LLM/moltbot
1
0
Fork 0
mirror of https://github.com/moltbot/moltbot.git synced 2026-05-11 12:58:34 +00:00
Code Issues Packages Projects Releases Wiki Activity

docs: update codex context engine sqlite plan

Browse Source
This commit is contained in:
Peter Steinberger
2026-05-09 18:00:57 +01:00
parent 9ccb61789c
commit e3608e7fa4
1 changed files with 13 additions and 14 deletions
Download Patch File Download Diff File Expand all files Collapse all files

27
docs/plan/codex-context-engine-harness.md
View File

@@ -263,26 +263,25 @@ supplementing thread history, swap this projection layer to use that API.
In `extensions/codex/src/app-server/run-attempt.ts`:
- Read mirrored session history as today.
- Determine whether the session file existed before this run. Prefer a helper
that checks `fs.stat(params.sessionFile)` before mirroring writes.
- Open a `SessionManager` or use a narrow session manager adapter if the helper
requires it.
- Determine whether SQLite already has transcript rows for `{agentId, sessionId}`
before mirroring writes.
- Use the SQLite transcript scope helpers; do not open a transcript file or
derive a locator.
- Call the neutral bootstrap helper when `params.contextEngine` exists.
Pseudo-flow:
```ts
const hadSessionFile = await fileExists(params.sessionFile);
const sessionManager = SessionManager.open(params.sessionFile);
const historyMessages = sessionManager.buildSessionContext().messages;
const transcriptScope = { agentId: params.agentId, sessionId: params.sessionId };
const historyMessages = readMirroredSessionHistoryMessages(transcriptScope);
const hadTranscriptRows = historyMessages.length > 0;
await bootstrapHarnessContextEngine({
hadSessionFile,
hadTranscriptRows,
contextEngine: params.contextEngine,
sessionId: params.sessionId,
sessionKey: sandboxSessionKey,
sessionFile: params.sessionFile,
sessionManager,
transcriptScope,
runtimeContext: buildHarnessContextEngineRuntimeContext(...),
runMaintenance: runHarnessContextEngineMaintenance,
warn,
@@ -366,15 +365,15 @@ best available message snapshot:
- Prefer full mirrored session context after the write, because `afterTurn`
expects the session snapshot, not only the current turn.
- Fall back to `historyMessages + result.messagesSnapshot` if the session file
cannot be reopened.
- Fall back to `historyMessages + result.messagesSnapshot` if the SQLite read
fails.
Pseudo-flow:
```ts
const prePromptMessageCount = historyMessages.length;
await mirrorTranscriptBestEffort(...);
const finalMessages = readMirroredSessionHistoryMessages(params.sessionFile)
const finalMessages = readMirroredSessionHistoryMessages(transcriptScope)
?? [...historyMessages, ...result.messagesSnapshot];
await finalizeHarnessContextEngineTurn({
@@ -384,7 +383,7 @@ await finalizeHarnessContextEngineTurn({
yieldAborted,
sessionIdUsed: params.sessionId,
sessionKey: sandboxSessionKey,
sessionFile: params.sessionFile,
transcriptScope,
messagesSnapshot: finalMessages,
prePromptMessageCount,
tokenBudget: params.contextTokenBudget,