Full reference

Run managed skills in a sandbox

Execute an imported skill against a stored input, inspect its completion state, and retrieve the output. Chain steps only after each step has a compatible input and output contract.

Updated

Reviewed against Console v0.2.2. Available actions depend on project access and runtime configuration; check your installed CLI version for command differences.

Prepare the skill and input

Import a reviewed managed skill, upload an input through Storage, and configure any required Vault secrets. Use the actual imported slug and returned storage key.

Run one bounded task

bash
stimulir sandbox run <skill-slug> \
  --input-key <storage-key> \
  --param limit=5 --json

The CLI waits for the task response. Inspect the state and reported error before using its output. Parameter names belong to the chosen skill; limit is illustrative and must be supported by that skill.

Pass a completed output to the next tool

--output-key-only prints the output key for a successful task. Keep stderr and the exit code so a failed run cannot be mistaken for an empty successful output.

Chain compatible skills

bash
stimulir sandbox pipe \
  --input-key <storage-key> \
  --step <first-skill> \
  --step <second-skill> --json

Each step receives the previous step’s output key. The pipeline stops when a step does not complete. Verify formats and required parameters before chaining; two valid skills do not necessarily have compatible schemas.

Inspect runs and retrieve results

bash
stimulir sandbox list --json
stimulir storage get --key <output-key> --output result.json

Retain task IDs, skill revision, parameters and input/output keys with your evidence.

Review access before execution

A sandbox is an execution boundary, not a guarantee that arbitrary imported code is trustworthy. Review network access, workspace secrets and output destinations. Keep secret values and sensitive input out of skill logs.

Command catalogue

Expand a command for its arguments. These commands follow the released CLI source; check stimulir --version and the command’s --help for your installation. Required flags are marked; some operations also validate combinations of inputs.

stimulir sandbox list

List past dedicated-task runs for the active workspace.

--json
Print raw JSON.
stimulir sandbox pipe

Pipe skills so each step's output feeds the next (synchronous, per step).

--step required
Ordered skill step: 'slug' or 'slug:k=v,k2=v2'. Repeat for each step, in order.
--input-key required
Storage key of the initial input.
--output-key-only
On success, print ONLY the bare final output_key (for KEY=$(...) shell capture).
--json
Print raw JSON (ordered per-step results).
stimulir sandbox run

Run a skill against a stored input in an isolated sandbox (synchronous).

skill required
Positional argument. Imported skill slug, e.g. evidence-clip.
--input-key required
Storage key of the input (must be under your workspace prefix).
--param, -P
key=value skill param (repeatable).
--output-key-only
On success, print ONLY the bare output_key (for KEY=$(...) shell capture).
--json
Print raw JSON.

SDK method catalogue

Expand a client to inspect public methods and parameter names. Read each operation’s task guide before supplying a payload or starting work. The SDK returns platform responses; handle errors and inspect completion state before using a result.

client.sandboxSandboxClient
list(workspace_id)
Returns the platform response; inspect its status and error detail.
run(skill, workspace_id, input_key, params)
Returns the platform response; inspect its status and error detail.
pipe(steps, input_key, workspace_id)
Run an ordered list of skills, threading each output into the next.