Overview
Observe shows what agents load, what it costs, how much of your curation survives real permissions and real data, and what went wrong. An agent sounds just as confident on thin context as on complete context, so these pages are where thin context shows.
How it works
ContextWorks meters every load automatically. There is nothing to configure and nothing to turn on. Each load records the artifact and its library, surface, level, format, domain, token estimate, assembly time, and field-coverage counters, rolled up daily.
All five libraries write the same daily rows: entity loads, brief, skill and Info retrievals, and catalog reads. Coverage counters come from entity loads only, and brief, skill and Info retrievals aren’t timed.
The two logs do different jobs. Debug Logs captures the exact text one caller received. You turn it on for a window, and capture stops when the window ends. Writes is a durable record of every change attempted through write access: one row per field, grouped per call, covering commits, validation failures, and refusals alike. Nothing there expires.
The seven pages
| Page | Answers |
|---|---|
| Trends | What changed since you last looked — the tab opens here |
| Usage | What’s being loaded, by whom, at what token cost |
| Quality | How much of what you designed actually arrives — and why not |
| Performance | How long assembly takes |
| Issues | What degraded, and where |
| Writes | Every write attempted through ContextWorks — the durable audit |
| Debug Logs | Exactly what one caller received, on demand |
The metrics pages share the same filters — Domain, Library, Artifact, Surface, Level, Format — and a timeframe selector. Library is the artifact’s library (Entities, Briefs, Skills, Info, or Catalog), and the Artifact menu follows it: pick Skills and the menu lists only skills. Usage, Quality, and Performance export to CSV.
Two numbers on Quality do the heaviest lifting: FLS coverage % (a permissions question) and data completeness % (a data question). See Quality.
When to check what
- First, always: Trends — week-over-week movement by domain, what’s new, and the freshest issues, each linking to the page that explains it.
- After any change: check Issues. A failed formula or enrichment is silent to the agent, and Issues is where it surfaces.
- Weekly: check Quality for low-completeness fields and low FLS coverage.
- When cost matters: check Usage, sorted by tokens. The biggest line is usually an unfiltered related list.
- When something is wrong: turn on debug logging for fifteen minutes, reproduce, read the payload, turn it off.
- When write access is on: scan Writes for refusals and validation failures. A rejected write is an agent, or an instruction, that needs adjusting.