chore(release): codex 1.1.7 (docs)

Author: rldyourmnd

plugins/rldyour-flow/README.md

@@ -5,7 +5,7 @@
 It provides Russian-first command skills:
 
 - `ry-init`: initialize project, module, or feature scope with Serena-first context.
-- `ry-start`: implement a task through context sufficiency, research, plan, worktree, commits, quality gates, reviews, and sync.
+- `ry-start`: implement a task through context sufficiency, research, plan, worktree, commits, quality gates, explicit-review routing, and sync.
 - `ry-newp`: design and optionally scaffold a new project after deep questioning.
 - `ry-review`: review a diff/scope plus affected integration graph.
 - `ry-repair`: normalize stale docs, memories, contracts, hooks, MCP/LSP config, CI, and AI-tool context while asking the owner for ADR/business decisions.
@@ -58,7 +58,7 @@ Loop prevention uses `.serena/.flow_sync_marker`, which is ignored by git. If th
 - `flow-verification-review`
 - `flow-security-review`
 
-`ry-start` and `ry-review` orchestrate these as parallel subagent review tracks when the command workflow calls for review agents. They are not broad implicit-entry skills; this keeps normal Russian prompts routed through command and domain skills first.
+`ry-review` and explicit-review `ry-start` requests orchestrate these as parallel subagent review tracks when the command workflow calls for review agents. They are not broad implicit-entry skills; this keeps normal Russian prompts routed through command and domain skills first.
 
 ## Sources
 

plugins/rldyour-flow/references/flow-lifecycle.md

@@ -49,7 +49,7 @@ Core order:
 8. Create or select branch/worktree and implement through atomic commits.
 9. Run progress checkpoints after meaningful milestones or every 2-3 plan groups: compare implementation against the plan, context pack, existing patterns, and touched integration path.
 10. Run quality gates and fix all issues in touched scope plus integration path.
-11. Run reviewer workflow. Use subagents when the `ry-start` workflow calls for parallel review.
+11. Run reviewer workflow only when the user explicitly asks for review, audit, security review, rules review, or `ry-review`. Use subagents only for that explicit review path.
 12. Run browser/security/design/LSP workflows when triggered by the change type.
 13. Synchronize Serena memories, agent-only files, AGENTS.md/CLAUDE.md when present, git, GitHub, `fullrepo`, and worktree cleanup through `flow-post-task-sync`.
 

plugins/rldyour-flow/references/reviewer-protocol.md

@@ -1,10 +1,10 @@
 # Reviewer Protocol
 
-Reviewer tracks are designed to be run as parallel subagents when `ry-start` or `ry-review` explicitly invokes the review phase.
+Reviewer tracks are designed to be run as parallel subagents when `ry-review` or an explicit-review `ry-start` request invokes the review phase.
 
 ## Subagent Permission
 
-The owner explicitly approved subagent usage for `ry-start` and `ry-review` review phases. Each spawned subagent must receive a self-contained prompt with task, scope, diff, constraints, expected output, and read-only status.
+The owner explicitly approved subagent usage for `ry-review` and for `ry-start` only when the user explicitly asks for review, audit, security review, or rules review. Each spawned subagent must receive a self-contained prompt with task, scope, diff, constraints, expected output, and read-only status.
 
 Managed rldyour reviewer roles are installed as Codex custom agents from `system/agents/*.toml` into `${CODEX_HOME:-$HOME/.codex}/agents/*.toml`. They must use `model = "gpt-5.5"` and `model_reasoning_effort = "medium"` so the parent workflow gets predictable reviewer behavior. If a parent workflow spawns an ad hoc review role that is not backed by a managed agent TOML file, set the spawn override to `model = "gpt-5.5"` and `reasoning_effort = "medium"`.