Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add instructions/fcavalcantirj/claude-code-eyes/claude-mdgit clone --depth 1 https://github.com/fcavalcantirj/claude-code-eyesWrote 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/fcavalcantirj/claude-code-eyes/claude-md)<a href="https://agentmods.dev/instructions/fcavalcantirj/claude-code-eyes/claude-md"><img src="https://agentmods.dev/badge/instructions/fcavalcantirj/claude-code-eyes/claude-md.svg" alt="Measured on agentmods" 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 | $0.01201 | $0.01201 |
| Opus 5 | $0.00600 | $0.00600 |
| Sonnet 5 | $0.00240 | $0.00240 |
| Haiku 4.5 | $0.00120 | $0.00120 |
Grade A, and why
claude-code-eyes CLAUDE.md scanned grade A with 1 finding 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 6d 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
- **Keep the soul.** The reason this beats a one-line `curl | Read` is the three How it starts
The opening of the file, as written. The whole thing — 91 lines — stays where its author put it; the contents beside it link to each section on GitHub.
claude-code-eyes — notes for Claude Code
This repository is a single Claude Code skill (repo root = the skill folder). If you're editing it with Claude Code, keep the invariants below — each was learned the hard way, and the tests/validator won't always catch you breaking them.
What it is
A camera-vision skill. snap.sh grabs a frame from a snapshot-capable camera and
prints its path; Claude Reads the path to see it. SKILL.md is the skill
itself; setup.sh writes the camera config and verifies it; the rest is
docs/metadata.
Load-bearing invariants (do not "simplify" these away)
- The
CCE_KEYSallowlist insnap.shandsetup.shis a security boundary. Config files (./.cce.env,~/.config/claude-code-eyes/config) are parsed, neversourced — a config file must not be able to execute code or set arbitrary variables. The allowlist check runs before the value ever reacheseval/printf -v. Keep it exactly there. - Never ship a backend you haven't actually run. The local
ffmpegwebcam path is documented in the README but deliberately not implemented insnap.sh, because it can't be tested headlessly (macOS gates camera access behind an interactive permission prompt). Mark untested things untested, or leave them out. - Keep the soul. The reason this beats a one-line
curl | Readis the three disciplines inSKILL.md: visual-verify (catch what green tests can't), the wiring-mentor protocol, and "know what your instrument can and cannot see." Don't trimSKILL.mdback down to a webcam grab. snap.shmust stay portable to bash 3.2 (the default/bin/bashon macOS) and Linux. Preserve the empty-array-under-set -uguard (${AUTH_ARGS[@]+"${AUTH_ARGS[@]}"}), avoid GNU-only flags, and write output to${TMPDIR:-/tmp}.
Frontmatter rules (enforced by the skill-creator validator)
SKILL.md frontmatter allows only these keys: name, description, license, allowed-tools, metadata, compatibility. name is kebab-case, ≤ 64 chars.
description is ≤ 1024 chars and must contain NO angle brackets (< >) —
write usage hints as words, not with bracketed placeholders. Validate with the
skill-creator plugin's quick_validate.py <skill-dir> (prints Skill is valid!,
exit 0).
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.
- 6d ago First seen · 91 lines · 1,201 tokens per session scan A cd61ca6b80c2
claude-code-eyes CLAUDE.md is an instructions file published in the GitHub repository fcavalcantirj/claude-code-eyes (44 stars, last pushed 15d ago), licensed MIT. It adds 1,201 tokens to every session, about $0.0060 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
recursive-decomposition-skill AGENTS.md
Instructions for massimodeluisa/recursive-decomposition-skill, covering agents: recursive-decomposition skill, read these first (mandatory), language policy, non-negotiables and commands.
chealth CLAUDE.md
Instructions for danielithomas/chealth, covering claude.md, project purpose, language & tooling, build & test and architecture.
C0BALT_CUT CLAUDE.md
Claude Code instructions for jperrello/C0BALT_CUT, covering project instructions for ai agents, beads issue tracker, quick reference, rules and session completion.
odoo-superpowers CLAUDE.md
Claude Code instructions for omas-odoo/odoo-superpowers, covering odoo superpowers — values & conventions, who this is for, hard rules (machine-enforced), core values and skills.
claude-goal-scaffold CLAUDE.md
Claude Code instructions for wilwaldon/claude-goal-scaffold, covering project: [your project name], stack, structure, design direction and content source.
super-spec CLAUDE.md
Claude Code instructions for lbk-open/super-spec, a project described as: Agent Skills that take a requirement — or a production alert — to a reviewed pull request: multi-agent TDD coding, a parallel review panel with severity-graded verdicts, living OpenSpec-compatible specs, and per-language guardrails.