From csdlc
Execute stories with sub-agents — work the stories, review each PR, wait for human approval before merging. Bundles Steps 3 (execution), 4 (review), and optionally 5 (QA). The supervised execution mode. Use when the user says "execute", "work the stories", "run these", "go build this", or after /breakdown produces stories.
How this skill is triggered — by the user, by Claude, or both
Slash command
/csdlc:csdlc-executeThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Execute the stories from: **$ARGUMENTS**
Execute the stories from: $ARGUMENTS
/execute is the supervised execution mode — Steps 3, 4, and optionally 5 bundled together. Work the stories with sub-agents, review each PR, and wait for human approval before merging or moving on.
| Command | Scope | Human gate | Merges? |
|---|---|---|---|
/execute | Multiple stories | Yes — waits for approval per PR | Only when told to |
/shipit | Full epic | Minimal — autonomous hurricane | Yes, auto-merges between batches |
/execute is for when you want to stay in the loop. /shipit is for full send.
Read the stories. From the epic doc in Foundry or $ARGUMENTS:
Present the plan. Show the user:
Stories to execute:
1. FND-E12-S1: Schema + DAOs (foundation, goes first)
2. FND-E12-S2: GitHub OAuth (depends on S1)
3. FND-E12-S3: .well-known endpoints (parallel with S2)
I'll work each story, open a PR, and wait for your go-ahead
before merging. Ready?
Execute each story. For each story in order:
a. Craft the agent prompt (using csdlc-craft-agent-prompt skill)
b. Spawn a sub-agent in an isolated worktree
c. When the intern completes, run the review checklist (Step 4):
PR ready: FND-E12-S1 — Schema + DAOs
Branch: fnd-e12-s1-schema-daos
PR: #15
Review: PASS (all criteria met)
Tests: 12 new, 188 existing — all green
Approve and merge? Or want to review first?
On approval: Merge the PR, move to the next story. On rejection: Note the feedback, fix or re-dispatch, present again.
Between stories: If the next story depends on a just-merged PR, pull main and verify the build before continuing.
When all stories complete: Run /present to package the full set for final review.
When stories can run in parallel (no dependencies between them):
/shipit territory. /execute always waits.Provides behavioral guidelines to reduce common LLM coding mistakes, focusing on simplicity, surgical changes, assumption surfacing, and verifiable success criteria.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub danhannah94/claymore-plugins --plugin csdlc