coulomb/intelligence-radar
5
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Add portable evidence bundles and budgeted OpenRouter trials

Browse source 
Assistant: codex
Assistant-Model: gpt-6-astra
Assistant-Session: 01a09cbc-910e-77a0-b336-3130194350b3
This commit is contained in:
tegwick 2026-09-13 23:50:19 +02:00
parent e7102ccb53
commit 919aefbef5
22 changed files with 1444 additions and 18 deletions
Download patch file Download diff file Expand all files Collapse all files

44
docs/data-contracts.md
View file

@ -92,3 +92,47 @@ 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.

27
docs/evidence/2026-09-13-resume.md Normal file
View file

@ -0,0 +1,27 @@
# Interrupted work recovery — 2026-09-13
Recovered the earlier user authorization: OpenRouter within USD 10. The previous
session's first inference returned HTTP 401 Unauthorized. A fresh read-only
`GET https://openrouter.ai/api/v1/key` also returned 401; no inference was retried.
The existing campaign has USD 0 recorded settled spend and USD 0.023712 pending.
Zero settled spend is not a verified billing statement. No real outcome is claimed.
Credential routing was checked with `warden route find` and `warden route show`.
The matching `openrouter-llm-connect` playbook explicitly scopes its consumer to
llm-connect in activity-core, not a generic OpenRouter client. That service key
was not retrieved. Resume needs a valid local OPENROUTER_API_KEY and reconciliation
of the existing reservation; no secret values belong in Git, Statehub or chat.
Completed local work: portable recommendation bundles, bounded source-size
configuration, budgeted trial execution with deterministic structured verification,
and default Makefile help. Review fixes persist campaign halts after an overrun,
stop other cases while billing is uncertain, and reject omitted bundle dependencies.
Validation: `make check` passed all 32 tests plus repository/archive integrity;
`make` lists targets; `git diff --check` passed. Source-backed evidence remains
incomplete for eligible comparable profiles. Vendor-reported results retain their
execution conditions and are not transferred to direct API profile rankings.
IR-WP-0004 remains active: T01 and T03 in progress, T02 waiting for credentials
and billing reconciliation. Hosted operation, curated publication, decomposition
comparisons and representative production calibration remain open in those tasks.