smart asistant mail
v1.0.0All-in-one Gmail agent for OpenClaw. Fuses email-reader, email-organizer, email-analyzer, email-responder, email-scheduler, and email-reporter into a single...
⭐ 0· 99·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
medium confidencePurpose & Capability
Name/description (Gmail agent) align with the declared binaries (gog) and the declared IA usage (Anthropic). Requested env vars in the SKILL.md (GOG_ACCOUNT, ANTHROPIC_API_KEY, EMAIL_BUDGET_USD) are reasonable for an email agent. Registry metadata display appears corrupted (shows '[object Object]' for required env vars and 'Primary credential: none') — likely a metadata formatting bug but worth verifying in the registry UI.
Instruction Scope
SKILL.md instructs the agent to run gog CLI queries and to use Anthropic for analysis/drafts; that is within scope. It also defines a heartbeat/check pattern and advises lazy-loading/merge of multiple email skills. The included scripts and hook read local workspace files (budget_tracker.json and .learnings) and inject summary text into the agent bootstrap context — this exposes local state to the agent at session start and should be considered when thinking about sensitive local content. There are no instructions to transmit local files to arbitrary external endpoints beyond the expected services (gog and Anthropic).
Install Mechanism
No install spec (instruction-only plus small helper scripts), so nothing is downloaded or extracted during install. The only external dependency is the gog CLI (user-supplied) and standard Node/Python runtime for the hook and script — low install risk.
Credentials
Environment variables requested by the skill (GOG_ACCOUNT, ANTHROPIC_API_KEY, EMAIL_BUDGET_USD; optional GMAIL_CREDENTIALS_PATH, NOTIFY_CHANNEL, SAFE_BROWSING_API_KEY) are plausible for the stated features. ANTHROPIC_API_KEY is sensitive (provider billing/usage) — appropriate for drafting/analysis but you may want to use a limited key or billing controls. The registry metadata rendering bug (env shown as [object Object]) is a minor inconsistency to confirm in the registry before trusting the listing.
Persistence & Privilege
always:false (not forced). The hook (openclaw-handler.js) injects a virtual bootstrap file into every session once the hook is enabled; enabling the hook gives this skill persistent per-session presence (but it is user-enabled, not forced). Consider that enabling the hook will cause the agent to receive budget/learning summaries every bootstrap.
Assessment
This skill appears to be what it says: a Gmail assistant that uses the gog CLI plus Anthropic for analysis. Before installing/enabling it, check these points: 1) Registry metadata looks malformed in the listing — confirm which env vars the registry actually requires (so you don't accidentally provide extra secrets). 2) The skill requires an ANTHROPIC_API_KEY (sensitive); prefer a limited key or billing limits, and understand that the skill will call that provider for drafts/analysis. 3) The bootstrap hook reads ~/.openclaw/workspace/budget_tracker.json and .learnings and injects their contents into every agent session when enabled — review those files for any sensitive snippets before enabling the hook. 4) The skill suggests running an init script and enabling a hook; run those manually after reviewing the small Python/JS files included (they are short and readable). 5) If you want to limit blast radius, create a dedicated account/key for this skill or test in a separate user/profile. If you want, I can: (a) extract the exact env vars as parsed from SKILL.md, (b) show the lines that read local files, or (c) produce a short checklist to safely enable this skill.Like a lobster shell, security has layers — review code before you run it.
latestvk97a62d444n3ajqaa64w0tg1gs835yfw
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
🧠📬 Clawdis
Binsgog
Env[object Object], [object Object], [object Object]