sase-var

name	sase_var
description	Attach named output variables to the current SASE agent run.

Before doing anything else, run this command to record that you are using this skill:

sase skill use sase_var --reason "<one-line reason for using this skill>"

Use this skill when you need a later SASE agent to consume a small string value produced by the current agent, or when you want the value to appear in the Agents-tab metadata for this run.

Workflow

1. Make sure the producing agent has a stable name with %name:<producer> or an agent-name template such as %name:build-@.

2. Set one or more output variables:

   sase var set KEY=VALUE [KEY=VALUE ...]
   

3. In later prompts, wait for the producer before referencing its variables. Every producer's variables live under a single agents dictionary keyed by agent name. For example, %name:build-@ can produce:

   sase var set result_path=dist/report.md status=ok
   

   A later waited agent can render {{ agents["build"].result_path }} after the producer has written the variable.

The key is always the agent's stable name. Agent-name templates use the template base, so %name:build-@ is {{ agents["build"].result_path }}, not build-0. The key is the raw agent name with no identifier munging, so dotted, hyphenated, and digit-leading names all work via bracket access: %name:research.@.final → {{ agents["research.final"].report_path }}, and %name:0n.cld → {{ agents["0n.cld"].report_path }}. Identifier-safe keys also support attribute access such as {{ agents.build.result_path }}.

Rules

• Run this only inside a SASE agent; the command requires SASE_AGENT=1 and SASE_ARTIFACTS_DIR.
• Keys must be valid Jinja attribute identifiers: [A-Za-z_][A-Za-z0-9_]*.
• Values are strings and are split on the first =, so sase var set token=a=b=c stores a=b=c.
• Quote assignments when your shell would otherwise split or expand the value, for example sase var set "summary=tests passed".
• Multiple calls merge into the same agent's variable map; later writes for the same key replace earlier values.
• Do not store secrets. Output variables are persisted in agent_meta.json and shown in ACE.

Use %wait:<producer> when a later agent needs a variable from another agent.

Stopping a %repeat / %r chain with STOP

STOP is a reserved output variable that only affects later %repeat / %r slots. Inside a repeat iteration, run:

sase var set STOP=1

before the iteration completes to skip every remaining repeat slot. Each later slot wakes, sees its repeat predecessor's STOP, finalizes as a successful completed (skipped) slot, and exits without running its prompt. Set STOP when the current iteration determines no further repeat work is needed.

STOP is conservative about truthiness: "", 0, false, no, and off (case-insensitive) are treated as not-stop, so a computed STOP=0 is a safe no-op; any other value stops the chain. It is otherwise an ordinary output variable: agents that simply %wait on this producer (outside a repeat chain) are not affected and can still read {{ agents["name"].STOP }}.