projectops
AGENTS.md6 starsChanged 32 days agoMITView on GitHub
projectops is three things in one repository: Do not read documentation to find options. Ask the CLI: Rules for running it non-interactively: Read CONTRIBUTING.md and ARCHITECTURE.md (code map and "how to add a flag / option / provider / language / skill") first. The detailed maintainer rules are in CLAUDE.md (Korean); the ones that bite most: Skill bodies live at skills/{name}/SKILL.md. If one applies, read it and follow it before you act. No slash-command UI is needed; read the file…
- Deletes or force-pushes
Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.
Of the AGENTS.md files we read, the 31,542 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.
In this repository17 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- llms.txtllms.txt
- pro-agent-testSkill
- pro-changelog-deploySkill
- pro-commitSkill
- pro-desktop-controlSkill
- pro-figmaSkill
- pro-githubSkill
- pro-init-worktreeSkill
- pro-launchSkill
- pro-oss-consultSkill
- pro-reportSkill
- pro-reviewSkill
- pro-skill-creatorSkill
- pro-sshSkill
- pro-testcaseSkill
- pro-ui-designSkill
The file
We split this AGENTS.md into the rules it gives an agent
8 rules. About 0 are common, 3 are distinctive, and 5 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 11
Do not read documentation to find options.
- Very specific: line 20
Always pass an explicit --mode and --force.
- Very specific: line 23
Do not edit files under .github/.projectops/ (run logs, baseline, incoming copies).
- Very specific: line 27
Read CONTRIBUTING.md and ARCHITECTURE.md (code map and "how to add a flag / option / provider / language / skill") first.
- Very specific: line 31
Commit through the pro-commit skill.
- Very specific: line 34
Run npm test and python3 -m pytest scripts/tests/ .github/scripts/test/ (the guard checks) and run the change itself before you say it …
- Distinctive: line 39
No slash-command UI is needed.
- Distinctive: line 87
Do not push unless the user explicitly asks for it.
≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
projectops — Agent Instructions
projectops is three things in one repository:
- A GitHub project automation template: versioning, release notes, CI/CD, issue automation (
.github/). - An installer CLI:
npx projectopscopies the template into an existing repo (src/,bin/). - An agent skill package:
skills/pro-*(Claude Code, Codex, Gemini CLI, Cursor, pi).
If you are an AI agent using projectops in a repo
Do not read documentation to find options. Ask the CLI:
npx projectops --mode options --json # every flag and every version.yml key, with values and defaults
npx projectops --mode doctor # read-only diagnosis of an existing installRules for running it non-interactively:
- Always pass an explicit
--modeand--force. Without them the CLI asks questions and blocks. --mode doctorand--mode optionsnever change files. Run them first.- Settings live in
version.ymlundermetadata.template.options. The option table above says which keys are safe to edit by hand. - Do not edit files under
.github/.projectops/(run logs, baseline, incoming copies). They are diagnostics.
If you are an AI agent working on this repository
Read CONTRIBUTING.md and ARCHITECTURE.md (code map and "how to add a flag / option / provider / language / skill") first. The detailed maintainer rules are in CLAUDE.md (Korean); the ones that bite most:
- Work on
develop.maindeploys; never commit to it directly. - Several agents may work in the same tree. Stage only your own paths, never use
git add -A, and never usegit reset --hard,git checkout .,git stash(whole tree) or force push. - Commit through the
pro-commitskill. The commit type decides the release version (feat:= minor,feat!:= major). - Common workflows exist twice:
.github/workflows/and.github/workflows/project-types/common/. Keep them identical. - A new CLI flag or
version.ymloption must also be added tosrc/core/options-schema.js;test/options-schema.test.jsfails otherwise. - Run
npm testandpython3 -m pytest scripts/tests/ .github/scripts/test/(the guard checks) and run the change itself before you say it works. Do not commit new tests.
Skills
Skill bodies live at skills/{name}/SKILL.md. If one applies, read it and follow it before you act.
No slash-command UI is needed; read the file directly.
| Intent | Skill |
| Commit with the issue number filled in | pro-commit |
| Create or manage GitHub issues, PRs, comments, labels, secrets, Actions logs | pro-github |
| Open the release PR from develop to main, retrigger automerge | pro-changelog-deploy |
| Write the implementation report for an issue or PR | pro-report |
| Review code | pro-review |
| Run the app / server / browser and look for bugs end to end | pro-agent-test |
| Launch an emulator, simulator, browser or server and capture it | pro-launch |
| Drive an installer wizard, OS dialog or desktop app on this computer's screen (not a web page or emulator) | pro-desktop-control |
| Compare an implementation with a Figma design | pro-figma |
| Show UI mockups to compare and pick, or request a design | pro-ui-design |
| Diagnose a repository as an open source project | pro-oss-consult |
| Generate QA test cases | pro-testcase |
| Create a git worktree for an issue | pro-init-worktree |
| Run commands on a remote server over SSH | pro-ssh |
| Expose a Synology service on a domain | pro-synology-expose (optional projectops-extras plugin) |
| Audit or refactor AI-written code, check a diff for stale comments, install coding conventions | pro-code-consult (optional projectops-extras plugin) |
| Create, review or improve a skill | pro-skill-creator |
Installing the skills
Claude Code:
claude plugin marketplace add Cassiiopeia/projectops
claude plugin install projectops@projectops-marketplace --scope userCodex: codex plugin marketplace add Cassiiopeia/projectops, then open /plugins and check the projectops entry.
Fallback without the marketplace:
git clone https://github.com/Cassiiopeia/projectops.git ~/.codex/projectops
mkdir -p ~/.agents/skills && ln -s ~/.codex/projectops/skills ~/.agents/skills/projectopsOther IDEs: npx projectops --mode skills.
Repository safety
This repository is also the source that initializes other projects. Agent package files belong here
but must not reach generated projects: AGENTS.md, GEMINI.md, llms.txt, CODE_OF_CONDUCT.md, gemini-extension.json,
.agents/, .claude-plugin/, .codex-plugin/, .cursor/, skills/.
They are removed by .github/scripts/template_initializer.py and skipped by src/core/exclusions.js.
Be careful when editing those two files and .github/workflows/.
Do not push unless the user explicitly asks for it.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.