career-ops-ui
AGENTS.md99 starsChanged 8 days agoMITView on GitHub
Read CLAUDE.md for the full project-level instructions, hard rules, conventions, and SDD pipeline. Everything in CLAUDE.md applies equally to Codex, Cursor, Aider, OpenAI's CLI, or any other agent driving this repo. This file exists so non-Claude CLIs find a canonical entry point. opencode.json loads the working set automatically (AGENTS/CONTEXT/PROGRESS + Claude's .claude/PROJECT-CONTEXT.md + the five docs/sdd/ manuals). The operating manual for any agent is docs/sdd/HANDOFF.md (rules, gates, the full ship train including the resumecraft.ru deploy); open work is in docs/sdd/BACKLOG.md,…
- Reads credentials
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 32,671 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 repository7 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- .github/copilot-instructions.mdCopilot instructions
- swarmSkill
- contributor-prSkill
- hermes-bridgeSkill
- parent-syncSkill
- swarmSkill
The file
We split this AGENTS.md into the rules it gives an agent
2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
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
- could be automatic: a setting could enforce it
- Very specific: line 36
Never commit
.env: Use .env.example placeholders only. - Very specific: line 37
Tests must be CI-isolated: Don't assume the parent project layout exists outside tests/fixtures/ or mktemp -d.
≈ “Common” is an estimate: at least 1 in 5 of the 41,581 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
career-ops-ui — Agent Instructions (Codex / generic CLI shim)
Read
CLAUDE.mdfor the full project-level instructions, hard rules, conventions, and SDD pipeline. Everything inCLAUDE.mdapplies equally to Codex, Cursor, Aider, OpenAI's CLI, or any other agent driving this repo. This file exists so non-Claude CLIs find a canonical entry point.
OpenCode and other agents — start here
opencode.json loads the working set automatically (AGENTS/CONTEXT/PROGRESS + Claude's
.claude/PROJECT-CONTEXT.md + the five docs/sdd/ manuals). The operating manual for any agent is
docs/sdd/HANDOFF.md (rules, gates, the full ship train including the resumecraft.ru deploy);
open work is in docs/sdd/BACKLOG.md, the order in docs/sdd/PLAN.md. OpenCode agents:
.opencode/agent/{fixer,locale-docs,release}.md + the three reviewers mirrored from
.claude/agents/ (spa-view-reviewer, test-isolation-reviewer, web-ui-route-reviewer);
commands: /continue, /ship X.Y.Z, /parent-sync, /sdd-status, /codebase-tour, /swarm.
Quick orientation
- This is
career-ops-uiv1.264.2 — an Express + vanilla-JS SPA that puts a polished web interface on top ofFighter90/career-ops. - It runs as a regular Node ≥18 process. It is CLI-agnostic at runtime — the choice of agent CLI affects only the development workflow inside this repo, not the deployed UI.
- Hard rules and parent-project boundary live in
CLAUDE.md(root). Codex's CODEOWNERS-style additions, if any, layer on top.
CLI-specific notes
| CLI | Entry point | Notes |
| Claude Code | CLAUDE.md | Native. Subagents in .claude/agents/, slash commands in .claude/commands/. SDD via gsd-* skills (superpowers@claude-plugins-official). |
| OpenAI Codex | AGENTS.md (this file) | Points to CLAUDE.md. No additional Codex-specific config required. |
| Gemini CLI | GEMINI.md | Points to CLAUDE.md. |
| Cursor / Continue | CLAUDE.md | Both honor CLAUDE.md by convention. |
| Aider | CLAUDE.md | Pass --read CLAUDE.md on first run if your aider version doesn't auto-detect. |
Things every CLI must respect
- Never edit anything outside
web-ui/. Parent career-ops files (../cv.md,../config/,../modes/,../data/,../reports/, …) are off-limits. - Never weaken security headers. CSP excludes
'unsafe-inline'fromscript-srcon purpose. - Never bypass
isValidJobUrl()for SSRF-relevant routes. Same forstripDangerousMarkdown()(CV markdown). - Never commit
.env. Use.env.exampleplaceholders only. - Tests must be CI-isolated. Don't assume the parent project layout exists outside
tests/fixtures/ormktemp -d.
The full list lives in CLAUDE.md under "Hard rules — do NOT violate".
Spec-Driven Development
This project uses the GSD pipeline (Claude-Code-native skills). Non-Claude CLIs can still follow the same discipline manually:
discuss → spec → plan → execute → verify → reviewSee docs/sdd/SDD-GUIDE.md for the full workflow + which artifacts to produce at each step. Specifications live in one folder, docs/sdd/specs/ (template _TEMPLATE.md); the program plan is docs/sdd/PLAN.md and the invariant→gate matrix docs/sdd/TRACEABILITY.md. Codex / Gemini / Aider users: write the artifacts as plain Markdown under .planning/phases/P-NN-<slug>/ and run them through whatever review loop your CLI supports.
Quick reference
| Command | Purpose |
npm start | Run server on 127.0.0.1:4317 |
npm test | Full unit suite (7082 tests as of v1.264.2) |
npm run test:e2e:browser | Playwright browser smoke (opt-in) |
bash bin/start.sh | One-shot launcher with browser open |
For deeper architecture: docs/architecture/OVERVIEW.md.
Working practice
This project follows Agentic Coding Design Patterns (ru) — executable guardrails, packaged workflows, a domain dictionary, ADRs, a progress log and workflow evals. Read CONTEXT.md for the vocabulary, PROGRESS.md for state and abandoned approaches, and docs/adr/ for standing decisions before changing anything they cover. The wiki's Agentic Practices page audits which patterns are actually in place.
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.

