菜单
Friction-removal retrospective for agent sessions. Use when the user says "lube", "$lube", "/lube", "friction", "several frictions were observed", "how do we unblock this", "avoid this in the future", or asks to prevent similar or adjacent blockers across future sessions.
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。
基于 SOC 职业分类
Order user-facing questions by dependency. Use when two or more clarifications may depend on one another, a high-impact fork should be resolved before details, a plan needs a decision-tree interview, or a visual choice should be previewed before asking.
Critically evaluate a planned or existing feature against the project's vision and decide whether to keep, sharpen, park, or cut it — favoring depth on the core loop over breadth of surface area. Use for "is this in scope", "does this fit the vision", "prune scope", "feature creep", "scope creep", "depth not breadth", "should we remove this", "audit features against the vision", "why does this feature exist", "agents keep adding WOW features", "keep this out of scope", maintaining an OUT_OF_SCOPE.md ledger, patrolling a repo for re-added parked/cut features, or gating proposed work items (beads/plans) before they enter the backlog. Read-only by default: it recommends and records; it does not delete code without explicit approval.
Run Codex in a persistent tmux session with signal-based completion so long tasks survive conversation death. Use when work may take 5+ minutes, needs background execution, or must remain retrievable after the calling session ends.
Deploy, rollback, debug, and operate Docker or VPS stacks, Pages or edge frontends, packages, and app-store releases using a skillbox client overlay. Also design or migrate repo-owned local self-test and self-release paths that move avoidable build, test, and deploy work off paid GitHub Actions. Use when deploying code, reducing Actions spend, creating a self-deploy or self-test path, checking health, reviewing logs, comparing env rollout plans, validating package publishing, or working in SSH dev environments where production constraints matter.
Decompose complex work into a Beads-backed execution graph with generated `WORKGRAPH.md` views, then run an NTM-style swarm by ready frontier with no write overlap. Use before parallel execution for large-ish, UI-facing, multi-file, multi-domain, naturally parallel, naturally orchestrated, subgoal-oriented, or review-sensitive tasks.
Plan new multi-repo domain slices, assess plan quality, or orchestrate implementation from an accepted slice plan. Use for "plan the X slice", multi-repo feature planning, API contract design, "implement the X slice", or slice-quality review; not for bug fixes, small refactors, or single-repo work. When a slice premise depends on live external reality rather than internal repo evidence, run a bounded GPT-5 Pro / Deep Research gate before committing to the plan.
| name | lube |
| description | Friction-removal retrospective for agent sessions. Use when the user says "lube", "$lube", "/lube", "friction", "several frictions were observed", "how do we unblock this", "avoid this in the future", or asks to prevent similar or adjacent blockers across future sessions. |
Start the first progress update with the exact prefix Using lube.
Preferred format: Using lube to <goal>. First I will <next concrete step>.
Do not change or omit that prefix. Usage review tooling depends on this stable marker; without it, lube invocations look like path-touch heuristics instead of confirmed skill use.
Several frictions were observed in our session that I believe could have been avoided. How do we unblock this and all similar or adjacent situations in the future?
Use the current session as evidence, then convert each friction into the smallest durable unblocker.
$skill-issue to create or improve a skill when the fix belongs in an agent workflow.For broad "what should we improve?" or "$cass $skill-issue $lube" requests, run the bundled miner instead of hand-assembling the same shell sequence:
python3 scripts/lube_evidence_miner.py --skills cass,skill-issue,lube --since month
The miner combines the portfolio-level skill-issue opportunity scan with
per-skill usage reviews, then emits the lube output shape below. It is evidence
gathering only; after reading it, still execute the smallest safe unblocker
directly when the fix is local and non-destructive.
For "mine cass for repeated agent pain" requests, do not hand-roll 10+
sbp cass search calls; run the frequency mode instead:
python3 scripts/lube_evidence_miner.py --mode frequency --top 5
It batch-searches a curated friction-signal list through the sbp cass search
backend, aggregates hits per pattern across sessions, ranks by
session-frequency × approximate matched tokens, and emits ranked JSON with
pattern, score, session_ids, sample_snippet, and a suggested
lube_target classification ready for bead creation. Add ad-hoc signals with
--terms "signal one, signal two"; override the backend with --cass-command
or LUBE_CASS_COMMAND (used by tests to mock the backend).
When the friction source is an orchestration or Skillbox runtime issue, inspect
recent events deliberately instead of reading the feed from the beginning. Call
skillbox_events once to see total_events, then tail with a cursor near the
end, for example cursor = max(total_events - 100, 0).
Classify repeated pulse.restart_failed or cross-client service-noise entries
separately from target-repo blockers. Only turn them into implementation work
when they explain the current goal's failed proof; otherwise record them as
process noise and keep the closeout focused on the exact approval or runtime
gate that still blocks completion.
When a client-scoped event feed returns rows whose client_id or
detail.client_id differs from the requested client, treat those rows as
cross-client noise unless another field directly explains the current proof
failure. Do not let subject/name matches alone promote another client's pulse
failure into the target repo's blocker list.
Do not turn this into a blame postmortem. Do not stop at advice when a safe concrete fix can be made in the workspace.