codex-a2a

codex-a2a is a skill for Claude Code, Codex from phamviet86/codex-a2a-gateway. It costs 44 tokens per session (745 once invoked), scanned A, original, Apache-2.0.

A workflow for sending tasks from Codex to a configured local Hermes or A2A peer and receiving the results. A2A means agent-to-agent communication, while Hermes is the connected peer used by this setup.

In plain words
What is it for?
Use it when another configured agent should handle work, especially longer jobs that run asynchronously. It can start, continue, retrieve, wait for, or cancel delegated tasks.
Why use it?
It supports durable delegation without treating uncertain network results as permission to send the same request twice. It also keeps follow-up work tied to the original conversation or task.

Skill for Claude CodeCodex

Written for no agent in particular: nothing here depends on one. Also seen: mentions Codex.

Good fit Use it when another configured agent should handle work, especially longer jobs that run asynchronously. It can start, continue, retrieve, wait for, or cancel delegated tasks.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/phamviet86/codex-a2a-gateway/codex-a2a
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.

Any agent
npx skills add phamviet86/codex-a2a-gateway --skill codex-a2a
Clone the repo
git clone --depth 1 https://github.com/phamviet86/codex-a2a-gateway

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 codex-a2a

README.md
[![agentmods](https://agentmods.dev/badge/skills/phamviet86/codex-a2a-gateway/codex-a2a/github.svg)](https://agentmods.dev/skills/phamviet86/codex-a2a-gateway/codex-a2a)
Your own site
<a href="https://agentmods.dev/skills/phamviet86/codex-a2a-gateway/codex-a2a"><img src="https://agentmods.dev/badge/skills/phamviet86/codex-a2a-gateway/codex-a2a/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for codex-a2a

Your own site · 80×15
<a href="https://agentmods.dev/skills/phamviet86/codex-a2a-gateway/codex-a2a"><img src="https://agentmods.dev/badge/skills/phamviet86/codex-a2a-gateway/codex-a2a.svg" alt="Reviewed on agentmods" width="80" 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 745 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
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 5.1 $0.00044 $0.00745
Opus 5 $0.00022 $0.00373
Sonnet 5 $0.00009 $0.00149
Haiku 4.5 $0.00004 $0.00075

Measured 3d ago against content hash 668549c38d6a, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-17, from the pricing page.

Security

Grade A, and why

codex-a2a 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.

src/codex_a2a_gateway/skills/codex-a2a/SKILL.md · 26 lines

How it starts

The opening of the file, as written. The whole thing — 26 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Use Codex A2A Gateway

Use the gateway when the user asks for the other agent or delegation materially helps the task. Do simple work locally. The outbound tools route to Hermes's active default profile; a requested specialty is an instruction, not selection of a separate profile. Model/reasoning selection is available only on negotiated inbound A2A → Codex requests.

Codex → Hermes through MCP

  • Start with hermes_status when readiness is unknown. If tools are missing, use the installed codex-a2a-setup skill; setup details are not part of each delegation.
  • Call hermes_chat with the task, a stable conversation_key, and an idempotency_key for a new request that could change anything. Separate independent jobs into separate contexts; reuse the returned context_id or conversation key for follow-ups. Include available real origin conversation/question IDs for attribution; never invent a Desktop ID.
  • Long work uses mode="async". Save the returned bridge_task_id; use hermes_task_wait/hermes_task_get to retrieve it. Wait expiry means the task may still be active. Ambiguous transport results are outcome_unknown, not permission to resend. Reconcile only saved task/message identities; preserve ambiguity when evidence is insufficient.
  • For input_required, send the answer through hermes_chat with the existing task_id, same context, and a new idempotency_key for the new answer. Ordinary follow-ups after completion use the same context and a new request identity.
  • Use hermes_tasks_list to find durable handles and hermes_contexts to inspect/close local mappings. Cancellation via hermes_task_cancel is best-effort; never claim computation stopped unless upstream evidence proves it.
  • Consume the retrieved result in the originating task. Where supplied, check origin/result identity and acknowledge using hermes_task_get(acknowledge_result_id=..., expected_origin=...). Results are pulled through get/wait, not automatically delivered into the Desktop conversation.

Read the full file on GitHub · 26 lines

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 · 26 lines · 44 tokens per session scan A 668549c38d6a

Subscribe to this mod's changes

codex-a2a is a skill published in the GitHub repository phamviet86/codex-a2a-gateway (0 stars, last pushed yesterday), licensed Apache-2.0. It adds 44 tokens to every session and 745 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 2026-09-14.