Download the exact captured file
--- name: omos-retro description: "Retrospect a finished run via $oh-my-openspec:omos-retro — compile the evidence bundle, dispatch the retrospector, relay ≤8 cited findings. Read-only. Invoke via $oh-my-openspec:omos-retro." --- > Codex host adaptation (generated from skills/omos-retro/SKILL.md — source of truth stays there): > - Dispatch roles with `spawn_agent` / `send_input` / `wait_agent` / `close_agent`. Subagents are asynchronous (default up to 6 concurrent threads), so a wave is one batch of `spawn_agent` calls followed by `wait_agent` collection — never poll. > - `openspec_validate` is provided by this plugin's bundled MCP server as `mcp__openspec__openspec_validate` — call it with `cwd` set to your session's project directory (the `change` filter optional); the VALID/INVALID/UNAVAILABLE/ERROR one-look report format is identical to OpenCode. > - This skill is invoked as a `$oh-my-openspec:omos-retro` mention (or picked from /skills). There is no argument substitution: arguments arrive as plain text in the user's message after the mention — read them from there; when absent, follow this skill's elicitation steps. > - Roles run from `~/.codex/agents/omos-*.toml`. If `spawn_agent` reports an unknown agent, stop and tell the user to run `$oh-my-openspec:omos-setup` once. # omos-retro Review the current run for defects in the plugin's prompts and for process friction or verification gaps visible in its evidence. Read-only: this command never edits anything — findings are for a future change. ## Flow ### 1. Compile the evidence bundle From the current session, assemble exactly: - commands run (`$oh-my-openspec:omos-init`, `$oh-my-openspec:omos-plan`, `$oh-my-openspec:omos-apply`, `$oh-my-openspec:omos-archive`) and their targets; - every role dispatch: role, one-line task, final report verbatim; - every scenario verification result and task completion/defer status present in the role reports, including `NO-REGRESSION-TEST:` evidence; - process evidence verbatim: repeated `BLOCKED:` or `FAIL` outcomes, re-dispatches, extra critic/verify rounds, `PLAN DIVERGENCE:` reports, user approvals or corrections, and unresolved scenario/task gaps; - `git diff --stat` at the end of the run. Keep the bundle raw and ordered — quoted lines and reports, not your paraphrase. Include only events and gaps evidenced by this run's artifacts; the retrospector's findings must be quotable from the bundle. ### 2. Dispatch Send the bundle to `omos-retrospector` verbatim. One dispatch, no rounds: the role either returns findings or `No findings — run clean.` ### 3. Relay Present the report as-is. Add nothing — no commentary, no severity re-shuffling, no "suggestions". Present the report as rendered Markdown, never inside a code block. If the user wants fixes, that is ordinary work in the plugin's own repository (`$oh-my-openspec:omos-plan` there, or direct edits), not an action of this command. ## Guardrails - Never edit `agents/`, `skills/`, `core/`, or any project file from this command. - The bundle is the only evidence: do not feed the retrospector interpretations, and do not fill gaps from memory. Prompt and process findings share the eight-finding limit. - An empty run (no role dispatches) → say so and stop; do not invent a session to analyze.