clickthrough/TODO.md

TODO

Project: Clickthrough v1

Current Status

• Draft implementation plan approved
• Build FastAPI server with screenshot + grid + zoom + actions
• Add auth + safety guardrails (token, dry-run, bounds)
• Add AgentSkill docs for operating the API reliably
• Add Gitea CI workflow for Python syntax safety
• Add usage docs + quickstart
• Run local syntax validation

Notes

• API responses now include request IDs, timestamps, and coordinate metadata
• Local syntax checks passed (py_compile, compileall)
• CI workflow runs syntax checks on push + PR

Next

• Add POST /exec endpoint (PowerShell/Bash/CMD) with timeout + stdout/stderr
• Add exec configuration via env (CLICKTHROUGH_EXEC_*)
• Document exec API + config
• Create backlog issues for OCR/find/window/input/session-state improvements
• Open PR for exec feature branch and review/merge
• Require configured exec secret + per-request exec secret header
• Upgrade skill with verify-before-click rules, confidence thresholds, two-phase risky actions, and Spotify playbook
• Add top-level skill section for instance setup + mini API docs
• Clarify user-owned setup responsibilities vs agent responsibilities in skill docs