Borrowing it
Nothing to install: this file belongs to VictorNain26/pexels-mcp-server. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/VictorNain26/pexels-mcp-server/main/CLAUDE.mdgit clone --depth 1 https://github.com/VictorNain26/pexels-mcp-serverWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/victornain26/pexels-mcp-server/claude-md)<a href="https://agentmods.dev/instructions/victornain26/pexels-mcp-server/claude-md"><img src="https://agentmods.dev/badge/instructions/victornain26/pexels-mcp-server/claude-md/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/instructions/victornain26/pexels-mcp-server/claude-md"><img src="https://agentmods.dev/badge/instructions/victornain26/pexels-mcp-server/claude-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.03072 | $0.03072 |
| Opus 5 | $0.01536 | $0.01536 |
| Sonnet 5 | $0.00614 | $0.00614 |
| Haiku 4.5 | $0.00307 | $0.00307 |
Grade A, and why
pexels-mcp-server CLAUDE.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 9d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 219 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
Operational guide for AI coding sessions in this repo. Read this first, then load the functional context from the docs listed below.
Read the functional context, do not duplicate it here
For what this server does and how it is wired, consult these — they are the single source of truth:
| Doc | Purpose |
|---|---|
README.md |
User-facing overview, tool table, OAuth flow, deployment guide. |
PRIVACY.md |
What the server processes, stores, logs, and forwards. |
CHANGELOG.md |
History of breaking and behaviour-affecting changes. |
If something is documented there, link from your work — do not paste a re-explanation into this file or into a code comment.
Code conventions in this repo
These are guidelines, not laws. If a rule below blocks a clearly better approach, override it and document the reason in the PR description. Don't keep a bad practice just to honour the letter of this file.
- Match scope. A bug fix touches the bug. A refactor goes in a separate PR.
- YAGNI. Three similar lines beat a premature abstraction. No feature flags or backwards-compat shims when the code can simply change.
- Trust internal, validate at boundaries. Pydantic models gate every tool
argument (
ConfigDict(extra="forbid")); internal helpers do not re-check. - Comments answer the why, never the what. No "added for ticket X", no "used by Y" — those rot. Source of truth for why is the commit message and the linked spec/RFC.
- Errors are agent-actionable, raised from the tool body. Validation
errors flatten via
server.py::_raise_invalid_params; Pexels errors (PexelsAuthError,PexelsRateLimitError,PexelsAPIError) propagate as-is. FastMCP catches each and marks theCallToolResultwithisError=trueper MCP spec 2025-11-25 SEP-1303 — never catch and return a string, that would silently mark the result as success. - Tool docstrings are written for the LLM caller, not the human dev.
Every tool MUST have: a one-line purpose, USE WHEN, DO NOT USE
WHEN, filter recap, return-shape teaser. Follow
Anthropic's "Writing tools for agents".
Keep them tight — every char ships in
tools/liston every conversation init. - Tool annotations are not optional. Every
@mcp.toolcarriesToolAnnotationswithtitle+readOnlyHint. The Anthropic Connector Directory rejects ~30 % of submissions for missing these. - Three MCP primitives, all read-only. This server ships the full
set: 8 tools (model-controlled), 3 resources (URI templates per RFC
6570), 2 prompts (claude.ai connector menu). See
README.mdfor the surface. Future primitives go through the same trim discipline — any type-level docstring leaks into$defsof every referencing schema. - No hand-rolled OAuth, no hand-rolled bearer. The MCP SDK owns the auth
surface. The only custom piece is
auth.py'sOAuthAuthorizationServerProviderimplementation and the public landing page atGET /registered via@mcp.custom_route. - No outbound URL fetching from tools. Removing the
pexels_preview_mediatool also removed the SSRF surface. Future tools that take URLs MUST allowlist hosts at the Pydantic layer; document the threat model. - Imports stay narrow.
mcp,httpx,pydantic,uvicorn,redis,cryptographyare the only runtime deps. The last two are gated behindREDIS_URL— the import path instorage.pyis lazy so a stdio install never pays the import cost. Anything new needs a justification in the PR.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 9d ago First seen · 219 lines · 3,072 tokens per session scan A 04305d549ebc
pexels-mcp-server CLAUDE.md is an instructions file published in the GitHub repository VictorNain26/pexels-mcp-server (0 stars, last pushed 1mo ago), licensed MIT. It adds 3,072 tokens to every session, about $0.0154 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
next.js AGENTS.md
AGENTS.md instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.
codex AGENTS.md
AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.
vscode buildNext.instructions.md
Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).
vscode oss-third-party-notices.instructions.md
Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, applying the notice (cutover) and fallback chain (never fail the build).
langchain AGENTS.md
AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.
deepseek-harness AGENTS.md
AGENTS.md instructions for deepseek-ai/deepseek-harness, covering agents.md, pre-stable apis and released session data, repository layout, commands and host sandbox failures.