Picoclaw Traffic Guardian
An opt-in, detect-and-log baseline for monitoring Picoclaw gateway traffic and exporting security posture data.
The evidence shows opt-in operation, detect-and-log defaults, no automatic system CA installation, no global proxy changes, bounded scanning, and redacted logs; signed release-manifest and checksum verification are also specified. Points are deducted because user confirmation, runtime isolation, rollback, dependency pinning, and failure recovery are incomplete, while installation verification depends on external GitHub downloads.
The SKILL.md, SPEC, and skill.json are broadly consistent that this is a specification scaffold without a runtime implementation, and they define architecture, schemas, and minimum behavior. Points are deducted because key paths lack implementation, tests, reproducible execution, and diagnosable failure handling; static review cannot establish that installation, proxying, detection, redaction, or profile export works.
The target platform, scenarios, triggers, scope, and out-of-scope behavior are clearly stated, with HTTP, optional HTTPS, and profile integration described. Points are deducted for limited evidence about input/output boundaries, actual Picoclaw compatibility, and false-trigger boundaries; Chinese-language support is absent, and core installation verification relies on GitHub and other overseas services that may be unreachable from mainland China.
The skill provides a version, changelog, README, SPEC, skill.json, SBOM, license metadata, and installation/verification examples, with a reasonably clear structure. Points are deducted for missing implementation-level dependency notes, FAQ, troubleshooting, named maintenance responsibility, explicit update path, and fuller long-term version governance.
The candidate provides a fairly complete design baseline for traffic monitoring, detection categories, safety constraints, and implementation requirements. Points are deducted because it explicitly ships no proxy or runtime implementation, so it cannot currently perform monitoring, detection, logging, or profile export without substantial builder work; static evidence does not verify usable outcomes.
Signed release-manifest instructions, a pinned public-key fingerprint, checksum checks, and defined JSON schemas provide limited auditability. Points are deducted because the selected skill has no committed tests covering its key paths or third-party execution evidence; repository-level CI mainly covers other skills and shared tooling, so conclusions remain only partially reproducible.
- This is a specification scaffold, not a directly runnable traffic monitor; installation does not provide the proxy, detectors, logging, or profile-export implementation.
- Do not treat release-signature verification as runtime security verification; confirm dependencies, distribution channels, key rotation, permission scope, and rollback procedures first.
- HTTPS MITM, CA trust configuration, and sensitive-traffic redaction still require implementation and testing; validate them in an isolated environment before deployment.
- The repository provides no skill-specific evidence for key-path tests and no Chinese documentation or mainland-China alternative download path.
What does this skill do, and when should you use it?
Picoclaw Traffic Guardian is a runtime traffic-monitoring specification skill in the ClawSec suite for Picoclaw. It defines lightweight gateway proxy inspection, outbound exfiltration detection, inbound injection detection, redacted local threat logs, and posture export. Version 0.0.1-beta5 is only a baseline specification and does not ship a proxy or runtime implementation. It is best suited to teams that need a defined security contract before building a platform-specific monitor.
It documents a standalone release-verification workflow that downloads and verifies a signed manifest, archive, SKILL.md, and skill.json. A future implementation is required to validate configuration, run in the foreground or an explicitly selected background mode, scope proxy environment variables to the target Picoclaw gateway process, and inspect HTTP request and response text within a bounded byte limit. It must emit JSONL findings with redacted snippets and export a small profile fragment for picoclaw-security-guardian. The current package contains no proxy or runtime implementation.
- A Picoclaw operator wants to plan opt-in monitoring for one gateway process without changing the system trust store or global proxy settings.
- A security engineer needs an implementation contract for detecting outbound exfiltration and inbound prompt injection before building the proxy.
- An operations team needs local JSONL findings with secrets redacted by default.
- A supply-chain maintainer needs to verify the release key, signed manifest, archive hash, and individual artifact hashes before installation.
What are this skill's strengths and limitations?
- Opt-in with detect-and-log as the default behavior.
- No automatic system CA installation or global proxy changes.
- Requires redaction before logs, summaries, and profile outputs.
- Provides signed-manifest, public-key-fingerprint, and SHA-256 verification procedures.
- Version 0.0.1-beta5 is a specification baseline with no proxy or runtime implementation.
- The first implementation is explicitly non-blocking and excludes transparent interception.
- It must not send traffic to external services or collect full request and response bodies.
- The supplied material does not include SPEC.md, an implementation, configuration schema, or test results.
How do you install this skill?
Install with the Vercel Skills CLI: npx skills add prompt-security/clawsec --skill picoclaw-traffic-guardian -a openclaw -y. For standalone installation, run the verification procedure in SKILL.md against checksums.json, checksums.sig, signing-public.pem, the release archive, SKILL.md, and skill.json; install or extract the archive only after verification succeeds.
How do you use this skill?
This package is a baseline specification, not a runnable proxy. Builders should read SPEC.md, place detector, redaction, profile-export, and report-format logic under lib/, helper scripts under scripts/, and tests under test/. The supplied source does not document a concrete start command, configuration schema, or executable test suite.
How does this skill compare with similar options?
Its scope is distinct from picoclaw-security-guardian and picoclaw-self-pen-testing: this skill defines the runtime traffic-monitoring boundary, while those skills should profile, drift-check, or review its status rather than own the proxy runtime.