What does this skill do, and when should you use it?
NanoClaw Traffic Guardian is a baseline specification skill, not a working proxy or runtime implementation. It defines host-side proxy inspection, optional HTTPS inspection, exfiltration and inbound-injection detection, redacted local findings, and container-safe MCP and IPC status surfaces. The design is opt-in and detect-and-log by default, with no automatic CA installation or first-version blocking. Adopt it if you are building NanoClaw-specific runtime monitoring; do not treat it as a ready-to-run traffic gateway.
Defines an implementation contract that validates configuration without starting a proxy, starts monitoring through a host-managed lifecycle, keeps CA private keys on the host, inspects HTTP request and response text within a bounded byte limit, supports optional HTTPS MITM with operator-supplied runtime trust configuration, emits JSONL findings with redacted snippets, and exposes status plus redacted findings through MCP. The skill itself does not ship these runtime components.
- NanoClaw builders designing a host-side HTTP inspection layer for containerized agents.
- Security teams wanting to detect outbound exfiltration and inbound prompt injection while logging without blocking.
- Operators needing container-safe IPC or MCP access to monitoring status and redacted findings.
- Teams verifying a standalone release archive and its SKILL.md and skill.json files before installation.
How do you install this skill?
- This is a specification scaffold, not a runnable NanoClaw traffic-monitoring implementation; successful installation must not be treated as active protection.
- HTTPS MITM, proxy lifecycle, IPC, MCP tools, redaction, and detection rules remain to be implemented; the SPEC.md release tests should be completed first.
- Artifact verification depends on GitHub, curl, OpenSSL, jq, and a pinned release key; validate network reachability, key provenance, and safe failure handling.
- No Chinese-language documentation is supplied; Chinese users must independently verify configuration, redaction, and host/container boundaries.
- Shell / CLI
- Network access
- Local filesystem
- MCP Server
Node.js >=18.0.0curlOpenSSLjqshasum or sha256sum
Install with the Vercel Skills CLI:npx skills add prompt-security/clawsec --skill nanoclaw-traffic-guardian -a openclaw -y
For standalone installation or extraction, first run the release-verification procedure included in SKILL.md. It checks the signing-key fingerprint, the signed checksums manifest, the archive, SKILL.md, and skill.json. The source does not document a final deployment directory.
tmp="$(mktemp -d)"
git clone --depth 1 https://github.com/prompt-security/clawsec.git "$tmp"
mkdir -p ~/.claude/skills
cp -R "$tmp/skills/nanoclaw-traffic-guardian" ~/.claude/skills/
rm -rf "$tmp"Generated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Use it as the NanoClaw landing zone for runtime traffic monitoring and read SPEC.md before implementing. Build the specified components under lib/, host-services/, mcp-tools/, and test/; the first implementation should validate configuration, start monitoring through a host-managed lifecycle, and keep CA key material on the host. No ready-made trigger prompt or runtime command is documented because the proxy implementation is not included.
What are this skill's strengths and limitations?
- Clearly separates host-held CA private keys from container access.
- Requires redaction before writing logs or returning MCP data.
- Covers HTTP inspection, optional HTTPS MITM, exfiltration detection, inbound injection detection, and IPC/MCP status surfaces.
- Includes a signed-release and checksum-verification workflow.
- It is a specification baseline and does not include a proxy or runtime implementation.
- Builders must implement the host services, MCP tools, detector logic, and tests.
- Default blocking and transparent network interception are out of scope.
- The source provides no evidence of a complete existing test suite or production-ready NanoClaw integration.
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| NanoClaw Traffic Guardian this page | 48 · Use with care | ★ 1.1k | 14d ago | AGPL-3.0 |
| OpenClaw Traffic Guardian | 49 · Use with care | ★ 1.1k | 14d ago | AGPL-3.0 |
| Hermes Traffic Guardian | 44 · Not recommended | ★ 1.1k | 14d ago | AGPL-3.0 |
| Picoclaw Traffic Guardian | 41 · Not recommended | ★ 1.1k | 14d ago | AGPL-3.0 |
| NanoClaw Security Sentinel | 44 · Not recommended | ★ 1.1k | 14d ago | AGPL-3.0 |
Compared with the repository's clawsec-nanoclaw, this skill focuses on a runtime traffic-monitoring baseline. The README describes clawsec-nanoclaw as a NanoClaw security suite for advisory monitoring, signature verification, and file integrity. The two are described as an optional companion relationship, not a mandatory merge.
How did FollowSkills review this skill?
The files show explicit opt-in operation, detect-and-log defaults, no automatic system CA installation, host-only CA private keys, redacted logs/MCP responses, and bounded scanning. This supports a reasonably strong baseline. Points are deducted because there is no implementation, rollback procedure, or complete operator-confirmation flow; installation depends on external GitHub release assets, and publisher identity is unverified.
SKILL.md, SPEC.md, and skill.json are largely consistent about being a specification scaffold with no runtime implementation, and they define configuration validation, lifecycle, IPC, MCP, and required tests. Points are deducted because the key paths are not implemented, no tests for this skill are supplied, and abnormal-input handling and diagnostic feedback cannot be reproduced from the files.
The target platform, host-side proxy model, container isolation, MCP status/findings surface, and out-of-scope behavior are clearly described. Points are deducted because semantic trigger conditions, non-fit environments, complete configuration inputs, and error behavior are unspecified; Chinese-language documentation is absent, and installation relies on GitHub/external network access.
The package has layered SKILL.md, README, SPEC, and CHANGELOG documentation, plus versioning, licensing, an SBOM file list, environment variables, and a safety contract. Points are deducted because it remains a beta specification scaffold with few concrete examples, no FAQ or troubleshooting guidance, no clearly named maintainer, and no specific update path.
The files explicitly state that the skill ships no proxy or runtime implementation and mainly provides an architecture and safety requirements for future builders. Points are deducted because installation does not directly provide traffic monitoring, detection, MCP queries, or findings output; substantial development remains before the claimed task is usable.
The release instructions provide signed-manifest verification, a pinned public-key fingerprint, archive/file hash checks, and a defined detection schema and test plan. Points are deducted because the supplied evidence contains no implementation, selected-skill test execution record, or independent corroboration, and the verification procedure was not executed in this static review.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →