How real projects brief Claude Code: the commands, conventions and warnings they put in CLAUDE.md.
vishalsachdevCLAUDE.mdvishalsachdev/canvas-mcp/CLAUDE.md
unused `requests`; `setup-python@v4 → @v6` (PR #105) - [x] Retired public hosted server (`mcp.illinihunt.org`) — security teardown + cleaned all references (memory, website, README/AGENTS/CHANGELOG) - [x] Issue #115: Gies/Azure hosted deployment — **DONE
bitwardenCLAUDE.mdbitwarden/mcp-server/.claude/CLAUDE.md
supply it via the `unlock` tool (see "Unlock tool" below) - Executes shell commands with security hardening - Returns plain text or JSON responses from CLI **2. API Interface (Organization Administration)** - Makes
hyusapCLAUDE.mdhyusap/deconstructor/CLAUDE.md
word processing logic - `utils/schema.ts` - Core data structure - `utils/supabase/client.ts` - Database client setup - `next.config.ts` - Headers, caching, security config ## Important Notes - Always use Bun commands, not npm - API route has 60s max duration
NVIDIACLAUDE.mdNVIDIA/OSMO/src/ui/CLAUDE.md
process.env.NODE_ENV === 'development')` mock checks - ✅ Use aliasing, separate entry points, build-time substitution **Why:** Security vulnerabilities, bundle bloat, runtime errors. ## SSR/Hydration Safety **CRITICAL:** Hydration mismatches occur when server-rendered HTML
SepineTamCLAUDE.mdSepineTam/mcp-for-stata/CLAUDE.md
tool. Do not add it back to `_TOOL_REGISTRY` without an explicit security review. ## Security invariants - Do-files may execute only from the configured working directory or the Stata
shep-aiCLAUDE.mdshep-ai/shep/CLAUDE.md
every failure**: You are the ONLY developer. Every test failure, CI failure, and security scan failure is YOUR responsibility. NEVER use the words "unrelated", "pre-existing", or "not our changes
HainrixzCLAUDE.mdHainrixz/cyber-neo/CLAUDE.md
analysis agent that runs as a Claude Code plugin. It scans local projects for security vulnerabilities across OWASP 2025 Top 10 and CWE Top 25 categories, then generates prioritized reports
franzvillCLAUDE.mdfranzvill/lad/CLAUDE.md
Zero-configuration discovery via mDNS/DNS-SD (`_lad-a2a._tcp.local`), well-known endpoints (`/.well-known/lad/agents`), and DHCP - Security model assumes local networks are hostile by default ## Common Commands ### Reference Implementation ```bash cd reference
awattarCLAUDE.mdawattar/claude-code-best-practices/CLAUDE.md
general-code-quality-debugger** - Code review, debugging, and quality assessment - **general-technical-project-lead** - Security assessments, strategic decisions, and architectural review **Agent Usage by Command:** - **`/custom-init`**: solution-architect, backend-developer
puritysbCLAUDE.mdpuritysb/AgentDeck/CLAUDE.md
surface is default-deny** (issue #145): the daemon binds `0.0.0.0` on purpose, so the security boundary is the pairing token enforced at one chokepoint per daemon (`bridge/src/http-auth-gate.ts` / `DaemonServer.httpAccessResponse`). An unauthenticated
entropyvortexCLAUDE.mdentropyvortex/meta-llm-charter/CLAUDE.md
patterns: pick one, name the discarded, flag cleanup. Break convention only for correctness or security, named. ## R8 Evidence Load-bearing claims on decision surfaces (report, PR, humanpending.md) carry [executed]|[inspected
a-FigCLAUDE.mda-Fig/Accordion/CLAUDE.md
webToken` is already accepted, which is what makes the door URL session-independent; the security posture is unchanged (same-user local processes could already read this data, hostile web pages
bx33661CLAUDE.mdbx33661/Wireshark-MCP/CLAUDE.md
repository, load the project skill from `skills/wireshark-traffic-analysis/SKILL.md`. Use it for: - `pcap` and `pcapng` analysis - security hunting and incident response - network troubleshooting Prefer the skill playbooks, evidence rubric, and report template
ghostwriternrCLAUDE.mdghostwriternr/claude-code-containers/CLAUDE.md
resources during development ## Current Implementation Status **✅ Completed:** - GitHub App Manifest setup and OAuth flow - Secure credential storage in Durable Objects with AES-256-GCM encryption - Basic webhook processing infrastructure with
0xiehnnktaCLAUDE.md0xiehnnkta/nemesis-auditor/CLAUDE.md
Nemesis Agent ## What This Is **NEMESIS** is an iterative deep-logic security audit agent for [Claude Code](https://docs.anthropic.com/en/docs/claude-code). It orchestrates two sub-agents in an alternating feedback loop
adamjgmillerCLAUDE.mdadamjgmiller/adamsreview/CLAUDE.md
regardless of score; set in Phase 3, re-asserted Phase 4. - **Lanes.** Deep = `correctness` / `security` (Opus 4a, passes through Phase 5). Light = `ux` / `policy` / `architecture` (Sonnet 4b, report-biased; Phase
tinyhumansaiCLAUDE.mdtinyhumansai/opencompany/CLAUDE.md
compromised container can reach every tenant's documents; db-per-tenant stays the security default. See `docs/spec/runtime/storage.md`. Unset (the default) is a full no-op. - The manager should also inject
KorroAiCLAUDE.mdKorroAi/mue-x/CLAUDE.md
algorithms, tools - **Research**: data analysis, papers, experiments, synthesis - **Creative**: writing, design, content generation, storytelling - **Security**: vulnerability analysis, hardening, pen-testing - **Data Science**: ML pipelines, statistics, visualization - **DevOps**: infrastructure, CI/CD, monitoring
cyanheadsCLAUDE.mdcyanheads/git-mcp-server/CLAUDE.md
injection (`tsyringe`). Service registration and tokens. | | `src/utils/` | Global utilities: `internal/` (logger, requestContext, ErrorHandler, performance), `security/` (sanitization), `telemetry/`, `metrics/`. | | `tests/` | Unit/integration tests mirroring `src/` structure. | --- ## III. Tool Development Workflow ### Step
telagodCLAUDE.mdtelagod/code-abyss/CLAUDE.md
installed persona's responses actually honor kernel precedence (correctness > voice, no capitulation openers, security authorization gates, etc. — `skills/_kernel/`) across 10 behavioral probes (`probes.json`). This is a small honest smoke battery
A file Claude Code reads at the start of every session. It holds the commands, conventions and warnings the agent needs for this project.
At the repository root. Claude Code also reads CLAUDE.md files in subdirectories when it works there.
Build and test commands, the project's layout, conventions that aren't obvious from the code, and mistakes to avoid. Short files tend to work better than long ones.
Claude Code reads CLAUDE.md; most other agents read AGENTS.md. Many projects keep one and point the other at it.
