plan-hardening
Team-lead delegates plan hardening to crl after the user has already discussed the plan details with crl.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
메뉴
Team-lead delegates plan hardening to crl after the user has already discussed the plan details with crl.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Orchestrate sc-lint-roslyn sprint work where team-lead coordinates, crl is the sole developer, and quality-mgr enforces the QA gate.
Repair Claude teammate routing after same-session compaction or resume when the sc-lint-roslyn team still exists on disk and the saved leadSessionId still matches the current SESSION_ID, but SendMessage or teammate reachability is broken.
Reusable QA orchestration skill for GitHub PRs. Use for multi-pass QA, CI monitoring, and template-driven findings and final quality reports.
Generate a sprint status report for the current phase. Default is --table.
| name | plan-hardening |
| version | 1.2.0 |
| description | Team-lead delegates plan hardening to crl after the user has already discussed the plan details with crl. |
| depends_on | {"codex-orchestration":"0.x"} |
Audience: team-lead only.
Use this only for phase-plan hardening before implementation starts or resumes.
If the user invokes this skill, that means that the plan details have already
been discussed and are fresh in crl context. Do not request details from
the user, the details will surface when the plan is delivered.
team-lead is responsible for routing, worktree creation, and assignment
metadata. team-lead is not the authority for rewriting the plan.
The user-discussed deliverable scope is authoritative. Hardening should welcome
improvements that clarify, tighten, split, or otherwise make the plan more
executable when those improvements are consistent with what the user already
discussed with crl. Hardening must resist and push back on substantial scope
changes that are at odds with that user discussion.
worktree_path and branch are knownsc-compose is availablecrlThe task must end with two required hardening passes:
Together they must produce:
Substantial scope changes are not valid hardening output. Examples:
Any remaining in-scope work without sprint ownership is a GAP. If more
sprints are needed, hardening must create them.
phase_idtask_iddescriptionworktree_pathbranchpr_targetsource_of_truthquestions_or_concernsreferences.claude/skills/plan-hardening/plan-hardening.xml.j2 with
sc-compose for pass 1 (sprint-scope hardening).crl.0.claude/skills/plan-hardening/plan-hardening-consistency.xml.j2
with sc-compose for pass 2 (consistency/ambiguity hardening).crl.0crl if the plan is still ambiguousquality-mgr for
a focused plan QA review.source_of_truth should point at the already-approved planning sources:
crlIf source_of_truth, questions_or_concerns, or repo documents imply a
substantial scope change from what the user already discussed with crl,
that is a stop condition. The hardening pass must not rewrite the sprint scope
to match it. Push back to team-lead, describe the scope conflict explicitly,
and require user discussion before proceeding.
If questions_or_concerns is present, crl should answer it in the ACK.
The ACK should also include a brief outline of the plan/work that crl
understands to be in scope. team-lead should wait for that ACK and outline
before raising scope concerns or discussing adjustments with the user.
After both hardening passes complete, team-lead should do a second,
critical review focused on whether:
quality-mgr can verify themDo not treat the hardening pass itself as the final review. The handoff is:
team-lead routes plan hardening to crlcrl completes sprint-scope hardening to zero findingscrl completes consistency hardening to zero findingsteam-lead critically reviews and pushes back if neededquality-mgr performs focused plan QA after that reviewRender:
.claude/skills/plan-hardening/plan-hardening.xml.j2.claude/skills/plan-hardening/plan-hardening-consistency.xml.j2Example:
sc-compose render \
--root .claude/skills/plan-hardening \
--file plan-hardening.xml.j2 \
--var-file /tmp/plan-hardening-vars.json
sc-compose render \
--root .claude/skills/plan-hardening \
--file plan-hardening-consistency.xml.j2 \
--var-file /tmp/plan-hardening-vars.json
Suggested vars file shape:
{
"task_id": "TASK-1234",
"phase": "phase-S",
"description": "Harden the second half of Phase S before implementation resumes.",
"worktree_path": "/abs/worktree",
"branch": "feature/pS-plan-hardening",
"pr_target": "integrate/phase-S",
"source_of_truth": "- User-approved planning discussion already completed with crl\n- docs/project-plan.md\n- docs/plan-phase-S.md\n- docs/requirements.md\n- docs/architecture.md",
"questions_or_concerns": "- Confirm whether missing follow-on sprints must be created on this branch if the current phase plan stops too early.",
"references": "- docs/project-plan.md\n- docs/plan-phase-S.md\n- docs/requirements.md\n- docs/architecture.md"
}
team-lead input conflicts materially with that scope, stop and push
back instead of normalizing the new scope into the docsteam-lead has critically
reviewed deliverables, sprint splitting, code-sample completeness, and
acceptance criteria