eric/audita
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 6 Wiki Activity

Document resilient reports and remove obsolete warning material

Browse Source
This commit is contained in:
Eric Rakestraw
2026-09-24 17:55:37 +00:00
parent 2f036b0356
commit a770c77a23
26 changed files with 346 additions and 536 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
docs/internal/llm-runtime.md
View File

@@ -6,7 +6,7 @@
prompt and schema assets, registers application backends, and inspects the
selected proposal and validation profiles offline. A role-bound `llm.RoleClient`
implements `structuredoutput.Client`; proposal and validator contracts alias that
interface during migration. Both roles share engine-local backend capacity.
interface. Both roles share engine-local backend capacity.
For each logical call, the adapter starts an operation context and a diagnostic
record before preparation. It supplies the reserved hardening text, the selected
@@ -15,12 +15,13 @@ contract. `PrepareExecution` produces the exact messages and hashes retained in
an Audita snapshot. The adapter checks the prepared backend and endpoint before
`RunPrepared`. It records available final content, usage, model, and profile
metadata and decodes only a validated result. Call hooks expose logical
start/finish records and PromptKit execution events for the report collector.
start/finish records and PromptKit execution events for debug utilization.
The [integration contract](../integrations/openai-compatible-llm.md) describes
wire ownership and failure behavior. `structuredoutput.Failure` has a safe
message and a typed cause; only exhausted structural repair is a malformed
output failure. Secrets from all configured backends, including unselected
message, fixed reason code, typed cause, and recoverable/fatal/canceled class.
Counts distinguish provider attempts, transient retries, and structural
repairs, including on errors. Secrets from all configured backends, including unselected
ones, are available to the diagnostics redaction boundary.
## Orchestration
@@ -29,10 +30,10 @@ CLI processing supplies two role clients from the same engine to the runner.
PromptKit's per-backend admission applies to both roles and every generation,
retry, and repair. The runner still processes module instances in order,
collects concurrent section work deterministically, and cancels remaining
section work on a runtime failure. `llm.DiagnosticsWriter` remains the protected
section work only on fatal failures. `llm.DiagnosticsWriter` remains the protected
request/response artifact writer. `llm.ObservationCollector` captures metadata
events under a short mutex and freezes them after runner work joins. Its logical
records include preparation failures; summary usage comes from terminal events
once per call and retains knownness and completeness for each field. Runner wall
and validator timings remain separate from provider attempts. The report and
artifact mapping is completed by the following roadmap work.
and validator timings remain separate from provider attempts. Ordinary reports
derive compact runner outcomes; full observations are debug-only.