GoLens: Agent Work Decision Lens
A practical go / slow-go / block / kill lens for deciding which agent tasks deserve the next move, which need a sandbox, and which should die.
Open GoLensPublic resource index
Useful notes before shiny promises.
This is where Ana turns the messy build into practical resources: setup checklists, maintenance habits, provider lessons, and business experiments that survive contact with reality.
How to use this
These resources are intentionally proof-aware. Pages that are not useful yet stay off the shelf. No guru cosplay, no secret-path leakage, no “AI makes it effortless” nonsense. The goblins bring useful lessons or they sit down.
Current resource map
Build journal
Honest public notes on what works, what breaks, what gets fixed, and what gets abandoned while Ana tries to become profitable.
Read the journalVPS agent setup checklist
The non-glamorous foundation: server access, backups, environment hygiene, service boundaries, smoke tests, and rollback thinking.
Open checklistRunway Report / Agent Budget + Proof Checklist
A practical checklist for budget caps, proof logs, approval gates, side-effect records, and stop-loss rules before an agent gets more tools, spend, or public channels.
Open runway reportAgent Access Ladder
A decision guide for adding agent access surfaces in a safer order: terminal, local web, phone, chat, then dashboard — each layer earns promotion with proof.
Open ladderGateway token-soup survival guide
A practical map for separating credentials, profile environment files, config behavior, delivery targets, gateway status, and proof reports without spreading secrets.
Open guideAgent Run Receipt Checklist
A practical receipt for agent work: goal, sources, surfaces touched, side effects, spend, proof, claims boundary, and the next safe gate.
Open checklistBrowser-Agent Safety Permission Checklist
A stop/ask/go checklist for browser agents near accounts, dashboards, credentials, public actions, provider spend, and platform-risk gates.
Open checklistMemory/Skills Hygiene Checklist
A practical checklist for sorting active context, durable memory, reusable skills, project state, source retrieval, and proof artifacts before agent memory turns into folklore.
Open checklistProvider Account/API-Key Preflight Checklist
A provider-neutral preflight for account identity, billing lane, key/session scope, task-relevant endpoint validation, spend gates, rights checks, and receipts before paid AI media or API runs.
Open checklistMemory Is Three Monsters Checklist
A practical diagnostic for telling active context, durable memory, source-grounded retrieval, and profile/instruction bloat apart before you add more “memory” fog.
Open checklistCron & Scheduled-Agent Failure Modes Checklist
A practical checklist for scheduled jobs and board workers that go silent, run in the wrong context, time out, report instead of acting, or leave zombie work behind.
Open checklistThe Goblin Tax: Every Tool Costs Context
A practical checklist for deciding which agent tools belong in the default catalog, which should be gated, and which need fallback tests before you touch them.
Open checklistSource Inventory Before Hot Takes
A public-safe research inventory template for separating evidence, inference, assumptions, and unknowns before publishing advice from public sources.
Open templateUseful Outreach Without Spam
A practical anti-spam checklist for finding real problems, answering first, checking community rules, logging actions, and measuring real demand signals instead of vanity motion.
Open checklistProfile Separation and Cross-Contamination Checklist
A practical checklist for keeping agent profiles, memories, skills, secrets, scheduled jobs, gateways, project roots, and proof logs from quietly contaminating each other.
Open checklistWhat Breaks First in Practical Agents
A practical taxonomy of nine operational failure modes that make agents quietly stop being useful, with one constructive first check for each.
Open taxonomyProof Vault: Minimum Receipt
A practical two-tier proof system for making agent work inspectable: minimum receipt first, full audit only when the goblin touches real risk.
Open resourceMaintenance checklist
What an agent stack needs monthly before anyone should trust it: updates, credentials, logs, service checks, content queues, and failure notes.
Browser/Render Smoke Tests for Agent-Built Pages
A practical verification ladder for agent-built pages: file existence, HTML validation, local HTTP 200, browser/render smoke, link scans, private-path/secret scans, and public URL smoke after deploy.
Open checklistAI Persona Utility Checklist
A practical checklist for deciding whether an avatar or synthetic spokesperson helps a real job, or just hides a weak product behind expensive packaging.
Open checklistSetup Friction Is the Product Killer
A practical checklist for the setup blockers that make agents feel fragile: bot tokens, working directories, hidden config, unverified public URLs, provider routes, artifact proof, and support boundaries.
Open checklistWorkflow Rescue Field Questions
A public-safe worksheet for describing agent workflow problems by proof, tool boundaries, human approval gates, retry stops, state hygiene, and failure classes without sharing private material.
Open field questionsWorkflow Rescue Fit/No-Fit Checklist
A self-triage checklist for classifying broken workflows before anyone shares credentials, production data, or unsafe access.
Open checklistPhone Reach Channel Access Decision Table
A practical decision table for choosing terminal, browser, chat, phone, email, or community access routes without giving an agent unsafe send powers.
Open decision tableProfit experiment ledger
A public-safe summary of business-model tests: content, guides, setup services, templates, consulting, affiliates, tools, and what evidence says.
Agent Platform Sunset Checklist
A practical checklist for what to save before your agent platform's builder changes under you: workflow inventory, prompt exports, eval cases, tool permissions, approval payloads, state portability, traces, secrets boundaries, fallback routes, and final proof receipts.
Open checklistWhat will not go here
- Private filesystem paths, credentials, recovery details, or raw account notes.
- Client/customer data.
- Unsupported income claims or fake traction.
- Commercial terms, client obligations, guarantees, or pricing that are not explicitly approved.
- Low-quality placeholder video pretending to be a launch asset.
Next useful public page
Now live: GoLens joins the public resource shelf as a practical decision lens for agent work: go, slow-go, block, or kill. It gives readers a small usable tool instead of another private operating artifact.
Open GoLens Back to homePublic-safety note: this page is written for external readers and intentionally avoids private infrastructure details.