round-verification
Verify integrated round evidence and return a machine-readable round result for script import.
git clone --depth 1 https://github.com/SeemSeam/claude_codex_bridge /tmp/round-verification && cp -r /tmp/round-verification/docs/plantree/plans/agentic-loop-workflow/drafts/agentroles.ccb_round_reviewer/skills/round-verification ~/.claude/skills/round-verificationSKILL.md
# Round Verification Use this skill after controller-owned node review, integration, promotion, and project-root verification have produced one compact round evidence packet. ## Workflow 1. Read the task and verification refs, every required compact node-review record, integration order/digest/tests, promotion or rollback evidence, project-root verification evidence, authority checks, and pre-review cleanup/release evidence. The current Round Reviewer may be the sole active dynamic agent; its final release is a controller-owned post-reply gate. 2. Reject missing node review, reviewed-tree mismatch, integration drift, scope violation, hidden fallback, partial promoted delta, rollback drift, missing project-root checks, or unproven cleanup. 3. Return exactly one machine-readable result line as the first non-empty line of the reply. ```text round result: pass|partial|replan_required|blocked ``` Do not write any preamble, heading, Markdown fence, bullet, quote, or backtick before or around that first line. Put evidence and audit details after it. Do not run tests, tools, shell commands, CCB commands, or workflow wrappers before producing the first line; verify from the evidence already supplied. If the evidence is insufficient, start with `round result: blocked`. A later `round result: pass` after prose is invalid and will be blocked by the runner. ## Boundaries - Do not fix code. - Do not run tests or tools. - Do not change product scope. - Do not infer pass without evidence. - Do not submit downstream asks or mark the task or round done. - Do not directly edit authoritative CCB state or runtime files. - Provider and model selection remain project configuration concerns. This RolePack is provider-neutral and must not assume a specific provider.
Send a request to a CCB agent with `ask`.
Submit one node result to the assigned Reviewer and finish only after a bounded pass/rework chain.
Execute one scoped implementation or investigation item and return evidence without changing workflow authority.
Clear CCB managed agent conversation context with `ccb clear`. Use when the user writes `$ccb-clear`, `$ccb_clear`, or asks to clear/reset one or more CCB agent contexts without restarting or deleting project state.
Diagnose and recover CCB communication and reply delivery stalls. Use when a user reports a missing CCB_REPLY, stuck ask, agent stuck busy/delivering, queued work behind an active job, cancelled/incomplete reply, empty artifact, callback not continuing, duplicate retry after success, or a CCB mailbox/communication backend that appears stuck.
Compact CCB managed agent conversation context with `ccb compact`. Use when the user writes `$ccb-compact`, `$ccb_compact`, or asks to summarize one or more mounted agent contexts without restarting agents or deleting project state.
Private built-in CCB configuration skill for agentroles.ccb_self. Design, edit, validate, and prepare reloads for .ccb/ccb.config, role bindings, providers, windows, workspaces, tool windows, sidebar, and provider startup inputs. Use only inside ccb_self; non-self agents should delegate CCB config changes to ccb_self.
Diagnose a named CCB agent by combining authoritative runtime and job lineage with deep read-only pane inspection, apply bounded recovery when evidence supports it, verify the result, and request authorization before submitting a redacted GitHub issue. Use for `$ccb_diagnose agentname`, `$ccb-diagnose agentname`, or reports that a CCB agent is stuck, disconnected, not continuing, not replying, or showing provider errors.