rampstackco/claude-skills / skills/dispatch-authoring
dispatch-authoring
VerifiedWrite the structured handoff from a planning session to a coding agent: a dispatch with a header, Gate 0 checks, scope, specification, verification, delivery rules, and stop conditions, so the agent can execute unattended and return a reviewable draft PR. Use this skill whenever the user wants to hand work to an agent, write an agent brief, write instructions for Claude Code, Codex, or a Cursor agent, queue work to run overnight, or turn a plan into something an AI coder can execute without asking questions. Triggers on dispatch, hand off to an agent, agent brief, agent handoff, task brief for an AI coder, instructions for Claude Code, Codex task, Cursor agent, background agent, run this overnight, give this to an agent, unattended run. Also triggers when the user has a plan or a spec and asks how to get an agent to build it, even if they never say dispatch.
| File | Kind | Size |
|---|---|---|
| README.md | markdown | 359 B |
| SKILL.md | skill md | 8.0 KB |
| references/dispatch-template.md | markdown | 3.7 KB |
| references/example-dispatch-new-skill.md | markdown | 5.5 KB |
| references/gate-zero-checklist.md | markdown | 3.1 KB |
| references/pr-body-template.md | markdown | 1.8 KB |
references/example-dispatch-new-skill.md
View on GitHub# Example Dispatch: Add One Skill to a Skills Catalog
Fictional setup: **Quillmark** maintains a public catalog of writing skills for AI assistants in the repo `quillmark-labs/writing-skills`. Each skill is a folder with a `SKILL.md` and a `references/` folder. A generator script builds the README index, and another builds a checksum manifest. The planner wants one new skill, `headline-testing`, added in the "editing" category.
Everything below is fictional.
---
```
# Dispatch: add the headline-testing skill
Date: 2026-11-03
Target repo: quillmark-labs/writing-skills (PUBLIC)
Default branch: main
Branch: feat/headline-testing
Lane: single agent, single worktree, single held draft PR
Planner: Dana (planning session). Executor: coding agent. Merger: Dana only.
## Gate 0 (run before any write)
1. Confirm the remote is quillmark-labs/writing-skills. Stop and report if it resolves anywhere else, including a fork.
2. Confirm the repo is public. Nothing internal goes in files, commits, or the PR body: no internal paths, no customer names, no private repo names.
3. Create a fresh worktree on feat/headline-testing. Never write to the canonical checkout or to main.
4. Read, in this order, before writing:
- AUTHORING.md (structure, frontmatter, voice)
- CONTRIBUTING.md (checklist)
- skills/subject-line-writing/ (closest existing skill; match its depth and tone)
- every file in scripts/; read each docstring
5. Record which script regenerates README.md's index and which regenerates MANIFEST.json, with the flag for each. Run both in check mode on the clean tree and record the result. If either fails on the clean tree, stop and report before writing.
6. If the repo has an agent instructions file, read and follow it.
## Scope
In:
- New folder skills/headline-testing/ with SKILL.md and at least three reference files: a template, a checklist, an example
- README.md index, regenerated by its script
- MANIFEST.json, regenerated by its script
- One CHANGELOG.md entry under the existing Unreleased heading
Out:
- Any other skill's files
- Renumbering skills outside the editing category
- Changes to scripts, CI, or AUTHORING.md
- Formatting fixes anywhere else, even if you notice them
## Specification
Source spec: docs/proposals/headline-testing.md on main. Implement it as written. Do not paste it into the skill.
Constraints:
- Follow AUTHORING.md exactly: required headers in order, the frontmatter fields it lists, SKILL.md under 250 lines.
- Category: editing. Place it immediately after subject-line-writing by display order. Read the existing display order values in the editing category and shift only what is needed.
- The example reference uses a fictional publication. No real outlets or brands.
- Cross-link subject-line-writing and copy-review in "When NOT to use."
Judgment wanted: the description's trigger phrases. Optimize for triggering on casual requests ("which title is better?") without colliding with subject-line-writing. List the phrases you chose and any collisions you checked in the PR body.
## Verification (run all, paste output in the PR body)
1. Lint script, as named in CONTRIBUTING.md. Expected: passes with no errors.
2. README index generator in write mode, then check mode. Expected: check passes.
3. Manifest generator in write mode, then the diff of MANIFEST.json. Expected: exactly one new skill entry, no other entries changed.
4. Read skills/headline-testing/SKILL.md back from disk. Paste the list of its level-two headers. Expected: the required headers in order.
5. Line count for every new file. Expected: within AUTHORING.md limits.
6. Search the new folder for the em dash character. Expected: no matches.
7. Count skill folders before and after, derived from the filesystem. Expected: after equals before plus one. Do not compare against a number from this dispatch.
## Delivery
- Two commits: "feat(skills): add headline-testing" for the skill folder, then "chore: regenerate index and manifest, changelog" for generated files and the changelog.
- Open ONE draft PR titled "feat: headline-testing skill". Body follows the PR body template: summary, scope table, verification output pasted verbatim, Gate 0 record, judgment calls, stop conditions hit, open questions.
- Leave it as draft. Do not request review. Do not merge.
- Cleanup: Dana removes the worktree and branch after merge. Leave both in place.
- Report back: PR URL, verification summary, any stop condition hit.
## Stop conditions
Stop and report rather than proceed if:
- The remote or visibility does not match Gate 0
- No script produces README.md's index or MANIFEST.json
- The lint script rejects the category or a header and AUTHORING.md does not resolve it
- The source spec is missing or contradicts AUTHORING.md
- Any step would require force-pushing, rebasing a shared branch, or touching main
## Open questions deferred to the planner (do not block on these)
- Whether the starter subset of the catalog should include this skill.
```
---
## What makes this dispatch work
- **Gate 0 runs generators on the clean tree first.** A later failure can then be pinned on the change, not on a broken baseline.
- **Scope "out" names the tempting extras.** Formatting fixes elsewhere are the most common agent drift.
- **The spec is pointed at, not pasted.** One source of truth.
- **Judgment is invited in one named place,** with what to optimize for and what to report.
- **Verification step 7 derives the count.** The dispatch never states how many skills exist.
- **Cleanup has an owner,** and the agent is told to leave the worktree in place.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 469k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 321k | Updated | Claude CodeCodex | Verified | Install |