intelligence-radar/docs/data-contracts.md

# Data contracts and commands

Python 3.11+; standard library only. Run commands from the checkout root.

## Source collection

```bash
python3 -m intelligence_radar ingest catalog/sources.json
python3 -m intelligence_radar validate
make check
```

Source registry: `sources` array with unique `id`, `publisher`, and HTTPS `url`.
The registry is trusted operator configuration; review URLs before adding sources.
Captures live in `data/captures/<id>.json`; original bytes are in
`data/artifacts/<content_sha256>`. Preserve both for replay. `ingestion-runs` records
success/failure per source. HTML changes may be incidental, not newsworthy.

Review actual source content before marking structured facts `reviewed`. Each
result, price and availability item requires `capture_id`, timezone-aware
`observed_at`, `review_status: reviewed`, and a named `reviewer`. Timestamps cannot
precede the supporting capture date. An evidence ID proves linkage, not correct
extraction. The curated initial catalog documents its missing measurement data.

## Capability Advisor

```bash
python3 -m intelligence_radar advise examples/source-backed-task.json \
  --catalog catalog/reviewed-2026-09-13.json --as-of 2026-09-13
python3 -m intelligence_radar report <recommendation-id>
python3 -m intelligence_radar replay <recommendation-id>
```

The source-backed example deliberately withholds winners. Its candidate benchmarks
lack verified versions/results and its deployment access has not been established.
This is distinct from the synthetic demonstration:

```bash
make demo-advisor
```

Task JSON: objective, acceptance, environment, risk, capability weights (0–1), hard
constraints and workload. See examples/task.json. Input and cache token buckets are
disjoint; output includes reasoning. Attempts are a declared scenario, never `1/fitness`.

Catalog JSON: benchmark definitions (id/version/constructs/health/limitations/scale),
exact profiles (id/model/harness/effort/provider/deployment/attributes/availability),
results (profile_id/benchmark_id/benchmark_version/raw_score/evidence), and prices
(profile_id/currency/valid_from/valid_until/usd_per_million/evidence). See
examples/advisor-catalog.json for a complete synthetic shape. A synthetic catalog
requires explicit `--synthetic` and never mixes with production evidence.

For benchmark health, provide all non-task-relevance weights named in the active
policy. Values are curator judgments, not automated validation. Duplicate IDs are
rejected. Missing, stale, future or conflicting results remain unknown. Price
validity is date-inclusive and was deliberately limited to the review day in the
initial live catalog. New days need new observations.

## Actual outcomes

```bash
python3 -m intelligence_radar outcome /tmp/outcome.json
```

An outcome contains recommendation_id, profile_id, unique run_id, accepted (boolean),
independent_verification (boolean), verifier_evidence (artifact reference or notes),
actual_usd, runtime_seconds, and completed_at (timezone-aware timestamp).
Outcomes are immutable and run IDs are idempotent; changed duplicate runs are rejected.
Keep the verifier artifact where its reference can be resolved. These are operator
assertions. No actual provider execution or paid inference happens in this command.

A subsequent recommendation uses matching outcomes available at its cutoff.
A historical replay uses its original frozen set. The current baseline reports
95% Wilson intervals, not a learned cross-task success predictor.

## Scheduled operation

```bash
python3 -m intelligence_radar run-due
python3 -m intelligence_radar health
systemctl --user status intelligence-radar.timer
```

`ops/schedule.json` configures daily 21:30 and Sunday 20:00 Europe/Berlin.
The shipped user timer checks every 15 minutes while this workstation's user service
manager is running. It does not promise execution while the workstation is off.
Each job is at most once per successful slot; failures remain retryable. Only the
latest missed slot is caught up. Runtime state lives in ignored `.runtime/`;
`data/scheduler-runs/` retains auditable receipts. Existing digests are preserved,
so review the receipt's digest_status before assuming a new draft was produced.

Install the reviewed units with `make install-timer`; uninstall/disable with
`systemctl --user disable --now intelligence-radar.timer`. That command leaves
source evidence and publications intact. No external app notifications are sent.


## Portable evidence bundles

```bash
python3 -m intelligence_radar export <recommendation-id> /tmp/report.tar
python3 -m intelligence_radar --root /tmp/radar-copy import /tmp/report.tar
```

Bundles retain the frozen recommendation, parent recommendations, referenced
capture receipts and original source artifacts. Import validates paths, checksums,
content IDs, dependency presence and replay in staging before writing to the
ledger. Existing conflicting evidence is refused. Import is bounded to 128 MiB
and 10,000 members; export is deterministic. External verifier references in
outcome notes are not automatically packaged.

Source configuration optionally accepts `max_bytes` from 1 through 20 MiB;
the default remains 5 MiB.

## Budgeted provider work samples

```bash
python3 -m intelligence_radar trials --config ops/openrouter-pilot.json \
  --suite examples/work-sample-suite.json --pilot
```

This command makes paid OpenRouter calls using `OPENROUTER_API_KEY` from the
process environment. The recorded authorization is USD 10 total for this campaign.
`--pilot` selects the first two cases for each configured model; omit it for all
20 cases. Do not change the campaign identifier to bypass existing accounting.

Reservations precede requests, completed trials retain response/usage/verifier
results, and successful identical runs return their retained result. There are
no automatic inference retries. Unsettled reservations stop new work; an actual
cost exceeding its reservation halts the campaign across restarts. The prompt
reservation uses UTF-8 bytes plus framing allowance and configured provider price
caps; it is an estimate, not a provider-enforced account spending limit.

The interrupted first request returned HTTP 401. Before resuming, supply a valid
credential and reconcile the retained reservation against provider billing.
Reconciliation is currently manual and must preserve old events. Never delete
reservations or assume a timeout was free. Controlled work samples and close
variants are not representative production outcomes; they are not automatically
added to advisor calibration.

### Review trials before admission

Add `--plan` to the trial command to render config/suite identities, exact run
IDs, request hashes and campaign accounting. It requires no key and makes no
network calls or ledger writes. Settled runs contribute actual recorded cost;
pending reservations are counted once; only unseen runs add new reservations.
Budget blockers are explicit. The plan always states `execution_authorized: false`:
it neither proves current provider prices/access nor grants native delivery.

`docs/evidence/2026-09-14-trial-{pilot,full}-plan.json` and
`2026-09-14-trial-inputs.json` are review snapshots. The latter hashes raw source
and input bytes; plan config/suite hashes use canonical JSON. The owner must
install and admit a fixed recipient separately, then rerender against current
billing before execution. Mutable ledger state and provider availability may
change after planning.