Skill
Skill lythos-labs/lythoskill/packages/lythoskill-red-green-release/skill
User-acceptance-driven release workflow using heredoc patch files. Each iteration produces a timestamped pr-<timestamp>-<desc>.sh that self-archives after execution. No tag without explicit user LGTM. Supports rollback via archived backups.From its SKILL.md
npx -y skills add lythos-labs/lythoskill --skill skillAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 3 stars3 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.
SKILL.md
3.9 KB, 855 tokens by cl100k_base, as published. Nobody here has run it
Red-Green Release Workflow
No tag without LGTM. Every change is a self-archiving patch. Every state is rollback-able.
Typical Scenarios
This skill shines when the agent does not have direct filesystem or git access to the user's project. Typical setups:
- Web chat + Repomix: User pastes a Repomix dump of their codebase into a web chat. The agent proposes changes as
pr-*.shheredoc patches. The user copies the patch locally and runsbash pr-*.sh. - Distributed / async review: Patches are posted in issues, emailed, or shared via web UI for a human to review and apply manually.
- No git on target: Environments (e.g., bare-metal servers, deployed configs) where
git revertis unavailable —.bakfiles provide rollback.
In these scenarios the agent cannot git commit, git diff, or write files directly. The heredoc patch is the only viable delivery format.
Core Principles
- User acceptance drives releases — "LGTM" / "对了" / "就是这样" → then and only then commit + tag.
- Atomic patches — Each iteration = one
pr-<timestamp>-<desc>.shwith heredoc content replacement. - Self-archiving — Patch copies itself to
archived-patches/and deletes itself after execution. - Rollback-ready — Every patch backs up originals before modifying. Rollback = restore backup.
Workflow (4 Phases)
Phase 1 — Plan: Understand requirement, describe approach, user confirms design.
Phase 2 — Create Patch (dry-run): Write pr-<timestamp>-<desc>.sh containing
backup + heredoc replacement + self-archive logic. Show to user for review.
User says "apply it" → proceed.
Phase 3 — Apply & Test: Execute patch (bash pr-*.sh). Restart service.
User tests. If issues → back to Phase 2 with a new patch. Iterate until green.
Phase 4 — Accept & Tag: User says LGTM →
git add -A
git commit -m "<summary>"
git tag -a <tag> -m "<summary>"
Patch File Format
Naming
pr-<timestamp>-<description>.sh
Forbidden words in description: final, done, fix, ok. These words
encourage skipping the red-green test cycle by implying pre-verified success.
Template
#!/bin/bash
# PR: <timestamp>-<description>
# 1. Create archive directory
mkdir -p archived-patches
# 2. Backup current files (for rollback)
cp src/target.ts "archived-patches/target.ts.$(date +%Y%m%d).bak"
# 3. Apply changes (declarative full-file replacement via heredoc)
cat > src/target.ts << 'PATCH_EOF'
// Full replacement content here — exact file state after patch
PATCH_EOF
# 4. Self-archive: copy patch to archive, then delete from root
cp "$0" "archived-patches/$(basename "$0")"
rm "$0"
⚠️ Heredoc delimiter (
PATCH_EOF) must be quoted in thecatcommand to prevent variable expansion. See references/archive-format.md for multi-file patches and escaping rules.
Supporting References
Read these only when the specific topic arises:
| When you need to… | Read |
|---|---|
| See complete heredoc syntax, multi-file patches, and escaping rules | references/archive-format.md |
| Understand the 4-phase workflow in detail | references/phase-details.md |
| See conversation examples of red-green iterations | references/conversation-examples.md |
What ships with it: 3 files
4.3 KB alongside SKILL.md
references/
- archive-format.md1.6 KB
- conversation-examples.md1.2 KB
- phase-details.md1.6 KB
Gives 0 of the 12 instructions most ship operate skills give in 855 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
- produce one self-archiving patch file per iteration
- back up original files before modifying them
- ensure every state is rollback-able
- write descriptive patch names excluding forbidden words
- show the patch to the user for review before applying
- execute the patch script using bash
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.