career-ops-ui
CLAUDE.md99 starsChanged 8 days agoMITView on GitHub
This file is deliberately a short index, not a knowledge base. Long instructions here get skimmed and then ignored; the detail lives in files that are loaded when they are relevant. - The parent project (..) is read-only except a user-authorised git pull. - Never point a test at the real parent. CAREER_OPS_ROOT=$(mktemp -d) plus dynamic imports inside before(). - Counts come from the live registry, never hand-edited into the gate tests. - Every release ships a QA prompt (qa/QA-REGRESSION-PROMPT-v<version>.md),…
In this repository7 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .github/copilot-instructions.mdCopilot instructions
- swarmSkill
- contributor-prSkill
- hermes-bridgeSkill
- parent-syncSkill
- swarmSkill
The file
Files
career-ops-ui — agent entry point
This file is deliberately a short index, not a knowledge base. Long instructions here get skimmed and then ignored; the detail lives in files that are loaded when they are relevant.
| Read this | When |
| CONTEXT.md | first — domain dictionary. One accepted name per concept. Settles source vs adapter (129 vs 124), mirror vs relay, telegram vs telegram-channel. |
| PROGRESS.md | at session start — current state, next step, known issues, and abandoned approaches you should not retry. |
| docs/adr/ | before changing anything the records cover — why a choice was made, and what would make us revisit it. |
| docs/sdd/CONVENTIONS.md | while writing code — module system, routes, sanitizers, i18n, testing. |
docs/sdd/ — PLAN.md, TRACEABILITY.md, specs/ | before starting work — the program plan, the invariant→gate matrix, and the only specifications folder (docs/sdd/specs/; there is no docs/specs/). |
| AGENTS.md | non-Claude CLIs — same rules, portable phrasing. |
| .claude/skills/ | parent-sync (parity releases), contributor-pr, hermes-bridge. Invoke the skill rather than improvising the pipeline. |
Non-negotiable
- The parent project (
..) is read-only except a user-authorisedgit pull. - Never point a test at the real parent.
CAREER_OPS_ROOT=$(mktemp -d)plus dynamic imports insidebefore(). - Counts come from the live registry, never hand-edited into the gate tests.
- Every release ships a QA prompt (
qa/QA-REGRESSION-PROMPT-v<version>.md), patches included. - Every release runs the full regression on BOTH stands — local
127.0.0.1:4317and prod (resumecraft.ru): all views, ×19 locales, live LLM (remote-qa -f live=true) and scanner (-f scan=true). Findings go to Linear (Russian) immediately and are executed until done. - Every release updates: CHANGELOG ×19 (EN + 18 locales, parity green), the wiki
(Home ×19 banners + counts + Roadmap), the site build (cvstart.org),
PROGRESS.md, and a spec indocs/sdd/specs/with real verification output. Linear statuses move with the work (In Progress → In Review → Done on deployed-and-verified). Full list:docs/sdd/HANDOFF.md→ Post-release standing rules. - Every PR is merged into
mainonce CI is green — squash-merge, branch deleted, work continues from freshmain. No PR lingers open; a later finding is a new commit/PR on top of main, never a stranded branch. - Verify before claiming done. A passing unit suite is not an end-to-end check — see the premature-success note in PROGRESS.md.
Working practice follows Agentic Coding Design Patterns; the patterns this repo implements are listed in the wiki under Agentic Practices.
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.

