with evidence and report what…
SkillBFreshness: B★ 4Claude Code +9 - 0.564score
handoff
DearCaat
Explicitly invoked when the user asks to hand the current session to a later session; compress goals, decisions, progress, artifacts, and next steps into a…
SkillBFreshness: B★ 4Claude Code +9 - 0.564score
grounding
DearCaat
Evidence before opinion: when requirements or context are unclear, or advice is needed, verify primary sources before making a judgment.
SkillBFreshness: B★ 4Claude Code +9 - 0.563score
delegating
DearCaat
委派与协调:准备把活交给 sub-agent、决定续用已有 worker 还是重新开始、已有 sub-agent 在跑需要督办、或它有返回需要核实时使用。不涉及委派、全程自己动手的执行循 executing-plans。
SkillBFreshness: B★ 4Claude Code +9 - 0.561score
truth-seeking
DearCaat
求真校准:当用户改变方向、要求做决策、重申其倾向,或 agent 准备随用户改变判断时自动触发;检查立场变化是否由新证据或原推理纠错驱动,并据实质疑用户。普通事实查证用 grounding,用户取舍收敛用 grilling,显式对抗审视用 red-teaming。
SkillBFreshness: B★ 4Claude Code +9 - 0.561score
brainstorming
DearCaat
Before starting a new research question, analysis, chapter, or idea, expand the candidate hypotheses, research questions, and explanatory angles before…
SkillBFreshness: B★ 4Claude Code +9 - 0.561score
executing-plans
DearCaat
When an agreed plan is ready to implement, execute it faithfully, verify every step, and stop at anything the plan does not cover.
SkillBFreshness: B★ 4Claude Code +9
Signals
- Stars
- 4
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Sep 2, 2026 (1 month ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from DearCaat
- 0.574score
writing-skills
DearCaat
When creating, modifying, reviewing, splitting, or merging a skill, maintain its entry point, structure, rule layering, and review discipline.
SkillBFreshness: B★ 4Claude Code +9 - 0.568score
writing-skills
DearCaat
写 skill、改 skill、review skill、拆分或合并 skill、把稳定流程沉淀为 skill 时使用;只处理 skill 本身的入口、结构、规则分层与 review,不用于执行 skill 所描述的业务任务本身。
SkillBFreshness: B★ 4Claude Code +9 - 0.566score
debugging
DearCaat
When there is a bug, error, failed test, regression, performance drop, unexpected result, or anomalous data, establish a tight feedback loop and find the root…
SkillBFreshness: B★ 4Claude Code +9 - 0.566score
grilling
DearCaat
When a research design, analysis, outline, revision strategy, or other trade-off needs user alignment, converge one decision at a time with explicit user…
SkillBFreshness: B★ 4Claude Code +9 - 0.564score
truth-seeking
DearCaat
When the user changes direction, requests a decision, restates a preference, or the agent is about to change its position, separate evidence, correction…
SkillBFreshness: B★ 4Claude Code +9 - 0.564score
handoff
DearCaat
Explicitly invoked when the user asks to hand the current session to a later session; compress goals, decisions, progress, artifacts, and next steps into a…
SkillBFreshness: B★ 4Claude Code +9
Cite this page
delegating skill. Agent Mart. https://agentmart.dev/skills/dearcaat-delegating (last updated 2026-10-24)