From 3f74d242c76a8ee26a5630b742f0e48fcac24d57 Mon Sep 17 00:00:00 2001 From: Nick Josevski Date: Tue, 15 Sep 2026 15:13:03 +1000 Subject: [PATCH] docs: document the Claude step's Response output variable The Run outputs table covered only what you look at on the task page, so the one output a later step can consume went unmentioned. Add it, and spell out the cross-step reference in full. Co-Authored-By: Claude Opus 5 (1M context) --- .../claude-agent-step/getting-started.md | 17 +++++++++++++++-- 1 file changed, 15 insertions(+), 2 deletions(-) diff --git a/src/pages/docs/octopus-ai/claude-agent-step/getting-started.md b/src/pages/docs/octopus-ai/claude-agent-step/getting-started.md index 81afe2b175..1e5031c9ac 100644 --- a/src/pages/docs/octopus-ai/claude-agent-step/getting-started.md +++ b/src/pages/docs/octopus-ai/claude-agent-step/getting-started.md @@ -1,7 +1,7 @@ --- layout: src/layouts/Default.astro pubDate: 2026-07-07 -modDate: 2026-07-09 +modDate: 2026-09-15 title: Getting started with the Claude Agent Step navTitle: Getting Started navSection: Claude Agent Step @@ -86,7 +86,7 @@ When the task finishes, the step's log ends with its usage lines and `Claude Cod ## Run outputs -Every completed run leaves the following outputs on the task page. +Every completed run leaves the following outputs. The first four appear on the task page; the last is an [output variable](/docs/projects/variables/output-variables) that later steps can read. | Output | Where it appears | What it contains | | --- | --- | --- | @@ -94,11 +94,24 @@ Every completed run leaves the following outputs on the task page. | Claude Usage Summary | A panel on the task page | Each Claude step with its model, token count, cost, and any budget cap, plus a total. Use it to keep an eye on what a run costs. | | Artifacts | The task's artifacts | Files the agent attached using the built-in `octopus-artifacts` skill, ready to download. See [Built-in skills](/docs/octopus-ai/claude-agent-step/tools#built-in-skills). | | Transcript | Stored on the Octopus Server, gated behind a dedicated permission | The full, verbose session, recorded for auditing. See [Security & Compliance](/docs/octopus-ai/claude-agent-step/security-and-compliance) for who can read it and how. | +| Response | An output variable set on the step | Everything the agent said, as plain text, for a later step to read. See [Use the response in a later step](#use-the-response-in-a-later-step). | :::figure ![The Claude Usage Summary panel on the task page](/docs/img/octopus-ai/claude-agent-step/claude-usage-summary.png) ::: +### Use the response in a later step + +When the step finishes, it sets the agent's response as an output variable named `Octopus.Action.Claude.Response`. Octopus scopes output variables to the step that set them, so a later step refers to it by the step's name together with the variable's full name: + +```text +#{Octopus.Action[Run Claude Agent].Output.Octopus.Action.Claude.Response} +``` + +Replace `Run Claude Agent` with the name of your agent step. `Octopus.Action.Claude.` appears twice because it belongs to the variable's name, not to the `Output.` prefix. + +The variable holds everything the agent said, across every turn. Its thinking and its tool calls are not included, and stay in the verbose task log and the [transcript](/docs/octopus-ai/claude-agent-step/security-and-compliance). + ## Investigate a failed deployment When a deployment fails, the fastest first responder is one that's already there. Add an agent step that runs only when an earlier step fails, and use it to attach a first-pass diagnosis to the failed task, while the deployment is still in its failed state, instead of waiting for someone to start digging.