subagent-driven

subagent-driven is a skill for Claude Code, Codex from AyushParkara/syntra. It costs 44 tokens per session (416 once invoked), scanned A, original, Apache-2.0.

A workflow that runs each plan step in a fresh sub-agent, meaning a separate AI worker, and checks the result before continuing.

In plain words
What is it for?
Use it for multi-step coding plans where each task should be isolated, reviewed, and handed off in a short summary.
Why use it?
It keeps unrelated context and failed attempts from carrying into later steps and adds a review checkpoint after each step.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/ayushparkara/syntra/subagent-driven
Any agent
npx skills add AyushParkara/syntra --skill subagent-driven
Clone the repo
git clone --depth 1 https://github.com/AyushParkara/syntra

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for subagent-driven

README.md
[![agentmods](https://agentmods.dev/badge/skills/ayushparkara/syntra/subagent-driven.svg)](https://agentmods.dev/skills/ayushparkara/syntra/subagent-driven)
Your own site
<a href="https://agentmods.dev/skills/ayushparkara/syntra/subagent-driven"><img src="https://agentmods.dev/badge/skills/ayushparkara/syntra/subagent-driven.svg" alt="Measured on agentmods" height="20"></a>
Per session 44 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 416 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00044 $0.00416
Opus 5 $0.00022 $0.00208
Sonnet 5 $0.00009 $0.00083
Haiku 4.5 $0.00004 $0.00042

Measured 5d ago against content hash 49a80e417e57, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

subagent-driven scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

syntra/skills/subagent-driven/SKILL.md · 32 lines

What it actually says

You are executing a plan where EACH step runs in a fresh sub-agent, then gets reviewed before the next starts. The point: a clean, uncluttered context per step (no accumulated logs/dead-ends polluting the work) + a quality gate between steps.

Method:

  1. For the next step, spawn a fresh sub-agent (via task) with a CRISP brief: just this step's goal, the specific inputs it needs, and its done-criterion. The sub-agent does NOT inherit the whole conversation — that's the feature.
  2. The sub-agent does the step and returns its deliverable + evidence.
  3. Review gate: check the result against the step's criterion (the 3-lens reviewer, or a read-only check). Pass → record + continue. Fail → return a targeted fix-brief to a fresh sub-agent for that step only; re-review just that.
  4. Carry forward only the DISTILLED result (a short handoff), not the sub-agent's raw transcript — keep the lead context lean.

Rules:

  • One step = one fresh sub-agent = one clean context. Don't let a step's debris leak into the next.
  • Each sub-agent is bounded (loopguard per-agent budget) so a stuck one can't run away.
  • Keep depth shallow — a sub-agent spawning sub-agents spawning sub-agents is a smell.
  • The lead holds the plan + state; sub-agents are crisp-cut workers.

Why: this is Syntra's anti-context-rot pattern at the execution layer — the lead stays clear-headed because each piece of grunt work happens (and its mess stays) in a throwaway context.

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 5d ago First seen · 32 lines · 44 tokens per session scan A 49a80e417e57

Subscribe to this mod's changes

subagent-driven is a skill published in the GitHub repository AyushParkara/syntra (5 stars, last pushed 25d ago), licensed Apache-2.0. It adds 44 tokens to every session and 416 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories

mcp-scripting

Write mcpScript JavaScript for discovering, inspecting, and calling MCP tools.

nicobailon/pi-mcp-adapter · 21 tokens

adr-skill

Create and maintain Architecture Decision Records (ADRs) optimized for agentic coding workflows. Use when you need to propose, write, update, accept/reject, deprecate, or supersede an ADR; bootstrap an adr folder and index; consult existing ADRs before implementing changes; or enforce ADR conventions. This skill uses…

vercel/ai · 89 tokens

add-harness-package

Guide for adding new AI SDK harness packages. Use when creating a new @ai-sdk/harness- package that adapts a coding-agent runtime to HarnessV1.

vercel/ai · 41 tokens

add-provider-package

Guide for adding first-party AI provider packages to the AI SDK. Use when creating a provider package under packages/ to integrate an external AI service.

vercel/ai · 33 tokens

update-provider-models

Add new or remove obsolete model IDs for existing AI SDK providers. Use when adding a model to a provider, removing an obsolete model, or processing a list of model changes from an issue. Triggers on "add model", "remove model", "new model ID", "obsolete model", "update model IDs".

vercel/ai · 68 tokens

migrate-ai-sdk-v6-to-v7

Migrate applications from AI SDK 6.x to AI SDK 7.0. Use when upgrading Vercel AI SDK packages, fixing v7 migration errors, or when the user mentions AI SDK v6, v7, upgrade, migration, breaking changes, system to instructions, fullStream, telemetry, tool context, or finalStep.

vercel/ai · 79 tokens