Skip to content
Agentmart
Submit a tool

deep-thinking

Hierarchical Deep Thinking plans with dynamic revision and sub-agent delegation

Skill by azerothl

IndexedMIT★ 5
Write a review

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/azerothl-deep-thinking-253fbe

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.

We use your email only for this newsletter. Privacy

About

Hierarchical Deep Thinking plans with dynamic revision and sub-agent delegation

Use cases: agent-building

No 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 and code. Checked Oct 26, 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/deep-thinking

What's inside

Skill name
deep-thinking
Description
Hierarchical Deep Thinking plans with dynamic revision and sub-agent delegation
License
MIT
Instructions
14 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# Deep Thinking

**Language:** English | [Français](../../docs/fr/skills/deep-thinking/SKILL.md)

Activated when `cognitive_mode` is `deep_thinking` (request flag or user phrase).

1. Call `plan.create` first with a **full hierarchical tree** (`steps` + optional `children`) — required before any side effect.
2. **Advisory / evaluation** goals (« if I wanted to… », limitations, evolution requests): plan to *analyze and answer*, then `goal.complete`. Do **not** `module.scaffold` / install unless the user explicitly asks to build now.
3. Mark heavy or parallel nodes; use `plan.delegate_step` with a **short self-contained brief** (≤3 sentences).
4. When a delegated child finishes, the runtime injects `[child-done]` and marks the step Done. If `agent.await` says still running, **retry await** — the child is not blocked; do not redo its notes/scaffold.
5. Keep internal detail in `plan.append_log` — never dump logs into the user-facing answer.
6. `plan.get` when you need to re-read the tree; the UI shows a collapsible plan + light traces.
7. Finish with `goal.complete` when all critical steps are Done (or the evaluation answer is ready).

From azerothl/akasha-os, MIT.

More from azerothl/akasha-os

Signals

Stars
5
Forks
1
Open issues
11
Last commit (this folder)
Sep 14, 2026 (1 month ago)
Last release
Oct 12, 2026 (17 days ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from azerothl

Cite this page

deep-thinking skill. Agent Mart. https://agentmart.dev/skills/azerothl-deep-thinking-253fbe (last updated 2026-10-24)