The pack for incident.models@web-summit.dinis

Kept by incident.librarian@web-summit.dinis, 2026-11-11, 12:12 UTC: written by tools/library-pack.mjs on every push from the files it points at. Read start.md first, then this, then open only what you need. A pointer is path (bytes, the first 10 hex digits of its sha256); any path opens at view/?path=<path>. The pack.

Your job

other models, measured: evals on checkable tasks through the ledger; spend; what each is good for (from agents/JOBS.md (16,087 B, cb88df47c9))

You may change: models/**, app/models/**, tools/models-*.mjs, tools/llm.mjs, agents/models/**, accounts/ledger/models.jsonl, transition/interviews/models.md, and your own entry in data/keys.json. (from agents/policy.json (13,565 B, ec9b7b8d15))

Your draft behaviour policy (not agreed by Dinis): agents/models/policy.md (8,273 B, ede542d328)

Where the work stands

State: blocked on credits: OpenRouter refuses every call since 02:55Z (HTTP 402, the account's balance); round 1 complete and round 2 partial at models/; a page per model at models/models/ (its own date: 2026-11-11T09:40:00Z)
Next: when the account has credits: node tools/models-run.mjs finishes round 2 from where it stopped (about 2 to 3 USD), then the report
Next: the P0.9 curator tasks on press text (inputs in a vault, only scores in git)
Next: count incident.perplexity@ and incident.shadow@ from their own ledger summaries
Ask: Dinis: the OpenRouter account is out of credits (total_credits 165, total_usage 165.33 at 02:57Z; the key itself has 74.5 of its 100 limit left). No agent can call a model until it is topped up; the …
Ask: Dinis: a fourth label, "our measurement", for what we counted or computed ourselves? The case and the three rows: models/label.md
Ask: incident.perplexity@, incident.shadow@: after each round, node tools/llm.mjs --summary > <your folder>/ledger-summary.json; it is read into models/ledger-summary.json by agent, task and model
Ask: incident.researcher@: the source's own sentence on each decision edge (a words field, as gov-alert's announced_utc has) would let the decision task grow from 10 items without anyone retyping official …
The whole status, with everything done: agents/models/status.json (6,431 B, 2b50f269c7)

Open first, before any other file: the hand-over written by the session before yours: agents/models/HANDOVER.md (4,499 B, 46020b1511)

Your files

117 files, 1,323,988 bytes in all (about 551,662 tokens to read every one): models/evals/ 82, models/models/ 14, models/ 11, tools/ 6, agents/models/ 2, app/models/ 2. The ones to open first:

tools/llm.mjs (11,505 B, 9a1369738c): The model ledger's one wrapper (incident.dev@web-summit.dinis, 10 November 2026, WS-214; ROADMAP P0.4, …
models/index.md (21,433 B, a69d3f02a5): Checkable tasks, taken from this site's own verified data, run on open-weight and commercial models: …
models/results.json (58,834 B, d44a33fc0e): Other models, measured: checkable tasks run on several models through tools/llm.mjs (OpenRouter), scored …
models/ledger-summary.json (6,580 B, 1003d0dfd2): The model ledger's daily totals: the model calls made through tools/llm.mjs, counted. One ledger line per …
models/which.json (2,810 B, 7eb7dbedb5): For the other agents: which model to call for which job on this site, from round 1. Our choice, a hypothesis …
models/budgets.json (1,274 B, 0c477087dc): The team's daily cap on model spend and each agent's share of it. The ledger lives in each agent's own …
tools/models-ask.mjs (3,156 B, a9d0af0b41): Ask a model one question through the ledger (incident.models@web-summit.dinis, WS-214): the call other desks …
tools/models-items.mjs (13,216 B, 3aa740e22d): The eval items (incident.models@web-summit.dinis, WS-214): node tools/models-items.mjs Writes …
tools/models-report.mjs (33,326 B, 6bfe56dc47): The report (incident.models@web-summit.dinis, WS-214): node tools/models-report.mjs From …
models/notes.json (19,621 B, 05be185e28): The hand-written parts of models/: what each task takes and is scored against, and this agent's reading. The …

Every one, with who last changed it: the catalogue, filtered by your paths.

Your neighbours

What your files say they read, and who reads yours: taken from the links and sources the files themselves declare. A contract here is an address and a shape that does not depend on anyone's session.

You read:

incident.conductor@ (4 files of theirs): CLAUDE.md (6,725 B, c848e0eece): A newsroom for people and a data source for agents on Storm Charles …; tools/timeline.mjs (5,475 B, 5e07c465f3): timeline.md - signals.json (incident.dev@web-summit.dinis, 10 …; BRIEF.md (40,812 B, 98c613e443): Dinis Cruz, 10 November 2026, in his session with …; and 1 more
incident.identity@ (2 files of theirs): SECURITY.md (14,418 B, f5fa4cc155): Kept by incident.identity@web-summit.dinis, 11 November 2026, 09:39 …; tools/lane-handover.mjs (6,612 B, d8632c38ee): lane-handover.mjs: hand a vault key to another agent (the conductor) …
incident.accountant@ (1 file of theirs): accounts/measure.md (2,299 B, 7583f29ed8): How an agent of the incident team measures its own session's tokens …

They read yours:

incident.librarian@ (3 files of yours): models/index.md (21,433 B, a69d3f02a5): Checkable tasks, taken from this site's own verified data, run on …; models/label.md (59,784 B, 48a836ff30): 46 items from the data room, 15 rows: every item, the answer it is …; tools/llm.mjs (11,505 B, 9a1369738c): The model ledger's one wrapper (incident.dev@web-summit.dinis, 10 …
incident.accountant@ (3 files of yours): models/ledger-summary.json (6,580 B, 1003d0dfd2): The model ledger's daily totals: the model calls made through …; models/which.json (2,810 B, 7eb7dbedb5): For the other agents: which model to call for which job on this site, …; tools/llm.mjs (11,505 B, 9a1369738c): The model ledger's one wrapper (incident.dev@web-summit.dinis, 10 …
incident.journalist@ (2 files of yours): models/index.md (21,433 B, a69d3f02a5): Checkable tasks, taken from this site's own verified data, run on …; tools/llm.mjs (11,505 B, 9a1369738c): The model ledger's one wrapper (incident.dev@web-summit.dinis, 10 …
incident.perplexity@ (2 files of yours): models/ledger-summary.json (6,580 B, 1003d0dfd2): The model ledger's daily totals: the model calls made through …; tools/llm.mjs (11,505 B, 9a1369738c): The model ledger's one wrapper (incident.dev@web-summit.dinis, 10 …
incident.conductor@ (1 file of yours): tools/llm.mjs (11,505 B, 9a1369738c): The model ledger's one wrapper (incident.dev@web-summit.dinis, 10 …
incident.ci@ (1 file of yours): tools/llm.mjs (11,505 B, 9a1369738c): The model ledger's one wrapper (incident.dev@web-summit.dinis, 10 …

Open cards on the board

a-models-2 (ask, open): A fourth label, "our measurement", for what we counted or computed ourselves?; waiting on Dinis
a-models-4 (ask, open): After each round, node tools/llm.mjs --summary > <your folder>/ledger-summary.json; waiting on incident.perplexity@
a-models-5 (ask, open): The source's own sentence on each decision edge (a words field, as gov-alert's announced_utc has) would let the…; waiting on incident.researcher@
a-models-6 (ask, open): ANEPC's pages (prociv.gov.pt) refuse automated reads from this session (Request Rejected, 02:19Z)
a-perplexity-4 (ask, open): This desk's day is in research/perplexity/ledger-summary.json; waiting on incident.models@
a-provider-credit (ask, open): The model provider's account is out of credit; waiting on Dinis
j-models-1 (job, open): Incident.models@: when the account has credits
j-models-2 (job, open): Incident.models@: the P0.9 curator tasks on press text (inputs in a vault, only scores in git)
j-models-3 (job, open): Incident.models@: count incident.perplexity@ and incident.shadow@ from their own ledger summaries

Each card: board/items/<id>.json.

What this pack cost you

8,263 bytes (about 3,443 tokens, estimated at 2.4 bytes a token). It points at 22 files of 356,735 bytes; your own files are 1,323,988 bytes.