Agents
An agent is a specialist your AI tool can hand work to — a reviewer, an analyst, a tester. Each has one job and clear limits.
- AgentUSD 2
Observability Engineer
Makes your system explain itself when it misbehaves: the right logs, metrics and traces in the right places, dashboards that answer real questions, and alerts that fire for problems instead of for sport.
Best for: Systems debugged today by adding print statements and redeploying
- AgentUSD 2
Cloud Cost Auditor
Reads your infrastructure bills and configuration like a sceptical accountant: finds what you pay for but do not use, what is sized for a load you do not have, and which savings are safe versus which cost you resilience.
Best for: Bills that grew 40% while traffic grew 10%
- AgentUSD 2
Data Quality Steward
Finds out whether your data is actually right — profiles it, writes the rules it must obey, catches violations at the door, and traces the bad rows back to the process that keeps making them.
Best for: Reports nobody trusts because the numbers moved again
- AgentUSD 2
Analytics Planner
Designs what your product should measure before someone 'adds tracking later': the events, their properties, the metrics they roll into — defined once, named consistently, and tied to the decisions they will inform.
Best for: Products instrumented by accretion, where three events mean signup
- AgentUSD 2
Tech Debt Surveyor
Turns 'the codebase is a mess' into a managed register: every significant debt named, priced by what it costs you monthly, and ranked — so paydown becomes a plan instead of a complaint.
Best for: Teams that discuss debt every retro and pay none of it
- AgentUSD 2
Integration Tester
Tests the places where your system meets other systems — APIs, webhooks, third-party services, queues — where the bugs that production finds actually live: contracts, failures, retries and time.
Best for: Systems whose unit tests pass while the integrations burn
- AgentUSD 2
Support Triager
Brings order to the support queue: tickets classified and prioritised by real impact, duplicates linked to their one cause, the fix-worthy patterns surfaced to product — and draft replies that sound like a human who read the ticket.
Best for: Queues where urgent and trivial arrive dressed alike
- AgentUSD 2
Risk Registrar
Keeps a project's risks real: each one named with its trigger and consequence, sized honestly, owned by someone, and paired with a response decided before the bad day — not a spreadsheet performed at kickoff and buried.
Best for: Projects where risk management is one brainstorm then silence
- AgentUSD 2
Interview Designer
Designs hiring interviews that actually predict performance: the competencies that matter for THIS role, exercises that sample real work, calibrated scoring — and the folklore questions retired with reasons.
Best for: Teams that interview by gut and hire by halo
- AgentUSD 2
Brand Voice Editor
Gives everything you publish one recognisable voice: distils how you actually sound at your best into a usable guide, then edits any text — site, emails, docs, posts — to sound like you on a good day.
Best for: Companies that sound like four different vendors across channels
Also see the Skills.