Skills
Companion Agent Skills for chrome-cdp, maintained alongside the CLI so they never drift from the commands they document.
drive-chrome-cdp — the core usage reference for an agent driving the CLI: the list → use → snap → act → verify loop, --by addressing (accessible name, ref, cell, label, in-row), the select verb for prompt/combobox/cascade widgets, session batching, the JSON envelope + exit-code contract, and the passkey/consent rules.
SKILL.md is a thin stub; the body lives in drive-chrome-cdp/references/:
core, widgets, debugging, batch-and-recipes, and examples.
Golden tasks for this skill live in drive-chrome-cdp/evals/evals.json.
check-logged-in — confirm a Chrome tab is signed in to a web app by page content, not URL; a three-state verdict (signed-in / login-page / unknown) other skills branch on before they drive the app.
Builds on drive-chrome-cdp.
fill-grid-and-confirm — fill several cells of a grid-style form with --by cell inside one session, read every cell back, and confirm the save actually persisted before reporting success.
Builds on drive-chrome-cdp.
Using a skill
Each skill is a directory with a SKILL.md (YAML frontmatter + instructions).
Point your agent harness at this skills/ directory, or copy an individual skill into your own skill collection.
chrome-cdp skill --full prints the drive-chrome-cdp core doc plus every reference, straight from the installed binary — run it whenever you want it without leaving the shell, or to confirm a vendored copy hasn't drifted.
chrome-cdp skill list names every embedded skill and drive-chrome-cdp reference; chrome-cdp skill get <name> prints any one of them, addressed by reference name, skill name, or <skill>/<reference>.
This repository is the canonical source for these skills — downstream collections should vendor from here rather than forking, so updates to the CLI and its skills land together.