Add runtime SQLite query targets

This commit is contained in:
npc0-hue
2026-09-01 17:15:09 +08:00
parent 65da73eff0
commit 163fbda394
17 changed files with 708 additions and 114 deletions
+1 -1
View File
@@ -76,7 +76,7 @@ The executor resolves lifecycle action templates under the scoped server workspa
- Remote database and RCON jobs must use scoped input/artifact refs rather than embedding query or command bodies in job results.
- Run self-update, dependency, and log backfill jobs must use declared capabilities, logical target keys, scoped refs, and bounded result refs.
- Job payloads must not include logs, artifact chunks, raw host paths, raw credentials, direct sockets, or large inline result bodies.
- Process stdout/stderr must be redacted and sent through the asynchronous live
- Process stdout/stderr must be relayed verbatim through the asynchronous live
relay rather than embedded in progress/result bodies. Relay failure or drop
must not block job execution or create a local log backlog.
- Job ack/progress/result/cancel/reconcile calls are lightweight lifecycle metadata and must be able to complete while artifact/file transfer work is active or retrying.
+6 -5
View File
@@ -16,8 +16,8 @@ channel. Game plugins own durable log storage and analysis.
## Payloads
- `LogBatchIngestRequest`: run ID, session token, server instance ID, stream ID, source, sequence range, compression metadata, checksum, and bounded entries.
- `LogEntry`: sequence, timestamp, level, line, parser metadata, and redaction state.
- `LogBatchIngestRequest`: run ID, session token, server instance ID, stream ID, source, sequence range, compression metadata, checksum, and opaque entries.
- `LogEntry`: sequence, timestamp, level, opaque line, and parser metadata.
- `LogBatchIngestResponse`: accepted sequence range, latest stream metadata sequence, compatibility duplicate/retry fields, and server time.
- `LogStreamCursorRequest`: stream ID, sequence cursor, and limit.
- `LogStreamCursorResponse`: compatibility ordered entries, next cursor, and latest stored metadata sequence.
@@ -35,9 +35,10 @@ progress. A full in-memory relay queue or a failed relay request drops the
current event and never blocks, retries, or changes lifecycle state. Run does
not create a log spool, resend backlog, or platform delivery watermark.
Log relay payloads carry bounded redacted entries only and must not include
artifact chunks, file bodies, host paths, raw credentials, or direct socket
details.
Log relay payloads preserve supervised stdout/stderr and plugin-declared file
tail text verbatim. Run does not inspect, filter, redact, or content-truncate
those payloads; the channel still must not carry artifact chunks or direct
socket APIs.
## Browser Channel
+2 -2
View File
@@ -32,7 +32,7 @@ safe result `protected_request_unknown`; other transport failures use bounded
diagnostics without forwarding handler errors or response bodies. These errors
affect only the current request.
Management-program stdout and stderr are bounded, redacted, and sent to the
live log relay with source `management-program` and streams
Management-program stdout and stderr are sent verbatim to the live log relay
with source `management-program` and streams
`management-program.stdout` / `management-program.stderr`. They are not file
execution logs and are never embedded in job result content.