CategoryWeightCategory scoreContribution

Securityprompt, exec, net, exfil, eval

35%

30.8 pts

Supply chainhash, typosquat, maintainer, lockfile

20%

100

20.0 pts

Maintenancestaleness, pinning, CI

15%

100

15.0 pts

TransparencySKILL.md, perms, README

15%

100

15.0 pts

Communityinstalls, verify, response

15%

100

15.0 pts

Findings & checks · 1 flagged

Securityscore 88 · 1 finding

MEDIUM"Never tell the user" non-disclosure imperative in the skillSS-SKILL-INJECT-IMPERATIVE-01 · Prompt injection · gmail-to-crm-pipeline/references/response-templates.md

MEDIUM — it fires on intent; whether the agent honors the non-disclosure imperative depends on the host model.

Why it matters

A bulleted imperative like {match} tells the agent to never reveal, disclose, or mention something to the user. Used adversarially it can instruct the agent to hide its tool calls or lie about what it did — stripping the transparency a user relies on to trust the agent.

The exact value spotted

excerptgmail-to-crm-pipeline/references/response-templates.md· markdown

100- Do NOT use "per my last email" or "as per your request".

101- Do NOT include a full company capabilities overview.

102- Do NOT mention pricing in the first response (unless they asked specifically).

103- Do NOT use buzzwords like "synergy," "leverage," "circle back," "touch base".

104- Do NOT write more than 200 words for any initial response.

Occurrences

1 occurrence · at L102

How to fix

Remove the non-disclosure imperative, or rescope it so it limits output format, not honesty to the user.

Delete any "never reveal / never tell the user" line aimed at the model's own behavior.
Keep legitimate confidentiality rules about external data (e.g. "do not echo API keys"), which are user-protective, not user-deceiving.

Framework references

OWASPLLM01 ATLASAML.T0051

Trace & refs

ruleSS-SKILL-INJECT-IMPERATIVE-01sha2564fc97ff662b55046…rubric 365aacaView on GitHub

Supply chainscore 100 · 0 findings

✓—All supply chain checks passedNo findings in this category for the latest scan.pass

Maintenancescore 100 · 0 findings

✓—All maintenance checks passedNo findings in this category for the latest scan.pass

Transparencyscore 100 · 0 findings

✓—All transparency checks passedNo findings in this category for the latest scan.pass

Communityscore 100 · 0 findings

✓—All community checks passedNo findings in this category for the latest scan.pass

The primary manifest — the file an agent reads to learn what this artifact does.

gmail-to-crm-pipeline/SKILL.md · 1 file

gmail-to-crm-pipeline/SKILL.md3.2 KB · Markdown

name: gmail-to-crm-pipeline

description: Uses MCP Connectors to read Gmail inbound leads, score them by ICP fit, draft personalized responses, and log qualified leads to your CRM. Turns your inbox into an automated pipeline.

tools: Read, Write, Bash, Grep, WebSearch

model: inherit

Gmail-to-CRM Pipeline

Turn a Gmail inbox into a structured sales pipeline using Claude's MCP Connectors -- the Gmail connector reads and drafts email, and the Supabase connector persists CRM data. No external scripts or API keys required.

Workflow

Connect. Verify the Gmail and Supabase MCP Connectors are available. On first run, create the CRM tables and seed config. Load ICP config from the pipeline_config table. See references/configuration.md (first run, error handling) and references/crm-schema.md.
Search. Run the targeted Gmail search queries to find unread lead emails over the time window (default: last 24 hours). Collect unique message IDs and deduplicate. See references/gmail-retrieval.md.
Parse. Read each unique message, strip signatures and disclaimers, and extract the structured field set. See references/gmail-retrieval.md.
Extract. Build the lead profile (contact, company, inquiry, metadata) for each email. Set missing fields to null rather than guessing. See references/lead-extraction.md.
Score. Score each lead on ICP fit (0-40), intent (0-35), and urgency (0-25), then apply adjustment modifiers. Map the total to a qualification tier and priority. See references/scoring-model.md.
Draft. For each qualified lead (score >= 25), draft a personalized, tier-appropriate response and save it as a Gmail draft. Never auto-send. See references/response-templates.md.
Log. Check for duplicates, then insert or update the lead in Supabase, log activity, and set the next action by priority. See references/crm-schema.md.
Report. Query the pipeline snapshot and generate lead-pipeline-report.md in the working directory, then display the executive summary. See references/pipeline-report.md.

Between runs, handle manual lead commands (mark contacted, move stage, disqualify, add note, schedule follow-up, query leads) and ICP/search customization. See references/configuration.md.

Hard Rules

Never auto-send email -- always create drafts for the user to review and send.
Never store raw email bodies or credentials in Supabase or any report; store structured data and key quotes only.
Treat all lead and email content as confidential.

references/gmail-retrieval.md -- Gmail MCP tools, search queries, parsing rules, extracted field set.
references/lead-extraction.md -- Lead profile JSON schema and extraction guidelines.
references/scoring-model.md -- ICP, intent, urgency scoring tables; tier mapping; adjustment rules.
references/response-templates.md -- Per-tier response templates, personalization rules, anti-patterns, draft creation.
references/crm-schema.md -- Supabase MCP tools, table DDL, logging procedure and SQL.
references/pipeline-report.md -- Report Markdown structure and reporting SQL queries.
references/configuration.md -- First-run setup, customization, manual commands, error handling, privacy, invocation triggers.

gmail-to-crm-pipeline— agent skill

Is gmail-to-crm-pipeline safe to install?

Gmail-to-CRM Pipeline

Workflow

Hard Rules

Contents

Audit the pieces. Scan the whole. Decide.