Catalog refresh, incremental persistence, and a manual refresh tool#16
Merged
msitarzewski merged 3 commits intoJun 22, 2026
Merged
Conversation
Add Claude Opus 4.8/4.7, GPT-5.5, GPT-5.4 mini, and Gemini 3.5 Flash; drop deprecated o3. Correct Anthropic GA context windows (Opus 4.6 and Sonnet 4.6 -> 1M; Sonnet 4.5 stays 200k, its 1M is beta-gated). Add gpt-5.5 to NO_TEMPERATURE_MODELS and _REASONING_EFFORT_MODELS after empirically confirming it rejects temperature. Model IDs sourced from live provider APIs, pricing cross-checked against the truefoundry/models feed, temperature behavior verified against the live OpenAI endpoint. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EkrekgzMAQko92UkjnXhHL
Introduce IncrementalPersister (memory/persist.py): the thread is created up front as "active", each round is committed as soon as it finishes, and the thread is finalized to "complete" at the end. A crash mid-run now leaves a real partial thread instead of nothing. The WebSocket streams a thread_started event with the real thread ID so clients can deep-link mid-run. Consolidates the three previously-duplicated persist paths (CLI, WS, and the batch convenience) into one shared module and adds ConsensusContext.snapshot_round() so a finished round can be persisted before it is archived. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EkrekgzMAQko92UkjnXhHL
scripts/refresh_catalog.py diffs the model catalog against the truefoundry/models feed and the live provider APIs. It reports price/context/status changes, hashes a per-model field projection against catalog_snapshot.json to show what changed since the last run, discovers new frontier models via the live APIs, and empirically probes OpenAI temperature support (the one field the feed gets wrong). Propose-only: it never modifies catalog.py. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01EkrekgzMAQko92UkjnXhHL
msitarzewski
deleted the
feat/catalog-refresh-and-incremental-persistence
branch
This was referenced Jun 22, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Three independent changes, one per commit.
1. Catalog refresh (
feat(catalog))statusfield)gpt-5.5toNO_TEMPERATURE_MODELS+_REASONING_EFFORT_MODELSafter empirically confirming it rejects temperatureSourcing: model IDs from live provider APIs, pricing cross-checked against the truefoundry/models feed, temperature behavior verified against the live OpenAI endpoint. No fabricated values.
2. Incremental persistence (
feat(persist))IncrementalPersister: thread createdactiveup front → each round committed as it finishes → finalizedcompleteat the endthread_startedevent with the real thread ID for mid-run deep-linking3. Manual catalog refresh tool (
feat(scripts))scripts/refresh_catalog.py— propose-only, run on demandcatalog_snapshot.jsonto show what changed since last run, discovers new frontier models, and empirically probes OpenAI temperature (the one field the feed gets wrong)catalog.pyTesting
Not included (follow-up)
/api/askstill uses its lite persist path; unifying it onto the incremental path is the noted next step.🤖 Generated with Claude Code
https://claude.ai/code/session_01EkrekgzMAQko92UkjnXhHL