All plugins

elianjuan-g

GitHub profile2 plugins in the catalog6

Plugins by elianjuan-g

2 plugins by elianjuan-g in the catalog, most-starred first.

elianjuan-ghermes-guardDefensive prompt-injection guard for Hermes Agent: a pre_llm_call hook scans every inbound message locally (deterministic patterns, no model calls) for Unicode smuggling, hidden instructions, special tokens, and Base64-decoded payloads; a pre_tool_call hook blocks RCE/subshell patterns, sensitive credential paths, and call-ID races. High-confidence attacks are blocked; suspicious messages are quarantined for manual review via the guard_review tool (release is manual re-send by design). Includes a Predator pattern store (human-corrected signatures), a SHA-256 hash-chained ledger, secret redaction, a live Posture Score, a 28-vector simulation arena (21 message + 7 tool-arg), and a standalone Obsidian & Gold HUD. Python stdlib + SQLite only. Disclosure — no network calls, no credentials read or stored, no data leaves the machine, no core overrides, no self-updater. Phase 3 verified live on Hermes 0.21.5 (validate 13/13, doctor OK, plugins list enabled, live hook-fire test: attack blocked on both hooks, benign passed); v1.1.2 re-verified the same way (validate 13/13, live hook-fire test 19/19 through the real plugin dispatch, 168/168 tests).Communitytools6elianjuan-ghermes-manas-aegis[ALPHA] Experimental Hermes guardrail (prototype, not standalone security): pattern-based pre-tool checks that can block or rewrite some risky tool arguments, with a canary honeypot, local telemetry and a loopback HUD, plus STIX 2.1 export. Opt-in, test environments only. Disclosure — alpha guardrail: its pre_tool_call / pre_llm_call / transform hooks can block tool calls or rewrite their arguments (honeypot redirect to a decoy file, safe-flag git rewrite) and scrub tool results; serves a token-protected loopback-only HUD on 127.0.0.1 (default port 18777, off with auto_launch_hud_server: false); fetches public CISA KEV and NVD CVE feeds over HTTPS (no key) for the threat-intel panel, refreshed every 6 hours while the HUD runs; the full scan reads files under the active Hermes home (skipping .env, auth.json, config.yaml, webhook_subscriptions.json, mcp-tokens/, *.bak backups and other profiles); writes redacted telemetry and a STIX 2.1 feed under the plugin data folder; no credentials read, no telemetry sent out.Communitytools