Conversation
jazelly
commented
Aug 18, 2026
jazelly
commented
Owner
- Prepare tokenless 0.6.0 as the release candidate from dev to main.
- Consume the 27 user-visible Changesets and persist a complete bilingual 0.6.0 section in packages/cli/CHANGELOG.md.
- Add the release-pending marker and aligned package-lock metadata so merge to main triggers the existing npm publish and GitHub Release workflow.
- Add durable tokenless agent run, read, resume, and cancel flows with authenticated daemon routing, local stdio MCP catalogs, exact approvals, and provider continuation.
- Add the standalone Web Agent Harness runtime with restart-safe local control-plane operations, durable SQLite state, and real MCP execution boundaries.
- Add the opt-in OpenAI- and Anthropic-compatible local API proxy with standard Chat Completions and Responses routes, structured outputs, tool-call history, continuation, explicit status codes, and tokenless/auto routing.
- Add setup-managed direct G4F provider integration, standard OpenAI streaming, provider catalog exposure, and retained native ChatGPT and Perplexity A/B adapters.
- Add Arena browser capabilities, Meta AI browser workflows, provider Browser/Direct mode controls, and evidence-backed provider routing diagnostics.
- Unify browser and direct image generation, reference-image editing, ChatGPT direct images, verified browser/G4F asset persistence, authenticated asset readback, and provider-specific image-surface fixes.
- Add FeatureBench agent runtime wiring, container tools, official evaluation verdicts, non-overwriting reports, and full real-provider E2E entry points.
- Improve dashboard daemon contracts, localized API job history, resident-browser tab cleanup, Page Ref ownership, provider response code-block preservation, and Linux x64 managed browser resolution.
- Validate the release metadata with release-pending verification and changelog extraction for 0.6.0.
- Validate the built product with npm run lint, npm run build, and the bounded full local test runner at 157 passing tests with zero failures or skips.
- Keep manual real-provider release gates as the final pre-publication prerequisite where the repository matrix requires them.
Existing projects that already speak the OpenAI or Anthropic wire format can now route their single-shot Q&A traffic through a visible provider session instead of a paid API, by pointing base_url at the daemon. New daemon routes, behind the existing control bearer token: POST /v1/openai/chat/completions GET /v1/openai/models POST /v1/anthropic/messages The proxy ships disabled. `tokenless api-proxy <status|enable|disable>` and an interactive setup step turn it on and pick how proxied turns map onto provider conversations: `new-conversation` flattens the transcript and starts a fresh conversation per request, while `continue-conversation` fingerprints all but the final user turn and reuses one provider conversation for that thread. `model` must name the provider explicitly as `tokenless/<provider>`. An unmapped model is rejected rather than redirected, so it stays visible which account and which subscription answered. Tool, function, and structured-output fields fail closed because visible pages expose no equivalent control. `stream: true` replays each dialect's documented event sequence as one terminal chunk, since a visible reply is only readable once it has finished rendering. Reported usage is always zero: Tokenless does not meter provider tokens. Also adds Linux x64 to the managed browser catalog with pinned CloakBrowser 146 and Chrome for Testing 146, discovers system browsers from their Linux install paths instead of falling through to the Windows branch, and extracts zip artifacts with unzip where GNU tar cannot read them. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
feat(harness): continue web-ai turns with a bounded MCP loop
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
2 participants
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.