- add agent inbox delivery, popup annotation routing, and relay or ops session handling upgrades across the plugin and extension
- repair daemon autostart persistence and status reporting while bootstrapping managed sessions from readable system Chrome data
- flush CLI json output on exit so long-running automation can terminate cleanly without truncating responses

Co-authored-by: Codex <noreply@openai.com>

- add checked-in figma file, node, and variable fixtures used by the canvas import coverage
- keep the canvas validation suite reproducible without relying on local-only fixture state

Co-authored-by: Codex <noreply@openai.com>

- improve provider workflows, shopping extraction, and direct-run release harnesses with stronger diagnostics and coverage guardrails
- mirror live command and tool metadata through the help and surface generators while tightening bundled skill resolution
- update CI and release gate automation to validate docs drift, skill assets, and direct-run evidence lanes

Co-authored-by: Codex <noreply@openai.com>

- sync the readme, docs, agents guides, and skill notes to the current canvas, runtime, and workflow surfaces
- capture the March 21 investigation notes and refresh release evidence wording around the direct-run gate policy
- ignore local-only .opendevbrowser data, .tmp directories, and extension checksum sidecars so they stay out of history

Co-authored-by: Codex <noreply@openai.com>

- ignore generated .opendevbrowser runtime artifacts by default
- keep .opendevbrowser/canvas/adapters.json trackable as repo input

Co-authored-by: Codex <noreply@openai.com>

- centralize provider issue classification and summary helpers
- preserve actionable constraint follow-up through shopping, research, product-video, CLI, and direct-run diagnostics
- add regression coverage for auth, render-required, and challenge-constrained flows

Co-authored-by: Codex <noreply@openai.com>

- mark the env-limited provider investigation resolved with current live evidence
- rewrite the anti-bot orchestration investigation around a shared global challenge mode

Co-authored-by: Codex <noreply@openai.com>

- delete the March 21 provider investigation docs
- scrub surviving guidance references to deleted canvas and investigation docs
- keep unrelated planning doc edits unstaged

Co-authored-by: Codex <noreply@openai.com>

- chunk oversized canvas relay responses in the extension runtime
- avoid redundant design-tab syncs for relay overlay commands
- require ops hello acknowledgment before marking relay readiness

Co-authored-by: Codex <noreply@openai.com>

- tolerate globally parsed flags after the artifacts cleanup subcommand
- keep cleanup argument parsing stable while CLI output flags are present

Co-authored-by: Codex <noreply@openai.com>

- raise daemon disconnect timeout when --close-browser is requested
- give the managed live-workflow wrapper extra timeout headroom
- lock both timeout contracts with focused regression tests

Co-authored-by: Codex <noreply@openai.com>

- add targeted branch coverage for ops session recovery and idle disconnect edges
- cover unsupported chrome locator platforms and plain community expansion failures
- keep the repo-wide pressure-test coverage gates above threshold

Co-authored-by: Codex <noreply@openai.com>

- add the anti-bot orchestration Part 2 and implementation plan docs
- align investigation, distribution, and roadmap docs with the latest findings

Co-authored-by: Codex <noreply@openai.com>

- revert the listed anti-bot planning docs from the public PR branch
- preserve those docs for the local main carryover step after remote merge

Co-authored-by: Codex <noreply@openai.com>

- disable global coverage thresholds for focused npm test invocations unless explicitly overridden
- cover the focused-run behavior so the cli-help-parity PR check matches local and hosted execution

Co-authored-by: Codex <noreply@openai.com>