subagent-driven-development

subagent-driven-development is a skill for Claude Code, Codex from carsteneu/yesmem. It costs 31 tokens per session (2,067 once invoked), scanned A, original, Apache-2.0.

A workflow for carrying out a coding plan by assigning mostly independent tasks to visible helper agents, then reviewing each result for requirements and code quality.

In plain words
What is it for?
Use it to split an implementation plan into parallel tasks, monitor helpers, pass them context, and review or stop them as work progresses.
Why use it?
It reduces the need to coordinate every task manually and makes the agents' work visible in separate terminals.

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/carsteneu/yesmem/subagent-driven-development
Any agent
npx skills add carsteneu/yesmem --skill subagent-driven-development
Clone the repo
git clone --depth 1 https://github.com/carsteneu/yesmem

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-development

README.md
[![agentmods](https://agentmods.dev/badge/skills/carsteneu/yesmem/subagent-driven-development.svg)](https://agentmods.dev/skills/carsteneu/yesmem/subagent-driven-development)
Your own site
<a href="https://agentmods.dev/skills/carsteneu/yesmem/subagent-driven-development"><img src="https://agentmods.dev/badge/skills/carsteneu/yesmem/subagent-driven-development.svg" alt="Measured on agentmods" height="20"></a>
Per session 31 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,067 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 $0.00031 $0.02067
Opus 5 $0.00015 $0.01033
Sonnet 5 $0.00006 $0.00413
Haiku 4.5 $0.00003 $0.00207

Measured 3d ago against content hash fbee2bc25803, 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-development 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 3d 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.

skills/bundled-skills/subagent-driven-development/SKILL.md · 199 lines

How it starts

The opening of the file, as written. The whole thing — 199 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Swarm-Driven Development

Execute plans by dispatching visible YesMem swarm agents per task, with two-stage review (spec compliance, then code quality) after each.

Overrides superpowers:subagent-driven-development — uses YesMem swarm (spawn_agent, scratchpad, send_to) instead of invisible Claude Code Agent tool. All agents run in separate visible terminals.

When to Use

  • Have an implementation plan with mostly independent tasks
  • Want to stay in current session as orchestrator
  • Want to SEE agent work in real-time (vs. invisible subagents)

Core Tools

Tool Replaces Purpose
spawn_agent(project, section, model?) Agent tool Spawn visible agent in own terminal
scratchpad_write(project, section, content) Agent prompt parameter Pass task context to agent
scratchpad_read(project, section) Agent return value Read agent results
send_to(target, content) Inline result Agent notifications
relay_agent(to, content) N/A (new) Inject instructions into running agent
stop_agent(to) Automatic Explicit agent cleanup
list_agents(project) N/A Monitor agent status

Process

digraph process {
    rankdir=TB;

    subgraph cluster_setup {
        label="Setup";
        "Read plan, extract all tasks" [shape=box];
        "Choose project name" [shape=box];
        "Write plan to scratchpad" [shape=box];
        "whoami() for caller_session" [shape=box];
    }

    subgraph cluster_per_task {
        label="Per Task";
        "Write auftrag-impl-N to scratchpad" [shape=box];
        "spawn_agent(section=impl-N)" [shape=box];
        "Wait for DONE via send_to" [shape=box];
        "Handle status" [shape=diamond];
        "Write auftrag-spec-review-N" [shape=box];
        "spawn_agent(section=spec-review-N)" [shape=box];
        "Spec PASS?" [shape=diamond];
        "relay_agent impl-N with fix instructions" [shape=box];
        "Write auftrag-quality-review-N" [shape=box];
        "spawn_agent(section=quality-review-N)" [shape=box];
        "Quality PASS?" [shape=diamond];
        "relay_agent impl-N with quality fixes" [shape=box];
        "stop_agent all three" [shape=box];
    }

    "More tasks?" [shape=diamond];
    "Final review + finishing-a-development-branch" [shape=box style=filled fillcolor=lightgreen];

    "Read plan, extract all tasks" -> "Choose project name";
    "Choose project name" -> "Write plan to scratchpad";
    "Write plan to scratchpad" -> "whoami() for caller_session";
    "whoami() for caller_session" -> "Write auftrag-impl-N to scratchpad";

    "Write auftrag-impl-N to scratchpad" -> "spawn_agent(section=impl-N)";
    "spawn_agent(section=impl-N)" -> "Wait for DONE via send_to";
    "Wait for DONE via send_to" -> "Handle status";
    "Handle status" -> "Write auftrag-spec-review-N" [label="DONE"];
    "Handle status" -> "relay_agent impl-N with fix instructions" [label="NEEDS_CONTEXT"];
    "Handle status" -> "Write auftrag-impl-N to scratchpad" [label="BLOCKED\n(re-dispatch)"];

    "Write auftrag-spec-review-N" -> "spawn_agent(section=spec-review-N)";
    "spawn_agent(section=spec-review-N)" -> "Spec PASS?";
    "Spec PASS?" -> "relay_agent impl-N with fix instructions" [label="FAIL"];
    "relay_agent impl-N with fix instructions" -> "spawn_agent(section=spec-review-N)" [label="re-review"];
    "Spec PASS?" -> "Write auftrag-quality-review-N" [label="PASS"];

    "Write auftrag-quality-review-N" -> "spawn_agent(section=quality-review-N)";
    "spawn_agent(section=quality-review-N)" -> "Quality PASS?";
    "Quality PASS?" -> "relay_agent impl-N with quality fixes" [label="FAIL"];
    "relay_agent impl-N with quality fixes" -> "spawn_agent(section=quality-review-N)" [label="re-review"];
    "Quality PASS?" -> "stop_agent all three" [label="PASS"];

    "stop_agent all three" -> "More tasks?";
    "More tasks?" -> "Write auftrag-impl-N to scratchpad" [label="yes"];
    "More tasks?" -> "Final review + finishing-a-development-branch" [label="no"];
}

Read the full file on GitHub · 199 lines

Files

What ships with it

3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.

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. 3d ago First seen · 199 lines · 31 tokens per session scan A fbee2bc25803

Subscribe to this mod's changes

subagent-driven-development is a skill published in the GitHub repository carsteneu/yesmem (39 stars, last pushed 5d ago), licensed Apache-2.0. It adds 31 tokens to every session and 2,067 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 2026-08-30.

Related

Other skills, from other repositories

linggen

Linggen — durable cross-host memory plus browser control, over two local MCP servers: ling-mem for memory, the Linggen engine for browser, X and agents. Memory: three-tier model (core + long-term + episodic staging) of who the user is, not a log of what was done; same ling-mem daemon and store in Claude Code, Codex…

linggen/linggen-memory · 127 tokens

estimate-tokens

Estimate token count for files using the bundled script. TRIGGER: "estimate tokens", "count tokens", "token usage", "context window size". DO NOT USE: for manual token math — always run the script.

xg-gh-25/SwarmAI · 48 tokens

decision

Full decision lifecycle in VeritasReason � record, query, find precedents (hybrid/advanced), analyze influence, explain, insights dashboard, list, and record exceptions. Uses AgentContext, ContextGraph, DecisionQuery, CausalChainAnalyzer, DecisionRecorder.

bibinprathap/VeritasGraph · 55 tokens

embed

Generate, inspect, and use node/text embeddings in VeritasReason — compute Node2Vec embeddings, find similar nodes, score link predictions, batch similarity, and pairwise similarity. Uses NodeEmbedder, SimilarityCalculator, LinkPredictor, and AgentContext. Sub-commands: compute, similar, similarity, predict-link…

bibinprathap/VeritasGraph · 0 tokens

visualize

Visualize the VeritasReason knowledge graph — topology, centrality, communities, paths, embeddings, decision insights, and temporal evolution. Uses GraphAnalyzer, CentralityCalculator, CommunityDetector, PathFinder, and ContextGraph analytics. Sub-commands: topology, centrality, community, path, decision-graph…

bibinprathap/VeritasGraph · 0 tokens

reason

Run reasoning over the VeritasReason knowledge graph — deductive logic, abductive hypothesis generation, Datalog programs, SPARQL queries, Rete network evaluation. Uses DeductiveReasoner, AbductiveReasoner, DatalogReasoner, SPARQLReasoner, ReteEngine. Sub-commands: deductive, abductive, datalog, sparql, rete, prove…

bibinprathap/VeritasGraph · 0 tokens