docs(features): split the response inventory by topic #512
No reviewers
Labels
No labels
move-to-repo
coilyco-bridge-deploy
move-to-repo
coilyco-flight-deck-agent-compose
move-to-repo
coilyco-gaming-eco-app
move-to-repo
coilysiren-inbox
move-to-repo
unknown
🔒⚠️📦⚠️🔒 SANDBOXED 🔒⚠️📦⚠️🔒
autonomy
async-consult
autonomy
epic
autonomy
headless
autonomy
live-collab
c#
priority
P0
priority
P1
priority
P2
priority
P3
priority
P4
role/ai
role/creator
role/design
role/director
role/engineer
role/exec
role/human
role/ops
role/qa
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set
Reference
coilyco-gaming/sirens-echo!512
Loading…
Reference in a new issue
No description provided.
Delete branch "eng/split-the-response-inventory"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
The inventory was 42 characters from its 4000-character cap, so the next shipped capability could not be recorded and the cheapest response for each agent was to skip the inventory. A few rounds of that and it stops being an inventory while still reading like one.
Two cuts, both by subject
features-observability.mdtakes what the service reports about itself: OTel ingress and joined traces, trace-correlated metadata logs, the metric families, liveness and readiness, the gateway heartbeat, and the caller/service fault split.features-admission.mdtakes who may reach it: invocation gates, summoning by edit, the access policy, and the admission tiers.That leaves the response-service file as one subject — how a turn is run, checked and answered — at 3312 characters. Telemetry is where six entries arrived in the last day, so it is the growth that most needed somewhere to go.
One cut would have met the acceptance and left 293 characters, which at the current rate is about a day. Two leaves each file with real room.
A stale number, replaced with the property
The inventory said "23 bounded SigNoZ exception groups". The catalog now holds 29 distinct outcomes across 32 codes — six of which I added today, so I am the reason it drifted.
I did not write 29 in its place. An untested count in prose drifts again on the next code, and this one already had. It now names the property instead: a build-time closed catalog tagged by stage, outcome, and fault. That is what the number was standing in for, and unlike the number it stays true.
Restored
The reply-assembly entry that was reverted when the cap refused it, in its author's wording, now that #413 has landed at
c71f1ff.closes #471