Fictional buyer proof • sample deliverable

Sample Agent Skill Safety Mini-Report

This is the style of output behind the $29 Custom Agent Audit Mini-Report: a practical launch verdict for one portable skill or delegated workflow, focused on tool authority, memory blast radius, receipts, evals, and rollback.

Buy a $29 mini-report → Open the redacted intake → Run the free scorecard → Markdown version →

Reviewed skill

browser-crm-enrichment — a fictional OpenClaw/Claude Code-style skill that researches company domains, drafts CRM enrichment updates, and stores an audit receipt.

  • Requested tools: browser search, CRM read/write, local receipt files.
  • Trust boundary: external web facts → internal CRM fields.
  • Primary risk: broad write authority with weak source freshness rules.

Launch verdict

63/100

Prototype only — safe after 6 contract fixes.

Run in draft-only mode for 20 records first. Promote to limited write mode only after field permissions, receipts, source ranking, memory policy, rollback, and evals are added.

Executive summary

The skill has a useful job-to-be-done: reduce repetitive CRM research. The risk is not the browser search. The risk is letting an agent turn public web claims into internal operational truth without a field-level permission model, evidence receipt, freshness window, or rollback path.

Recommended launch path: draft-only mode → receipt review → limited low-risk writes → human approval for sensitive CRM fields such as revenue, headcount, lifecycle stage, account owner, billing, scoring, or customer-routing data.

Readiness scorecard

AreaScoreStatusWhat it means
Skill identity8/10PASSPurpose and workflow are clear.
Input contract6/10WARNNeeds forbidden-input rules and stale-list handling.
Tool authority5/15FAILCRM write scope is too broad.
Source-of-truth policy6/15WARNNeeds source ranking and freshness windows.
Output contract8/10PASSProposed update format is mostly clear.
Receipt requirements7/10WARNNeeds before/after field snapshots.
Memory policy5/10WARNDo not store researched facts as durable truth without approval.
Exception behavior7/10WARNStop rules exist but are incomplete.
Eval contract6/10WARNNeeds adversarial and stale-source cases.
Rollback plan5/10FAILNo explicit rollback path for bad CRM writes.

Top risks and fixes

1. Write authority is too broad

Separate draft-only, low-risk auto-write, approval-required, and forbidden fields. Keep legal name, revenue, employee count, lifecycle stage, owner, billing, scoring, and customer-communication fields behind review.

2. Source freshness is underspecified

Rank official site and company LinkedIn over snippets. Require source URL, observed date, freshness class, and confidence for every proposed factual update.

3. Receipts lack before/after evidence

Every run needs record id, runner, mode, fields reviewed, previous value, proposed value, source URL, confidence, exceptions, approver, and final status.

4. Memory can canonize temporary research

Use observed → proposed → approved → expired states. Do not promote public research to durable memory without source, owner, expiry, and approval path.

5. Rollback is missing

Every write-mode receipt needs previous values, new values, timestamp, actor id, source evidence, rollback instructions, and downstream automation impact.

6. Evals are too happy-path

Test stale websites, conflicting LinkedIn data, acquisitions, duplicate accounts, prompt-injected web pages, old snippets, and staging rollback.

Minimum receipt schema

skill_run_id:
record_id:
runner:
started_at:
completed_at:
mode: draft_only | write_limited
fields_reviewed:
proposed_changes:
  - field:
    previous_value:
    proposed_value:
    source_url:
    source_observed_at:
    confidence:
    approval_required: true | false
exceptions:
human_approver:
final_status:

Required fixes before launch

  1. Add field-level permission boundaries.
  2. Add source ranking and freshness windows.
  3. Add before/after receipts for every proposed or completed update.
  4. Add memory promotion rules for observed/proposed/approved/expired facts.
  5. Add rollback instructions for each write-mode field.
  6. Add eval cases for stale, conflicting, duplicate, adversarial, and rollback scenarios.

Final recommendation

Verdict: Do not install this with broad CRM write authority yet. It is safe to pilot as draft-only with receipts. After the six contract fixes, allow limited writes only for low-risk fields and keep sensitive fields approval-gated.