Read and verify the audit trail over the wire, not by client-side file parsing
I maintainer di solito rispondono entro 1 giorno
@kikashy ci sta già lavorando.
Dal 23/10/2026.
Valutazione
- Difficoltà
- 4/5
- Tempo stimato
- 3-5 giorni
- Idoneità per principianti
- 48/100
Direzione di ricerca
Start by locating the MCP tool registration and the audit code used by audit verify and audit repair; the issue names no files or tests. Trace how the audit directory stores evaluations, signatures, and stamps, then identify the existing MCP reference and conformance rows. Done means the tools provide paged trail data and a verify report with --expect and --trail equivalents, with tests proving byte equality and updated reference documentation.
Scritto dal modello di indicizzazione a partire dal testo della issue.
Descrizione
Remaining scope and proposal
The bounded tools and larger history sources below are included in the cumulative Runtime v0.38.0 candidate, prepared and merged by #325 (d895416771b2f8de2b5b37bafd9c30c2404d42dc). No v0.38.0 tag or publication is implied. This issue remains open under v0.39.0 for larger raw exports, larger verification and staged supplied-input contracts.
Merged #326 (e88492f482519ea6e36bd00ed186142aecfb4db9) records a proposal for bounded resumable raw downloads. Its independent review and targeted fix confirmation resolve the dispatch/context and terminal cancellation findings; the document remains PROPOSED / NOT IMPLEMENTED. Candidate budgets, platform capture, bounded transport and implementation evidence remain adoption gates. It changes no current tool or source limit and does not cover larger verification.
Merged history follow-up
Merged #324 (b51646e57d207f5b733d25fde68f26d4d2d2fdb5) decouples history's source-size limit from its fixed 4 MiB read/recheck window. It admits configured sources above 64 MiB up to the exact address ceiling 9007199254740990 (2^53-2), while preserving output, session and authority budgets. This is an addressability bound, not a filesystem-capability promise. Real 64 MiB boundary tests, synthetic high-offset tests, four caught mutants, all nine local checks, independent review and final-head CI passed. ADR 0069 is accepted. Raw export/configured verification and supplied-input limits remain unchanged, so this issue stays open.
Delivery status — 2026-10-23
Merged #313 supplies experimental_list_audit, experimental_list_audit_history and experimental_verify_audit, with ADR-0065, MCP reference documentation and Runtime-specific contract rows. Tool-only tests reconstruct all three files byte-for-byte and reproduce Desk-shaped history; CLI/MCP verification tests compare the full reports. These are Runtime convention checks, not a new Core conformance claim.
The delivered parity is explicitly bounded: raw-export captured files and configured verification trails are at most 64 MiB; history uses the existing Desk 4 MiB window and newest-50 selection, with the larger source-address bound delivered by #324; supplied verification trails are at most 8 MiB, and supplied checkpoints at most 1 MiB. The dedicated history tool handles the raw listing's 64 KiB structured-line and Unicode-selection differences within those budgets.
This issue remains open. Unrestricted parity beyond those source/input budgets is not claimed. Completing that broader scope needs an explicit bounded continuation/verification contract and corresponding consumer evidence; it must not silently raise resource limits or change Desk pagination behavior. Desk integration belongs to its separate active session and has not been changed here.
Evidence: MCP audit reference, contract rows.
Original request
Why
The Desk reads the audit trail (evaluations.jsonl, signatures.jsonl, stamps.jsonl) by parsing the files the runtime wrote, for the definition history and the trail download, and shells out to audit verify and audit repair over the same files. A client that does not share a filesystem with the trail cannot show or check it, and the reading of a record is best kept in the component that writes it. A wire method is preferred over client-side file parsing.
Scope
- Experimental MCP tools over the configured audit directory: a paged listing of the trail (evaluations, signatures, stamps, with the same members the files carry) and a verify report as
audit verifygives it, with--expectand--trailequivalents. - The CLI unchanged; the tools call the same code.
Acceptance
- The Desk's definition history and trail download can be produced from the tools alone, byte-equal to the file-based reading, shown by a test.
- Documented in the MCP reference; conformance rows.
- Lingua principale
- Go
- Stelle
- 2
- Fork
- 6
- Merge medio
- 3h 3m
- PR unite (30g)
- 99
Preparare l'ambiente
- Nessun Dockerfile né file Docker Compose
- Ha un modello di pull request
- Leggi la guida per i contributori
Come iniziare
- Leggi tutta la issue e poi la guida ai contributi del progetto.
- Commenta sulla issue per dire che te ne occupi tu — evita che due persone facciano lo stesso lavoro.
- Fai un fork del repository e lavora su un branch.
- Apri una pull request che faccia riferimento al numero della issue.
Altre issue di Judgment-Pack/judgment-pack-runtime
-
Experimental line: evaluate reusable subgraphs per RFC 0022 (spec #129)Forse già presa @kikashy l’ha presa 1 giorno fa. Apertaenhancement spec
Difficoltà 5/5 Più di una settimana Idoneità per principianti 25/100
Judgment-Pack/judgment-pack-runtime#307 · 1 assegnatario ·
I maintainer di solito rispondono entro 1 giorno
Tutte le issue di Judgment-Pack/judgment-pack-runtime
Issue simili
-
Project submission:Aperta
Difficoltà 1/5 Meno di un'ora Idoneità per principianti 72/100
slavakurilyak/awesome-ai-agents#845 ·
I maintainer di solito rispondono entro 1 giorno
-
bug
Difficoltà 2/5 1-3 ore Idoneità per principianti 68/100
I maintainer di solito rispondono entro 3 giorni
-
Difficoltà 2/5 1-3 ore Idoneità per principianti 85/100
-
Difficoltà 2/5 1-3 ore Idoneità per principianti 74/100
JakeChampion/lang#13132 ·
I maintainer di solito rispondono entro 1 giorno
-
[BUG] deploy.restart_policy without condition drops the restart policyForse già presa @akshat-lakhera l’ha presa oggi. Aperta
Difficoltà 2/5 1-3 ore Idoneità per principianti 82/100
docker/compose#14383 · 1 commento ·
I maintainer di solito rispondono entro 1 giorno