delegate

A work-planning tool that splits a larger task among subagents, which are separate AI workers, using defined inputs, outputs, and rules for combining their results.

In plain words
What is it for?
Use it to decide whether work can be parallelized, assign subtasks, define their contracts, and combine results using a chosen strategy.
Why use it?
It reduces coordination problems when several workers handle parts of the same job. It makes dependencies, failure handling, and result merging explicit.

Skill for Claude CodeCodex

Part of the agent-capability-standard plugin — 42 skills, 2 hooks shipped together

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/synaptiai/agent-capability-standard/delegate
Any agent
npx skills add synaptiai/agent-capability-standard --skill delegate
Clone the repo
git clone --depth 1 https://github.com/synaptiai/agent-capability-standard

Made for: Claude Code, Codex.

Or install agent-capability-standard, the plugin that ships this one along with the rest of its 42 skills, 2 hooks.

Per session 33 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,736 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 $0.00033 $0.02736
Opus 5 $0.00016 $0.01368
Sonnet 5 $0.00007 $0.00547
Haiku 4.5 $0.00003 $0.00274

Measured 3d ago against content hash 73cae0dff7fd, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

delegate 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/delegate/SKILL.md · 386 lines

How it starts

The opening of the file, as written. The whole thing — 386 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Intent

Delegate a complex task to one or more subagents by defining clear contracts, input/output interfaces, and strategies for merging results. Ensure coordinated execution with conflict resolution.

Success criteria:

  • Task decomposed into delegatable subtasks
  • Each subtask has explicit contract (inputs, outputs, constraints)
  • Interface between tasks is well-defined
  • Merge strategy handles conflicts and failures
  • Dependencies between subtasks are clear

Compatible schemas:

  • schemas/output_schema.yaml

Inputs

Parameter Required Type Description
task Yes string or object The overall task to delegate
agents No array Available agents/workers for delegation
constraints No object Global constraints (timeout, resource limits)
merge_strategy No string How to combine results (first_wins, consensus, aggregate)
failure_policy No string What to do on subtask failure (abort, continue, retry)

Procedure

  1. Analyze task: Understand the overall objective

    • Identify the goal and success criteria
    • Determine if task is parallelizable
    • Identify shared state or resources
    • Assess complexity and scope
  2. Decompose into subtasks: Break into delegatable units

    • Each subtask should be independently executable
    • Minimize dependencies between subtasks
    • Identify natural parallelization boundaries
    • Use decompose capability patterns
  3. Define contracts: Specify expectations for each subtask

    • Input: what data/context each subtask receives
    • Output: what each subtask must produce
    • Constraints: limits on time, resources, scope
    • Verification: how to check subtask completion
  4. Design interfaces: Specify data flow between subtasks

    • Format of inputs and outputs
    • Required fields and optional extensions
    • Error formats and status codes
    • Handoff protocols
  5. Plan merge strategy: How to combine results

    • Handle successful completions
    • Resolve conflicts between subtask outputs
    • Aggregate partial results
    • Determine final output format

Read the full file on GitHub · 386 lines

Files

What ships with it

2 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 · 386 lines · 33 tokens per session scan A 73cae0dff7fd

Subscribe to this mod's changes

delegate is a skill published in the GitHub repository synaptiai/agent-capability-standard (4 stars, last pushed 3d ago), licensed Apache-2.0. It adds 33 tokens to every session and 2,736 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.