Skip to main content
Lab notebook

Lab notebook

Every project my agent pipeline works on is written up as a climb: the problem it started from, a measurable summit, the rules it may not break, and every merge on the way, each tied to the roadmap line it served or marked as a detour. This is the notebook, one page per project.

Committed snapshot· generated 2026-11-07 15:31 UTCJSON →Factory run record →

These climbs are the committed snapshot, not live data. Served because: operator not configured.

4 climbs over the last 30 days, 1008 merged steps: 42% joined to the route by an exact key, 1% by inference, 58% off the written route: 34% owner merges, 5% exhaust and tech-debt rows, 3% canaries, 16% plan-less rows, 0% other. 138 weather episodes and 0 false summits (reverted merges) in the window.

Tier 1 · invest

Active investment: these projects get new work.

loom-core

services/loom-core

Tier 1 · invest

Agent swarms produce activity, not delivery: work that nobody scoped, verified or merged. loom-core's Mills operator turns roadmap intent into scoped items, proves every change in a merge train, and keeps cost, failures and operator controls visible, so the factory can be trusted to run unattended on the lab's top projects.

Auto-merge rate (30d)
86.4%
target ≥ 95%
+43.7 pts since Oct 8, toward target
Enqueue-to-merge p50 (press entries)
0.6 hours
target ≤ 1 hours
+0.4 hours since Oct 9, away from target
Cost per merged change
2.2 usd
target ≤ 3 usd
−3.1 usd since Oct 8, toward target
Roadmap gates met
14.3%
target ≥ 100%
2 of 14 checked, 0 verified (2 claimed, unverified)
±0 pts since Nov 7

Altitude toward target, all summit metrics; the dotted line is the target.

How the walk joined the route · 726 steps

exact key
48%
inferred
0%
off route
51%

51% off the written route: 18% owner merges, 6% exhaust and tech-debt rows, 4% canaries, 23% plan-less rows, 0% other.

Conditions in the window

weather
112 episodes, 557.4 h, holding
false summits
0

Latest leg

Four waypoints served on 2026-11-06 as auto-merge rate slipped away from target.

Open the loom-core climb →

fi-fhir

libs/fi-fhir

Tier 1 · invest

Healthcare integration engines are closed, expensive and hard to observe. fi-fhir is an open, Kubernetes-native engine with profile-driven parsing, canonical events, durable ingestion and delivery, and a Mapping Studio. It is pre-1.0: its capabilities and tests do not yet certify every supported deployment profile or vendor feed, and the climb is toward a 1.0 whose supported matrix they do.

Roadmap gates met (Now list)
59.4%
target ≥ 100%
19 of 32 checked, 10 verified (9 claimed, unverified)
±0 pts since Nov 7
Main pipeline success ratio (30d)
90.9%
target ≥ 95%
−9.1 pts since Oct 9, away from target
Slice-to-merge p50 on the Mills lane
n/a
target ≤ 2 h
n/a over the window

Altitude toward target, all summit metrics; the dotted line is the target.

How the walk joined the route · 36 steps

exact key
83%
inferred
0%
off route
17%

17% off the written route: 17% owner merges.

Conditions in the window

weather
4 episodes, 14.8 h
false summits
0

Latest leg

No changes merged on 2026-11-06

Open the fi-fhir climb →

edilint

libs/edilint

Tier 1 · invest

Invisible EDI defects (homoglyphs, short counts, mixed line endings, trailer miscounts) pass through healthcare clearinghouses and fail downstream. edilint is a public Go linter and fixer for X12, HL7v2 and EDIFACT that makes those defects visible in CI, climbing from v0.6.0 toward the v1.0 "legitimately useful" exit criteria.

Roadmap gates met (Now list)
58.3%
target ≥ 100%
7 of 12 checked, 4 verified (3 claimed, unverified)
±0 pts since Nov 7
Merges per week
13 merges
target ≥ 3 merges
+8 merges since Oct 9, toward target
GitLab pipeline success ratio
91.7%
target ≥ 95%
−8.3 pts since Oct 9, away from target

Altitude toward target, all summit metrics; the dotted line is the target.

How the walk joined the route · 15 steps

exact key
53%
inferred
0%
off route
47%

47% off the written route: 47% owner merges.

Conditions in the window

weather
1 episode, 12.3 h, holding
false summits
0

Latest leg

No changes merged on 2026-11-06

Open the edilint climb →

flexinfer

services/flexinfer

Tier 1 · invest

Self-hosted inference on a mixed homelab fleet (two gfx1100 cards, a Radeon VII and a GTX 980 Ti) breaks in ways nobody measures: lanes that hog a card, crash on load or time out behind a cold start. FlexInfer is the Kubernetes operator and OpenAI-compatible proxy that serves the lab's LLM, embedding, rerank, speech, image, audio and video lanes with scale-to-zero, breakers, reroute and per-GPU certification, climbing toward a fleet where every request the proxy routes is answered and every backend change is certified before it serves.

Proxy request success ratio (rolling 1d)
99.5%
target ≥ 99.5%
−0.4 pts since Oct 30, away from target
Roadmap gates met (Now list)
50%
target ≥ 100%
2 of 4 checked, 1 verified (1 claimed, unverified)
±0 pts since Nov 7
Master pipeline success ratio (7d)
67.1%
target ≥ 90%
+67.1 pts since Oct 9, toward target
Merges per week
109 merges
target ≥ 40 merges
+101 merges since Oct 9, toward target

Altitude toward target, all summit metrics; the dotted line is the target.

How the walk joined the route · 231 steps

exact key
13%
inferred
3%
off route
84%

84% off the written route: 84% owner merges.

Conditions in the window

weather
21 episodes, 433.4 h, holding
false summits
0

Latest leg

17 changes merged on 2026-11-06

Open the flexinfer climb →

Sanitized from the Mills chronicle read model: run, backlog and roadmap ids are hashed; titles, headlines and narrated sentences are scrubbed of paths, URLs, credentials, addresses and unpublished repository names; only climbs on published repositories are listed. Merge request numbers and short commit hashes are kept because they are public; so is the name of the model that narrated each leg, while its backend, prompts and traces are not.

Lab Notebook | Cody Blevins