hermes-attestation-guardian

v0.1.0

Hermes-only runtime security attestation and drift detection skill for operator-managed Hermes infrastructure.

⭐ 0· 87·0 current·0 all-time

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for davida-ps/hermes-attestation-guardian.

Previewing Install & Setup.

Prompt PreviewInstall & Setup

Install the skill "hermes-attestation-guardian" (davida-ps/hermes-attestation-guardian) from ClawHub.
Skill page: https://clawhub.ai/davida-ps/hermes-attestation-guardian
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Bare skill slug

openclaw skills install hermes-attestation-guardian

ClawHub CLI

Package manager switcher

npx clawhub@latest install hermes-attestation-guardian

Security Scan

Capability signals

CryptoRequires walletRequires sensitive credentials

These labels describe what authority the skill may exercise. They are separate from suspicious or malicious moderation verdicts.

VirusTotal

Benign

View report →

OpenClaw

Suspicious

medium confidence

ℹ

Purpose & Capability

The code and SKILL.md implement Hermes-focused attestation, signature verification, advisory feed checks, baseline diffing, and optional cron scheduling — all coherent with the declared purpose. However the package metadata (registry requirements) omitted the 'node' runtime that SKILL.md and the included .mjs scripts require; that's an inconsistency in declared requirements.

✓

Instruction Scope

Runtime instructions are narrowly scoped to Hermes assets: they read/write files under HERMES_HOME, verify signed artifacts, refresh/verify a remote advisory feed, and manage optional scheduled jobs. The instructions do not ask for unrelated credential stores or arbitrary system data. They do include an explicit pre-install release verification step (downloads from GitHub releases + OpenSSL checks) which is appropriate for the purpose.

✓

Install Mechanism

There is no automatic install spec (no remote downloads executed automatically). The skill is instruction-only regarding install, and the code files are bundled in the skill. This minimizes surprise remote install behavior. The SKILL.md recommends manual verification of release artifacts before install (good).

Credentials

Registry metadata lists no required environment variables, but the code and SKILL.md rely on many environment variables (HERMES_HOME, HERMES_ADVISORY_ALLOW_UNSIGNED_FEED, HERMES_ADVISORY_FEED_STATE_PATH, gateway toggles like HERMES_GATEWAY_TELEGRAM_ENABLED, and others). That mismatch means the skill may behave differently or require operator configuration not documented in the registry entry. The skill also reads and can write files under the user's HERMES_HOME and advisory state; these are reasonable for the stated purpose but represent sensitive local state that the operator should review.

ℹ

Persistence & Privilege

The skill does not request 'always: true' and allows model invocation (default). It exposes scripts to add managed cron blocks (apply mode) which will mutate the user's crontab/scheduler if used; this is expected for recurring checks but is a privilege that alters system state. The code takes precautions (print-only default, explicit --apply) and confines written paths to HERMES_HOME and rejects symlinks, which is good practice.

What to consider before installing

This package appears to implement the Hermes attestation and advisory-check functionality it claims, but take these precautions before installing or running it: - Ensure you have Node installed; SKILL.md and the scripts require node even though the registry metadata omitted it. - Verify the release artifacts exactly as the SKILL.md shows (checksums.json, checksums.sig, and the pinned signing public-key fingerprint). Do not skip the OpenSSL verification step. - Review and set HERMES_HOME explicitly (the tool confines writes to HERMES_HOME). Inspect existing files under that directory; the tool will read/write under $HERMES_HOME/security and can create state and cached feed files. - Audit the hard-coded feed URL and pinned feed public key (DEFAULT_REMOTE_FEED_URL and PINNED_FEED_PUBLIC_KEY_PEM) to ensure they match your operator trust policy and the release verification fingerprint. - Be cautious when using any --apply or --allow-unsigned flags: --apply will modify your crontab/scheduler (the scripts call schedule-bin operations) and --allow-unsigned is explicitly emergency-only and bypasses signature checks. - Because the registry metadata did not declare env vars the code uses, plan to run the scripts in a controlled sandbox (container or VM) first and run the provided tests (or inspect test scripts) to confirm behavior in your environment. If you want higher confidence, ask the maintainer for: corrected registry metadata listing 'node' and the env vars the skill reads, a signed release tarball whose checksums match the published manifest, and documentation of the default scheduleBin used for crontab operations.

✗

lib/cron.mjs:116