eunomia.dev
CLAUDE.md244 starsChanged 5 days agoMITView on GitHub
Guidance for agents (Claude Code, Codex, and others) in this repository. AGENTS.md is a symlink to this file. Source for the eunomia-bpf project website (https://eunomia.dev): tutorials and docs for eBPF, eunomia-bpf, bpftime, and related projects, plus the eunomia-bpf community's site operations, media publishing, and external communications. Public media separates voice from ownership: personal accounts carry the maintainer's judgment; org/project accounts carry formal releases. Eunomia.dev is the institutional canonical archive (eunomia-bpf, AgentSight, ActPlane, bpftime, papers, tutorials, talks), community-first in identity, eBPF…
- Reads credentials
In this repository20 other files it gives its agents.
- app/AGENTS.mdAGENTS.md
- app/CLAUDE.mdCLAUDE.md
- blog-writerSkill
- blog-writing-styleSkill
- content-launch-plannerSkill
- devto-publisherSkill
- eunomia-community-radarSkill
- eunomia-content-patrolSkill
- eunomia-research-reportSkill
- eunomia-social-radarSkill
- hackernews-publisherSkill
- juejin-publisherSkill
- linkedin-publisherSkill
- lobsters-publisherSkill
- medium-publisherSkill
- reddit-publisherSkill
- seo-geoSkill
- xiaohongshu-publisherSkill
- x-publisherSkill
- zhihu-publisherSkill
The file
We split this CLAUDE.md into the rules it gives an agent
3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 31
Rule hygiene: keep rules few and non-contradictory.
- Very specific: line 117
Test with cd app && npm run dev and npm run verify before committing.
- Very specific: line 118
Never change existing public paths, route slugs, or nav hrefs without an explicit request for that exact change.
≈ “Common” is an estimate: at least 1 in 5 of the 28,029 CLAUDE.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
CLAUDE.md
Guidance for agents (Claude Code, Codex, and others) in this repository.
AGENTS.md is a symlink to this file.
Project Overview
Source for the eunomia-bpf project website (https://eunomia.dev): tutorials and docs for eBPF, eunomia-bpf, bpftime, and related projects, plus the eunomia-bpf community's site operations, media publishing, and external communications. Public media separates voice from ownership: personal accounts carry the maintainer's judgment; org/project accounts carry formal releases. Eunomia.dev is the institutional canonical archive (eunomia-bpf, AgentSight, ActPlane, bpftime, papers, tutorials, talks), community-first in identity, eBPF as the technical anchor, AI agents as a primary workload/user. Eunomia Labs, Inc. may be named transparently as steward/builder, but don't turn the site into a sales-first company page; keep pricing/CTAs on product surfaces unless explicitly asked otherwise.
Precedence Rule
Inside this repo's already-authorized goals (site/content changes, scheduled patrols, publishing runs, skill-routed tasks), continue, repair, and deliver automatically — don't stop to ask for approval. Ask or wait only where a genuine external constraint requires it: the team's branch/PR/review process (below), publishing beyond what's already authorized (see Publishing), spending money, or handling credentials. Nothing else is an approval gate. Report a genuine external blocker honestly and keep working on unaffected tasks rather than stalling the whole run.
Rule hygiene: keep rules few and non-contradictory. Before adding a rule here or in a skill, delete or merge an existing one instead of stacking an exception on top. Each rule has one home — skills/automations link to this file instead of restating it. Incident stories and dated logs belong in git history or an issue; a one-line pointer here is enough.
Workflow: Branch, Commit, Push
Work directly on main: no new branches/worktrees, no PRs, unless the user
explicitly asks for one. This checkout may be shared with the user or other
agents, so run git status --short --branch before any branch/stash/rebase/
reset/commit, stage only the explicit paths you intended to change, and
rebase onto origin/main when it has advanced before pushing main directly.
For OSS code/docs/sync/CI/release changes, follow oss-change-workflow for
scope, validation, and review. Every configured agent/model may do real work
within its task's already-authorized scope; coordinator roles only prevent
duplicate work, never reduce another model to read-only. Don't set a timeout
on a subagent — let it finish; stop it only if cancelled, obsolete, or stuck.
Skills & Planning Material
See .agents/README.md for how the skills bridge (submodule + sync script)
works and how to change a shared vs. repo-specific skill. Track short-term
fixes as GitHub issues, not draft docs; only reusable, durable, public-safe
decisions belong in documentation/skills.
Publishing
- Standing authorization: a request to publish/post/submit, or a queue item
marked
排队, authorizes completing that publication end to end (prep, preview, publish, QA, ledger update) without re-asking at the last step — except private messages, follows, likes, votes, account settings, payments, and deletions. Stop only for a draft/preview-only task or a real external blocker (report it, keep working on other eligible tasks). - Medium/DEV.to are API-first: use local
MEDIUM_API_KEY/DEV_TO_API_KEY, never print or commit them; follow with visible-browser QA. Other platforms are visible-browser-only. - Prepare the platform artifact locally first (typically
draft/media/YYYY-MM-DD/<source-slug>/<platform>.md); use the platform editor for import/settings/preview/QA, not large rewrites. Record any new concrete platform problem/workaround in the matching publisher skill instead of a disposable per-run file.
Blog Writing & Confidentiality
blog-writing-style owns finished-prose style (English/Chinese);
blog-writer owns source prep, writing/editing, verification, and
publication-integrity checks — neither duplicates the other.
PUBLIC repository: never write business strategy, fundraising plans, pricing,
customer info, competitive analysis, personal constraints, or papers under
review here (including draft/) — that belongs in the private
~/workspace/eunomia-strategy repo. Site ops, SEO/content plans, and brand
style guidance are fine here.
Tech Stack, Commands, Architecture
Custom Next.js (pages router) + React 19 + Tailwind frontend in app/,
statically exported (output: "export") — not a runtime MkDocs site. Content
is Markdown in docs/** (tutorials/ synced from bpf-developer-tutorial —
edit there, not here); app/scripts/generate-* builds it into JSON artifacts
consumed by the Next.js pages. mkdocs.yaml is the permanent, single source
of truth for site IA/navigation (parsed by app/lib/content/mkdocs-config.ts);
components render generated data from it, never their own route tables or
hard-coded hrefs. Details: app/README.md, app/ARCHITECTURE.md. Deploy via
.github/workflows/app-static-pages.yml (app/out -> GitHub Pages);
mkdocs.yml is legacy/manual-only.
cd app && npm ci # Node.js 22+
npm run dev # dev server :3000
NEXT_PUBLIC_SITE_URL=https://eunomia.dev npm run build # static export -> app/out
npm run lint && npm run typecheck && npm run verify # quality gates (verify = CI)make tutorial # clone/update bpf-developer-tutorial; also: make bpftime / cuda-exp / cupti-expMultilingual: /zh/** mirrors English routes; RSS at feed.xml/zh/feed.xml.
Project home pages live in their own repo's README, not here. eBPF tutorial
code: each tutorial has its own libbpf Makefile (.bpf.c for BPF programs,
.c for user space; make / make clean).
Invariants
- Test with
cd app && npm run devandnpm run verifybefore committing. - Never change existing public paths, route slugs, or nav hrefs without an
explicit request for that exact change; new pages are fine if they don't
move/reparent existing content.
/bpftime/**stays under bpftime.
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.

