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 skills add luca-mastrostefano/agent-native-setup --skill extract-profilegit clone --depth 1 https://github.com/luca-mastrostefano/agent-native-setupWrote 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/skills/luca-mastrostefano/agent-native-setup/extract-profile)<a href="https://agentmods.dev/skills/luca-mastrostefano/agent-native-setup/extract-profile"><img src="https://agentmods.dev/badge/skills/luca-mastrostefano/agent-native-setup/extract-profile/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/skills/luca-mastrostefano/agent-native-setup/extract-profile"><img src="https://agentmods.dev/badge/skills/luca-mastrostefano/agent-native-setup/extract-profile.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 | $0.00073 | $0.02786 |
| Opus 5 | $0.00036 | $0.01393 |
| Sonnet 5 | $0.00015 | $0.00557 |
| Haiku 4.5 | $0.00007 | $0.00279 |
Grade B, and why
extract-profile scanned grade B 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 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.
Reads MCP configurationmediumAgent snooping
mcp.json carries server URLs and auth tokens; reading it lets a mod discover and abuse other integrations.
`.mcp.json` registration (env-var token expansion, never a secret) and pre-approve it in How it starts
The opening of the file, as written. The whole thing — 156 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Extract the agent-native setup surface of the repo at the source repo path given in the task into a complete, standalone
agent-native-setup profile (working name: as given, default <repo>-setup), following the
proven procedure below. The profile must reproduce the source's agent workflow faithfully —
parameterized, verified, and ready to publish. Work in a sibling directory, never inside
the source repo.
1. Inventory (fan out, classify everything)
Sweep the source repo for its agent-native surface — not its product code:
- Contracts: AGENTS.md / CLAUDE.md / GEMINI.md (symlinks or shim files?), CONTRIBUTING, SECURITY, code-of-conduct.
- Agent tooling:
.claude/(agents, commands, skills, settings),.cursor/,.gemini/,.github/prompts/, MCP config (.mcp.json— and check whether tools the contract references (e.g.mcp__x__*allowlists) are actually registered anywhere; a permission allowlist is not a registration). - Docs conventions: ADR/RFC systems (index, template, lifecycle rules), architecture docs, onboarding/getting-started docs — separate the conventions and structure (reusable) from the product content (not).
- Gates: git hooks (and how they're armed — husky prepare script? pre-commit install? core.hooksPath?), linter/scanner configs, quality-ratchet files, CI workflows that enforce the contract (vs product build/release pipelines).
- Env surface:
.env.example, secret names in workflows. - Distribution vehicles: some repos are the setup, packaged for a marketplace
(
.claude-plugin/manifests, root-levelskills/dirs). The packaging is SKIP; the payload ships relocated to project-native paths (skills/x→.claude/skills/x). Docs about the setup (README, examples, per-tool install guides) are SKIP too — the profile writes its own README.
Classify every file: VERBATIM (ships as-is), PARAMETERIZE (project name/slug,
instance-bound IDs, org URLs), SKIP (product code, fixtures, lockfiles, generated
artifacts, product-specific doc content). Red-flag pass: embedded secrets/tokens, absolute
paths, personal instance IDs (ticket-system section IDs, dashboards), non-English docs, and
per-machine/local state (.claude/settings.local.json, any *.local.json, .env /
.env.local) — always SKIP and .gitignore these; shipping one commits the adopter's own
approvals and tokens.
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 · 156 lines · 73 tokens per session scan B 4cf7637c6fdc
extract-profile is a skill published in the GitHub repository luca-mastrostefano/agent-native-setup (5 stars, last pushed today), licensed MIT. It adds 73 tokens to every session and 2,786 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it B with 1 finding (reads mcp configuration). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
pentest-credentials
Credential testing methodology — hash crack selection, wordlist generation, password spray (advisory), default cred audit. Triggers on credential testing, hash crack, Hashcat, John the Ripper, password spray, wordlist generation, cupp, CeWL, Crunch, hashid, default password, credential stuffing.
pentest-ctf
CTF (Capture the Flag) challenge solving advisory — HackTheBox, TryHackMe, PicoCTF, web/pwn/rev/crypto/forensics. Triggers on CTF, HackTheBox, HTB, TryHackMe, THM, PicoCTF, OverTheWire, pwn, reverse engineering, crypto challenge, forensics challenge, stego.
pentest-exploit-chain
Multi-step exploit chain analysis — linking low/medium severity findings into a critical chain, stealth+impact scoring advisory. Triggers on exploit chain, attack chain, chain analysis, multi-step attack, kill chain, attack path, stealth scoring, impact analysis.
pentest-orchestrator
Authorized penetration testing engagement orchestrator — scope declaration, OPSEC tagging, evidence-handling discipline. Triggers on pentest, penetration test, red team, engagement, ROE, scope, security assessment, vulnerability assessment, ethical hacking. ADVISORY/DEFENSIVE: does not run live exploits; methodology…
pentest-privesc
Privilege escalation methodology — Linux + Windows + container escape advisory. LinPEAS/WinPEAS analysis, SUID/capability abuse, kernel exploit selection. Triggers on privesc, privilege escalation, LinPEAS, WinPEAS, SUID, capability, sudo abuse, kernel exploit, Windows token, UAC bypass.
pentest-report
Penetration test report writing — executive summary, technical writeup, CVSS scoring, remediation roadmap advisory. Triggers on pentest report, executive summary, technical writeup, CVSS, remediation roadmap, finding writeup, retest report.