Review a configured queue of pull requests, checking code changes, linked issues, tests, and review discussions. Report actionable findings and supporting evidence, then assess approval or merge separately against the permissions and conditions configured for each action.
Look across feedback, operational data, errors, and delivery failures over a defined period to find recurring problems. Trace patterns to shared causes, examine earlier fixes, and recommend systemic changes, implementing and verifying them only when configured policies allow.
Open a running local app in Design to compare routes and screen sizes, collaborate on visual changes, and hand those changes to a coding agent for application to the source.
Collect unresolved work that needs human judgment from configured pull requests, issues, feedback, errors, and delivery signals. Group related problems into a decision queue with source links, supporting evidence, and visible gaps in coverage.
Guide public skill additions and updates through file placement, catalog documentation, installation discovery, integration settings, and validation. Distinguish ordinary skills from instruction-based, app-backed, and generated skills so each receives the appropriate checks.
Monitor another AI agent’s work and independently investigate the same task. Compare the original request with actual changes and test evidence, report gaps and concrete follow-up steps, and make limited fixes when authorized.
Organize large coding tasks with Claude Fable directing lower-cost subagents. Delegate bounded research, edits, and tests, then have Fable verify evidence, resolve conflicting findings, integrate changes, and review the final result.
Configure a software delivery workflow that turns selected feedback and maintenance signals into verified, reviewed changes. Define information sources, schedules, working directories, and separate permissions for actions such as implementation, merging, deployment, and notifications.
An experimental workflow for finding and resuming coding runs interrupted by a host restart or automation failure. It checks original instructions, worktree ownership, and external state, then reports which runs were resumed, skipped, or held.
Open Agent-Native apps beside the conversation through Dispatch MCP, a connection for app tools. Create or edit content using the current screen and selection, then read saved results back to confirm the changes.
Organize work so an expensive AI model handles planning, judgment, and final review while cheaper helper agents perform bounded research, coding, and testing. Define ownership and verification requirements, then inspect the evidence and integrate their results centrally.
Monitor one explicitly authorized pull or merge request, inspect current reviews and checks, and fix actionable findings within scope. Apply separate permission and verification conditions before publishing fixes, replying, approving, or merging.
Collect product feedback, usage data, runtime errors, and issue reports from configured sources. Classify and group findings with supporting evidence, then apply separate project rules for implementing fixes, requesting clarification, or closing reports.
An experimental workflow for delivering software changes through configured checks, pull requests, review updates, and merge verification. It follows project permissions and release conditions, and reports missing evidence or checks before treating delivery as complete.
Monitor explicitly authorized software delivery tasks, verify which work has stopped, and send reminders only when a concrete next step is due and the configured notification policy allows it.
Compare competing plans from AI agents against the actual task and available evidence. Choose one, combine useful parts, or request revisions, then produce a single execution handoff with ordered steps, verification checks, and reasons for rejected alternatives.
Keep implementation work moving by turning routine uncertainty into explicit assumptions. The agent follows existing project patterns, makes small reversible changes, validates results, and reports its decisions, completed changes, checks, and remaining risks.
Add a short status line to the end of agent responses so readers can see whether work is finished, a specific follow-up remains, or progress depends on their input.
Guide implementation, upgrades, and debugging by checking official documentation and local specifications first. Identify the relevant version, confirm supported behavior, and use those findings to produce an implementation or answer with appropriate validation.
Recover recent work context from local Clips Rewind recordings. Search by words, app, or time range, then inspect relevant text and screen frames to clarify what the user recently saw or said.
Manage long-running or parallel agent work around five-hour and weekly usage limits. Check consumption between batches, pause new work at 95 percent, preserve a clear handoff, and check actual usage again before resuming.
Turns a repeatable workflow from a conversation, skill, spreadsheet, or Claude/ChatGPT project into an Agent-Native app. It creates populated screens with controls that hand tasks to an in-app agent and return results to the relevant objects.
Turn a text development plan into a visual document with diagrams, file maps, annotated code, and open questions. For interface work, add wireframes or an interactive prototype so reviewers can examine screens and behavior before implementation.
Turn a pull request, branch, commit, or git diff into an interactive review overview. Show interface changes, affected files, API behavior, data structures, and annotated code differences so reviewers can understand the overall change before inspecting individual lines.
Is this your account? Link GitHub to bring it onto your creator page (Japanese)