birdhouse

Run, write, and manage Birdhouse agentic tests. Each test exercises Birdhouse end-to-end using a real browser and a real sandbox or explicitly requested isolated OpenCode instance. Use when asked to run agentic tests, run a specific test, or add a new agentic test.

Add isolated Birdhouse web testing to an existing worktree-based agent workflow. Teaches agents to delegate browser validation to child agents, capture screenshots and MP4s, and clean up isolated runs when Cody asks.

Re-apply the Birdhouse opencode fork commits onto a new upstream tag one commit at a time. Load this skill when rebasing Birdhouse's custom opencode patches onto a fresh upstream release.

Compute Fibonacci numbers using recursive agent delegation where each agent spawns children for sub-problems. Use when you want to run a fibonacci test, recursive fibonacci test, or fibonacci test.

Birdhouse frontend architecture patterns for SolidJS and Corvu. Load when working on modals, dialogs, popovers, typeaheads, reactive component trees, buttons and navigation, or any component layering in birdhouse-workspace. Cross-references the generic `solidjs-best-practices` and `corvu-best-practices` skills.

Search message content across Birdhouse agents to find agents by what they did or said. Use when asked to find an agent, search agents, look up what an agent did, or locate a past conversation.

Complete guide to Birdhouse database migrations for agents. Covers both data.db and agents.db — generating migrations, verifying via tests, and applying to specific databases when explicitly asked. Load this skill whenever working on database migrations.

Inspect running production Birdhouse OpenCode instances correctly by mapping workspaces to directories and querying per-directory tool/config state.

List all Birdhouse CI checks and how to run them locally. Load this skill when you need to run CI checks, verify a branch is clean, or check if changes would pass GitHub Actions.

Warm up agents to present their plan and wait for approval before implementation. Use when you want to warm up or warm up an agent before starting work.

Generate information-dense, searchable agent titles optimized for browser tabs and future discoverability. Use for creating agent titles or title generation.

Use this skill when you need to test Birdhouse, AAPI, or plugin behavior in a separate Birdhouse server that you control via curl. Best for isolated read-mode testing, plugin changes, and real end-to-end agent workflows.

Complete guide to Birdhouse logging for agents. Covers where to find logs (dev vs prod, Birdhouse vs OpenCode), how to write logs correctly (subsystems, levels, structured data), and how to tune log verbosity. Load this skill whenever reading, writing, or debugging logs.

Goal-driven UI test suite for Birdhouse. Runs an agent through real user flows in an isolated server using browser automation, producing timestamped artifacts (screenshots, logs, report).

Export complete agent trees to markdown with clickable links and structured transcripts. Use when you need to export agent tree, export full tree, export tree to markdown, export all agents, or export entire tree.

Generate a Birdhouse app usage report from PostHog showing active users, event counts, locations, and install trends across 1/3/7 day windows. Use when asked for usage report, who is using the app, active users, installs, or user activity.

Test agent read modes and exchange reading functionality across multiple agent interactions. Use when you want to run an agent read exchange test, test agent read modes, or test latest exchange reading.