- Verified that side-channel turns maintain stable prompt cache parity with main turns.
- Applied secret and tool obfuscation to ephemeral assistant side-channel data to prevent leakage.
- Updated `Agent` and `AgentSession` test suites to validate consistency under native dialect environments.
Mirror of the provider-anchored handoff test: a ~20k-token stored conversation
whose provider usage was deflated to 1k (as a before_provider_request compressor
would) now triggers pre-prompt compaction, proving #estimateStoredContextTokens
floors the decision through the real call-site (not just the helper math).
CI caught that flooring by the raw local estimate falsely triggers compaction on
thinking-heavy turns: estimateTokens counts the opaque thinkingSignature /
redactedThinking payloads (providers bill them on replay, #2275), but their local
byte size diverges wildly from what the provider actually charges — so a turn
with a large encrypted-reasoning blob but small provider usage would trip the
floor (broke agent-session-handoff 'provider-anchored usage' test).
estimateTokens now takes { excludeEncryptedReasoning } and the compaction floor
(#estimateStoredContextTokens) uses it: the floor counts only reliably-countable,
on-wire-compressible content (text, tool results, tool calls), while the provider
usage arm of compactionContextTokens still accounts for encrypted reasoning. This
keeps the encrypted-reasoning case provider-anchored while still flooring upward
when a before_provider_request hook compresses tool results.
A before_provider_request extension (a context-compression proxy like Headroom,
an obfuscator, or inline snapcompact) can shrink the outgoing request below the
real stored conversation. The provider then reports deflated prompt tokens, so
the auto-compaction threshold never fires and the stored history grows unbounded
until it overflows the context window and can no longer be compacted at all.
Add compactionContextTokens(provider, storedEstimate) = max(provider, estimate)
and apply it to both the pre-prompt and post-response compaction decisions,
flooring the provider-reported tokens by the agent's own estimate of the stored
conversation. Display and cost accounting still use exact provider usage; only
the compaction trigger takes the floor.
- Added `buildSideRequestContext` to the `Agent` class to generate prompt-cache-friendly provider contexts.
- Updated ephemeral side-channel turns to forward the full tool catalog to maintain prompt cache hit rates.
- Injected a `developer` role reminder into ephemeral turns to instruct the model to suppress tool calls.
- Implemented automatic post-processing to strip any tool calls from ephemeral turn responses.
- Exported message and dialect helper functions in `agent-loop.ts` to support context construction.
- Fixed session history desynchronization during `rewind` by performing a full context rebuild after applying branch changes.
- Prevented `rewind` tool output and assistant side-channel data from polluting the prompt cache by flushing and sanitizing session state.
- Added explicit test coverage for context reconstruction and assistant message sanitization after rewind events.
The prompt-inventory test sliced on the '# Inventory'/'ENV' markers from the
PR's merge-base prompt.md. On current main (chore: prompt reorder) the heading
is '# Tool Inventory' and the 'ENV' marker is gone, which is why the file was
deleted there. Accept either layout so the resurrected tests (incl. the SDK
'render provided tools' contract) pass after the cherry-pick.
fork() reset mnemopi conversation tracking directly but skipped the shared new-transcript reset, so the folded/promoted first-turn memory stayed in #baseSystemPrompt. The next turn re-recalled and the change-detection saw no diff, taking the fallback promotion path and injecting the <memories> block twice into the forked session prompt. Route fork() through #resetMemoryContextForNewTranscript() like the other reset paths and add a regression test asserting the forked prompt contains recalled memory exactly once.
The #3099 test omitted the startInAllScope flag, so it passed against the
buggy baseline too (empty folder defaults to folder scope regardless). Force
the removed flag via a cast to pin the real contract: even when a caller asks
for all-projects scope on an empty folder, the picker must stay folder-scoped.
Proven: passes on head src, fails on baseline src (renders '(all projects)').
resolveModels("all") expanded the full TINY_LOCAL_MODELS registry, which now
includes the qwen3-1.7b entry marked unsupportedReason. loadPipeline() throws
for such specs, so the download worker reported it as failed and the bulk
command exited with "One or more tiny title models failed to download" even
when every usable model downloaded. Filter unsupported specs out of the `all`
prefetch path; explicit single-model requests are unchanged. Addresses the
unaddressed Codex P2 on PR #3133.
Run threshold compaction maintenance when an active goal turn ends through a successful yield, while preserving the final-yield skip for non-goal completions.
Fixes#3146
The chunked welcome path cleared its snapshot progress timer before
writing the replica file and switching sessions. If that apply work
failed, the frame-apply catch only logged the error, leaving the
initial join promise pending with no welcome/progress timer left to
settle it.
Reject the pending initial join when a welcome or snapshot-chunk apply
fails before the join has completed, preserving reconnect-time logging
for already-joined guests. Add a regression test that forces the replica
write to fail and asserts /join rejects instead of hanging.
Fixes#3144
The host used to ship the entire transcript inside a single welcome
frame, so a multi-MB session spent the guest's 30s first-welcome
timeout on the relay transfer itself: ~1.3 MB took ~3s, ~4.2 MB took
~12s, and ~13.6 MB never arrived before the guest gave up with
'timed out waiting for the host's welcome'.
Bump COLLAB_PROTO to 2 and split the welcome:
- welcome carries metadata only (header, state, agents, entryCount,
readOnly) and lands in well under one second.
- a train of snapshot-chunk frames (SNAPSHOT_CHUNK_BYTES = 512 KB,
oversize entries ship alone) carries the transcript. Last chunk
flips final: true; an empty snapshot still emits one final chunk.
- the host queues welcome + chunks synchronously inside #handleHello,
preserving the host comment's ordering invariant (later broadcast
frames cannot interleave between them).
- the TUI guest accumulates chunks under a SNAPSHOT_PROGRESS_TIMEOUT_MS
that resets per chunk; only after final does it write the replica
jsonl, switchSession, and render. The first-welcome timeout still
guards arrival of the small welcome.
- the collab-web GuestClient streams entries into the snapshot as
chunks arrive and flips phase to 'live' on final.
Includes a contract test (in-process relay) asserting the welcome is
metadata-only, the chunk train fans the 1.5 MB synthetic transcript
across multiple frames with only the last marked final, and the
flattened entries match the source snapshot.
Fixes#3144
Stopped treating subprocess crash notifications as model-specific inference failures. Added regression coverage proving an unrelated queued title model can spawn a replacement worker after the crashed worker faults all pending requests.
Fixes#3132
Blocked the unsupported Qwen3 1.7B ONNX memory model before loading transformers and remembered local model execution failures so the client returns null instead of spawning another __omp_worker_tiny_inference process for the same failed model.
Fixes#3132
Stopped routing internal URL directory reads through the filesystem tree renderer so vault:// and '/data/workspaces/can1357__oh-my-pi__3116/.omp-session/2026-06-20T09-33-10-396Z_019ee460-817c-7000-8139-f8e2927809dd/local' keep their custom navigable listings.
Allow file-backed internal URL handlers to return existing directories as resources so read can list them and search/find can walk their source paths.
Fixes#3116
Restored and refreshed promoted memory prompts through the shared new-transcript reset path used by new sessions, handoff, branch, /btw, and cross-session switches.\n\nAdded coverage for newSession after first-turn memory recall so stale recalled memories cannot leak into the next transcript when recall returns no context.\n\nFixes #3111
Reset memory recall state before rebuilding the prompt during session switches, and clear fallback-promoted memory prompts when moving to another session.\n\nAdded a regression test that switches sessions after first-turn memory recall and verifies the next session does not receive stale memories.\n\nFixes #3111
Promoted first-turn memory recall into the stable base prompt so append-only sessions do not drop the memory block on the next turn and rebuild the provider prefix.\n\nAdded a regression test covering a memory backend that recalls once before the first model request.\n\nFixes #3111
The ASCII table renderer in `sqlite-reader.ts` shrank columns down to
`MIN_COLUMN_WIDTH=1` to fit the 120-cell budget. With ~20+ columns
(the reporter had 33) every multi-char cell collapsed to a lone `…`
and the final per-line `truncateToWidth(..., MAX_RENDER_WIDTH)` then
chopped the right edge — so the read tool returned a table of nothing
but ellipses with the rightmost cells missing entirely.
Bump the per-column floor to 3 (so cells always show at least two real
glyphs alongside the ellipsis) and, when the column count alone forces
the floor over budget, fall back to a per-row vertical block layout —
mirroring `psql`'s expanded display mode. Each row becomes a
`column: value` group with column names padded so colons align and
the value line truncated to the same 120-cell budget.
Fixes#3107
- Added a `proseOnlyThinking` configuration setting to suppress raw code blocks in AI thinking traces.
- Implemented `formatThinkingForDisplay` utility to replace code blocks with ellipses in the UI.
- Integrated runtime toggling and live refreshing of message components via streaming reveal controllers.
- Added a live tokens-per-second indicator to the assistant thinking pulse.
- Verified logic with new unit and integration tests for thinking block presentation.
- Added a windowed `SpeedTracker` to report average tokens-per-second during reasoning streams.
- Updated thinking animation from a dot pulse to a starburst effect with a dynamic speed badge.
- Engineered badges to fade from gray to accent color based on streaming throughput.
- Implemented automatic badge suppression during streaming lulls or for providers without live usage reporting.
- Added session-wide reset logic to prevent rate leaking between consecutive message turns.
can1357