agentsclimarketplace
Sign in

Publish bundle

Skill 0SxD/ce-rd-os/ce-rd-os/skills/publish_bundle

Use this skill to package a CE_RD_OS workstream output into the dual-track delivery format: GitHub repo plus 10-file text zip. Trigger when the user says "ship it", "package this", "publish v0.X.Y", "make me a release", "pack the bundle", or when a workstream reaches its termination criteria. Runs verify_self.sh as the privacy and conformance gate, runs pack_text_bundle.sh to produce the zip, fills the approval_gate template honestly, and surfaces to the architect for verdict. Never self-issues the release verdict.From its SKILL.md

Install
npx -y skills add 0SxD/ce-rd-os --skill publish_bundle

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its file declares

Copied from the file, not written here

The file declares its own license as Apache-2.0. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

5.4 KB, ~1.2k tokens by cl100k_base, as published. Nobody here has run it

publish_bundle

Packs a CE_RD_OS workstream output into the dual-track delivery format.

Two artifacts from one source of truth

The repo IS the source of truth. The zip is generated from the repo.

Track A1: GitHub repo

  • The repo at the user's GitHub remote (default https://github.com/<handle>/ce-rd-os).
  • AGENTS.md at root, conformant to the AGENTS.md spec.
  • 6 SKILL.md files under skills/, each conformant to agentskills.io.
  • references/ folder with the 10 supporting documents.
  • LICENSE (Apache-2.0), LICENSE-DOCS (CC-BY-4.0), CHANGELOG.md (Keep a Changelog 1.1.0), NOTICES.md, CONTRIBUTING.md.
  • Makefile, registry.yaml, authors.yaml, .gitignore.
  • .github/workflows/{verify.yml, release.yml}.
  • packets/ for vendored release artifacts and PROCEED.md.
  • scripts/ for verify_self.sh and pack_text_bundle.sh.

Track A2: 10-file text zip

Generated by scripts/pack_text_bundle.sh. Contents:

  1. AGENTS.md
  2. README.md
  3. SETUP.md
  4. system_directive.md (extracted from skills/system_directive/SKILL.md)
  5. setup_intake.md
  6. source_ingestion.md
  7. trinity_rubric.md
  8. workstream_creator.md
  9. publish_bundle.md
  10. PROCEED.md

The zip drops into NotebookLM, Claude project, custom Gem, custom GPT. The host reads AGENTS.md first, then runs setup_intake from setup_intake.md, then proceeds through the workstreams.

Publish-readiness rubric (15 row matrix)

The publish_bundle skill maintains a 15-row publish-readiness matrix that scripts/verify_self.sh executes mechanically.

#TagCheck
1EssentialAGENTS.md exists at repo root
2Essentialeach SKILL.md frontmatter validates
3Essentialskill name field equals parent directory name
4Essentialdescription length 1 to 1024
5EssentialLICENSE recognized header (Apache-2.0)
6Essentialno em or en dashes anywhere
7Essentialno privacy blocklist hits
8ImportantCLAUDE.md and GEMINI.md are 3-line pointers
9ImportantCHANGELOG.md has [Unreleased] and [0.X.Y] sections
10Importantregistry.yaml rows match skills tree
11Importantgit config user.email is not a system default
12Optionalcompatibility lists 3 or more host envs
13Optionalreference docs follow progressive disclosure
14Pitfallno eager multi-body load (no skill imports another's body)
15Pitfallno reference doc body over 5000 lines

verdict: PROMOTE if all Essential pass and no Pitfall fails. verdict: HOLD otherwise.

Procedure

The skill walks these steps in order. Steps stop on first failure.

  1. Read setup configuration from the active workstream.
  2. Run verify_self.sh. If exit non-zero, surface failures to the architect, do not proceed.
  3. Run pack_text_bundle.sh. Produces dist/ce-rd-os-vX.Y.Z-text-bundle.zip.
  4. Generate SHA-256 sidecar for the zip.
  5. Fill approval_gate template at build/APPROVAL_GATE_TEMPLATE.md. Sections 1 through 5 are filled by the agent honestly. Section 6 (architect verdict) is left blank.
  6. Surface to architect. Wait for verdict.
  7. On APPROVED verdict only: tag the release (git tag -a vX.Y.Z), push, run gh release create (or rely on the release.yml workflow triggered by the tag).
  8. On REVISIONS verdict: implement fixes, return to step 2.
  9. On REJECTED verdict: package state into HANDOFF_PACKET_v2.md, surface to architect, stop.

Architect-only verdict

The agent fills the approval gate but does not issue the verdict. The architect issues the verdict via the section 6 / 7 of the template at build/APPROVAL_GATE_TEMPLATE.md.

This is a hard rule. No agent ever auto-promotes a release.

Multi-bundle reuse

This skill works for any CE_RD_OS-shaped bundle, not just CE_RD_OS itself. The Makefile and scripts can be vendored into adjacent bundles (the user's other product tracks) by:

  1. Copying the bundle skeleton (AGENTS.md, registry.yaml, authors.yaml, Makefile, scripts/, .github/workflows/).
  2. Replacing skill bodies with the new bundle's skills.
  3. Updating registry.yaml entries.
  4. Updating CHANGELOG.md, NOTICES.md, README.md.
  5. Running verify_self.sh against the new bundle.

The user has indicated other product tracks will use this same publish flow. The skill is built for that.

Citations

  • AGENTS.md spec, Linux Foundation Agentic AI Foundation.
  • Agent Skills spec, Anthropic, agentskills.io.
  • Keep a Changelog 1.1.0, keepachangelog.com.
  • Conventional Commits 1.0, conventionalcommits.org.
  • SemVer 2.0.0, semver.org.
  • The vercel-labs zip-filename-equals-directory-name convention.
  • The openai-cookbook and anthropics/claude-cookbooks registry.yaml schema.

End publish_bundle.

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Gives 0 of the 12 instructions most ship operate skills give in ~1.2k tokens

Counted across 779 of the 1,178 authors here whose files we hold, read 2026-08-07

  • Document a rollback plan before deploymentin 41 of 779, across 22 files
  • Update the changelogin 21 of 779, across 19 files
  • Run the test suitein 20 of 779
  • Create an annotated git tagin 20 of 779
  • Clean up feature flags after full rolloutin 18 of 779, across 10 files
  • Verify deployment health after launchin 18 of 779, across 10 files
  • Test both feature flag statesin 17 of 779, across 9 files
  • Verify the working tree is cleanin 17 of 779
  • Make database migrations backward-compatiblein 16 of 779, across 8 files
  • Set up error monitoring before launchin 15 of 779, across 7 files
  • Monitor metrics at each rollout stagein 14 of 779, across 5 files
  • Create a GitHub releasein 14 of 779

Said here and by no other author read

  • read setup configuration from the active workstream
  • surface script failures without proceeding
  • run the packer script
  • generate a checksum sidecar for the bundle
  • fill the approval gate template honestly
  • leave section 6 of the approval gate blank

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 326,871. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.