action

A Peekaboo command for invoking named accessibility actions on interface elements, such as pressing a button or changing a value.

In plain words
What is it for?
Use it to press controls, show menus, increment values, and target a specific app, process, window, or previously captured interface snapshot.
Why use it?
It performs supported interface actions through the accessibility system instead of simulating mouse or keyboard input.

Command

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 commands/openclaw/peekaboo/action
Clone the repo
git clone --depth 1 https://github.com/openclaw/Peekaboo
Per session 0 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 505 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 $0.00000 $0.00505
Opus 5 $0.00000 $0.00253
Sonnet 5 $0.00000 $0.00101
Haiku 4.5 $0.00000 $0.00051

Measured 2d ago against content hash 311bd7cb104b, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

action 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 2d 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.

docs/commands/action.md · 39 lines

What it actually says

peekaboo action

action invokes a named accessibility action without synthesizing mouse or keyboard input. The action name is positional; --action remains available as an alternative form.

peekaboo action AXPress --on "$ELEMENT_ID" --snapshot <snapshot-id> --foreground
peekaboo action AXIncrement --on Stepper --app Calculator
peekaboo action --action AXShowMenu --on "$ELEMENT_ID" --pid 1234 --foreground

Use --app, --pid, --window-id, --window-title, or --window-index to resolve the intended target without activating it. Window title/index selectors require an app or PID. Actions that can raise or expose foreground UI, including AXPress and AXShowMenu, require --foreground; state-only actions such as AXIncrement remain background-capable. Foreground mode also permits web-content discovery to focus the page when required. The equivalent MCP tool is action, whose background-only policy refuses foreground-exposing actions before dispatch.

Every action resolves through a current UI snapshot. Pass --snapshot explicitly, reuse the latest unmodified see snapshot, or supply app/PID/window target flags so Peekaboo captures a fresh targeted snapshot before dispatch. If no snapshot can be established, the command refuses rather than searching the user's frontmost app. Process-scoped snapshots pin the process generation; exact-window snapshots additionally pin window identity and bounds. Missing or changed generation evidence is refused before element resolution or dispatch.

Remote element actions require Bridge protocol 1.37 and processGenerationBoundElementMutations; older or receiptless hosts are refused before the request is sent. The final resolved Accessibility element must report the same PID as the snapshot receipt.

Successful JSON and MCP results include the canonical target_identity and target_receipt, including the process-start identity needed to distinguish a live process from PID reuse.

When JSON reports requires_fresh_observation: true, or the host cannot return a canonical outcome, that snapshot remains readable for diagnostics but cannot drive another mutation. Run peekaboo see again and use its new snapshot ID; replaying the old ID is refused before dispatch.

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. 2d ago First seen · 39 lines · 0 tokens per session scan A 311bd7cb104b

Subscribe to this mod's changes

action is a command published in the GitHub repository openclaw/Peekaboo (5,098 stars, last pushed yesterday), licensed MIT. It costs nothing until one of its globs matches a file; then it loads 505 tokens. 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 2026-08-30.