菜单
Sweep the project's LEARNINGS.md inbox and turn accumulated entries into durable skill, rule, or documentation changes. Run when ready to process learnings — not mid-task.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
Delegates a task to OpenAI Codex running as an interactive session in a herdr pane - uses the user's ChatGPT subscription, visible in herdr, steerable mid-session. Use when delegating work to codex, offloading a task to a GPT model (sol, terra, luna), or running a second opinion from a non-Claude frontier model. Not for Claude Code's own Agent tool or local-LLM delegation (delegating-to-local-llm).
Delegates a task to a local LLM running as a Pi coding-agent session in a herdr pane - the subagent is visible in herdr, can be steered mid-session, and costs no cloud tokens. Use when delegating work to a local model, spawning a Pi subagent, offloading a task to omlx (qwen, gemma), or running agent work offline. Not for Claude Code's own Agent tool or subagents inside Pi.
Loads the rageatc-code-oss operating rules directly from the plugin when the SessionStart hook has not delivered them. Use when the rageatc-code-oss failsafe line appears in session-start hook output, when the plugin's operating rules are absent from context, or when the injected rules look truncated.
Loads the rageatc-core-oss operating rules directly from the plugin when the SessionStart hook has not delivered them. Use when the rageatc-core-oss failsafe line appears in session-start hook output, when the plugin's operating rules are absent from context, or when the injected rules look truncated.
Raises an artefact's signal-to-noise ratio by removing text that earns nothing for its reader — filler, repetition, over-explanation, ornamental phrasing — and relocating conditional detail to references. Use to slim a skill, agent, or document, cut bloat, tighten prose, or shrink a file that has grown too long. Tuned for SKILL.md files.
Creates a design system for software with a UI. Use when a project has a user interface and architecture is confirmed — whether creating from scratch or extracting patterns from existing code.
| name | codify |
| description | Sweep the project's LEARNINGS.md inbox and turn accumulated entries into durable skill, rule, or documentation changes. Run when ready to process learnings — not mid-task. |
| disable-model-invocation | true |
Read LEARNINGS.md at the project root. Candidates are entries without a Codified → marker. If there are none, report that and stop. If the file doesn't follow the standard ## YYYY-MM-DD — Title entry format (e.g. a numbered takeaways list), treat each discernible item as an entry and place markers where they unambiguously attach to the item.
Group candidate entries that share a root cause or a target file. A pattern across several entries is stronger evidence than any single entry — codify the pattern, not each instance.
The entry recorded the symptom and (usually) the likely cause at capture time. Now do the analysis the capture deliberately skipped: ask why the system allowed the failure, and what change prevents recurrence.
Test every proposed fix: would it have prevented the original incident even if nobody remembered the learning? "Be more careful" or "remember to X" fails this test — put the knowledge where the future reader will already be looking (the skill they will load, the rule that is always in context, the template they will copy). Prefer the smallest change that passes.
Some entries need no fix: a durable technical fact (version pin, API quirk) with no better home is already serving its purpose where it is. Others are not ready — a single incident with an unclear cause stays in the inbox until the pattern repeats.
| Learning kind | Target |
|---|---|
| Workflow or process failure in plugin-orchestrated work | rageatc skill, agent, or rule — record as a proposal; implement in the rageatc workshop, never in the installed plugin |
| Project-specific convention or constraint | Project CLAUDE.md or .claude/rules/ |
| Technical fact about the project's stack | Project docs (ARCHITECTURE.md, reference docs) — or the entry itself if no better home exists |
| Behaviour that should apply across all projects | ~/.claude/CLAUDE.md |
Present the proposals as a compact list — entries covered, root cause, proposed change, target file — and get a decision on each. Implement only what is approved.
Apply approved changes to in-reach targets now. rageatc proposals are handed to the user for the workshop — do not edit plugin files from a consuming project.
Append a marker line to each processed entry:
_Codified → <target file> (YYYY-MM-DD)_ — change applied_Codified → proposal: <target> (YYYY-MM-DD)_ — rageatc proposal recorded, pending workshop_Codified → retained in place (YYYY-MM-DD)_ — the entry is itself the durable recordThis marker is the one sanctioned exception to the file's append-only rule.
Summarise per entry: codified where, proposed for what, or left in the inbox and why.