GitHub Release: can1357/oh-my-pi v18.1.0 (01.09.2026)

$ git clone https://github.com/can1357/oh-my-pi.git Clone

@oh-my-pi/pi-ai

Added

• Added an optional completeSimple callback that observes every result, including results from internal thinking-loop retries.


• Added compatibility options for Anthropic-compatible proxies that reject context_management and OpenAI Responses proxies that provide incomplete reasoning-summary streams.


• Added ClinePass API-key authentication via the official CLINE_API_KEY environment variable, with account validation, actionable subscription and quota errors, support for eligible ClinePass model rosters, and rolling quota-window reporting in omp usage.


• Added Devin router-model support, including assignment of the concrete model before each request, routed-model metadata, credit usage reporting, and plan, quota-window, and account details through omp usage.

Changed

• Provider behavior is now driven by each model's resolved compatibility, identity, thinking, and behavior policies rather than model-name matching, improving support for model-specific request formatting, vision, reasoning, routing, pricing, and quota handling.


• Devin integrations now use the current released CLI identity and support parallel tool calls when the model declares that capability.

Fixed

• Fixed OpenAI remote-compaction replay for persisted sessions, allowing sessions with previously stored compaction items to resume successfully.


• Fixed Cursor Fable requests failing when advertised tools used JSON Schema composition keywords.


• Fixed Z.AI (GLM Coding Plan) browser sign-in by using the registered CLI callback address.


• Fixed OpenAI Codex/Responses tool results being lost when composite call identifiers could not be paired with the corresponding assistant call.


• Fixed native OpenAI Responses history replay becoming stuck on malformed or truncated function-call arguments; invalid history items are now discarded so the session can recover.

@oh-my-pi/pi-catalog

Added

• Added GitLab Duo model support.


• Added provider support for Llama.cpp, LM Studio, and Minimax.


• Added catalog entries for Qwen 3.8 27B, Granite 4.2 8B, Abliterated variants, GLM 5.3, and Qwen 3.8 Flash Next.


• Added compatibility metadata for context management and reasoning summaries on supported provider-compatible model endpoints.


• Added the native ClinePass provider with live model discovery, subscription and free-tier model listings, current limits and pricing, model modalities, and per-model reasoning controls. Subscription models display API-equivalent pricing while free-tier models display as free.


• Added native Devin provider discovery with current Cascade model capabilities, pricing, limits, reasoning controls, selector aliases, and model descriptions and recommendation/beta metadata. Added static SWE-1.6 defaults so the provider can resolve a default model before account-scoped discovery completes.


• Added per-tier model pricing and long-context pricing support, along with expanded catalog metadata for model limits, API routes, input modalities, and provider-specific model aliases.

Changed

• Updated pricing data across the catalog to reflect current provider rates.


• Standardized model display names for greater consistency.


• Improved model compatibility classification and variant selection using structured model identities, providing more reliable detection of model families, revisions, reasoning variants, and provider-specific capabilities.

Fixed

• Fixed compatibility detection for vendor-prefixed GLM models on Mistral and Cerebras, restoring the appropriate tokenizer and reasoning-history behavior.


• Removed an obsolete OpenCode provider entry that exposed unavailable Zen models in the model picker.


• Fixed model revision detection for parameterized model IDs, including correct classification of Qwen, Fireworks Kimi, and Cursor-wrapped Grok models.


• Fixed Kimi K3 reasoning-effort handling on OpenAI-compatible hosts such as LiteLLM and vLLM.


• Fixed pricing for Codex Daybreak aliases, including worker variants.


• Fixed reasoning-effort variants for local Ollama models and DeepSeek V4, and corrected prompt-cache support for applicable Bedrock Nova deployments.


• Fixed Devin model variants with separate thinking and context-size options so all supported routing combinations and context limits are preserved.


• Corrected Devin SWE-1.6 and SWE-1.6 Fast capability metadata so image attachments use the appropriate text-only fallback instead of being silently discarded.


• Devin discovery now warns when the service returns an empty native model catalog, helping identify stale or incompatible CLI credentials.

Removed

• Removed legacy DeepSeek V3 variants from the Novita catalog.

@oh-my-pi/pi-coding-agent

Added

• Added the /trace slash command to display session trace URLs in the stats dashboard.


• Added support for OpenAI-compatible gateways whose model-list endpoint is rooted at a versioned URL, with an injectV1: false discovery option to request {baseUrl}/models directly.


• Added provider-reported credits and concrete routed-model counts to /session statistics.


• Added CLINE_API_KEY to CLI environment help for native ClinePass subscription inference.


• Expanded Devin model selection to support native CLI aliases, dotted upstream model names, and raw effort-route identifiers.


• Added provider-supplied model metadata to /models, including new, beta, and recommended badges plus model descriptions.


• Standalone CLAUDE.md files in project and ancestor directories are now loaded as context alongside AGENTS.md, while preserving config-directory precedence.


• Added an Activity view to Agent Hub with searchable and filterable timelines spanning live progress and persisted transcripts; /hub is now the live-operations entry point while /agents retains Control Center behavior.


• Added an icon.advisorClosed symbol-theme token: the advisor eye in the status line now closes once the advisor has finished reviewing and will not add further comments.

Changed

• Disabled hashline editing for Kimi, Mimo, DeepSeek Flash, and Stepfun models for improved stability.


• Reworked /usage into a fullscreen dashboard overlay (no transcript output): a compact per-provider subscriptions grid with untouched providers collapsed into one line, a GitHub-style daily activity heatmap fed by local stats, and the classic full report one keypress away.


• Reworked transcript navigation with a fullscreen rewind selector opened by double-Escape, supporting rendered-item navigation, user-turn jumps, branching rewinds, and alternate session-tree branch selection.


• Updated /copy to use the fullscreen transcript selector, allowing users to copy a turn or navigate into nested content such as code, quotes, commands, and tool output.

Fixed

• Improved edit-tool error guidance for operations missing the » separator, identifying redundant context-only operations


• Fixed OAuth provider modifyModels projections being silently dropped after a discovery refresh introduced live-config headers.


• Edit-tool ＋/－ line operations now match their anchors leniently across whitespace drift (indentation, blank-line miscounts) instead of failing with a byte-for-byte error; a note reports the lenient match.


• Fixed an edit-tool REWRITE consisting only of ＋ add lines silently replacing (deleting) the matched text; it now inserts after the kept MATCH.


• Edit-tool no-match errors now name MATCH lines that exist nowhere in the file and suggest marking them with ＋, and errors without a located region no longer append a misleading file-head "closest match" preview.


• Fixed ordinary CLI startup eagerly loading the computer worker graph (native desktop addon and early environment), restoring lazy startup and profile .env ordering.


• Fixed online auto-thinking classifier usage being omitted from session token and cost totals.


• Fixed image generation with custom provider endpoints when using openai-codex credentials and a non-OpenAI chat model.


• Fixed custom hook UI factories not receiving the documented keybindings argument.


• Fixed MCP OAuth token exchange for authorization endpoints that use a different resource indicator.


• Fixed custom extension web_search tools being shadowed by the built-in search tool.


• Fixed Agent Hub task boards collapsing to summary rows after returning from a focused session.


• Improved Linux ARM64 browser startup messaging when managed Chrome for Testing builds are unavailable, with guidance for using system Chromium or PUPPETEER_EXECUTABLE_PATH.


• Fixed resuming image-heavy sessions that previously terminated while replaying transcripts.


• Fixed custom agents declaring hub being incorrectly treated as read-only.


• Restored compatibility for legacy Pi extensions that import calculateContextTokens or use the synchronous SettingsManager.create() API.


• Fixed custom model overrides being lost during configuration updates.


• Clarified that the default task-delegation setting follows the selected model's policy.


• Fixed /rename without a title interrupting active session activity.


• Fixed the Nerd Font notification persisting incorrectly after theme configuration.


• Fixed sampling parameter errors with newer Anthropic models.


• Long OpenCode Go usage-limit waits now switch replay-safe turns to a configured alternate provider when the delay exceeds retry.maxDelayMs.


• Fixed OpenAI Codex Responses tool results being lost when composite and plain tool-call identifiers did not match.


• Fixed /tan background agents failing to resolve credentials for providers supplied by extensions.


• Fixed Mnemopi saving session transcripts on exit when automatic retention is disabled.


• Fixed configuration writes through chained symlinks so the final target and intermediate links are preserved.


• Fixed direct tool calls using full xd:// device URLs.


• Fixed command-backed headers in custom discovery providers being resolved for discovered models.


• Fixed Windows drive paths pasted under WSL being resolved through their /mnt/<drive> mounts for images and file reads.


• Improved sloppy/SPARSE edit no-match guidance so low-confidence matches are clearly presented without unsafe copy-ready operations.


• Fixed agents in Hub wait loops failing to respond to user steering messages.


• Fixed /tan sessions inheriting parent costs and overstating subagent totals.


• Fixed prompt action labels being truncated.


• Fixed assistant text being truncated when a tool call begins during streaming.


• Fixed the advisor dropping concerns when catching up on multiple turns and improved review context with bounded tool-result excerpts plus complete ask exchanges.


• Fixed bash command timeouts being delayed by child processes holding output pipes open, while improving timeout reporting and cleanup.


• Fixed retry countdowns and capped-wait errors displaying floating-point noise in millisecond durations.


• Prevented browser app.path from terminating existing same-executable applications when no reusable CDP endpoint is available.


• Fixed top-level errors overwriting the active composer before terminal restoration.


• Fixed Enter being ignored during the first turn when omp starts with an initial prompt.

@oh-my-pi/pi-natives

Fixed

• Fixed TTY output backpressure reporting so pending write progress is accurately reflected during large writes.

@oh-my-pi/snapcompact

Added

• Added a declarative compatibility rules system for consistent model identification, capabilities, policies, and provider-specific behavior across model classes, families, and revisions.


• Added the compat-compiler CLI for managing model identity and capability rules through KDL configuration files.

Changed

• Standardized model revision handling and compatibility resolution across model discovery and runtime behavior.

@oh-my-pi/omp-stats

Added

• Added API for daily activity tracking, including cost, requests, and token usage aggregates


• Added a Traces dashboard for detailed session analysis, including an interactive timeline, searchable and filterable transcripts, token and cost summaries, and aggregate tool-usage and timing insights.

Fixed

• Fixed omp stats omitting usage from online auto-thinking classifier calls.

@oh-my-pi/pi-tui

Fixed

• Improved terminal stability when resuming image-heavy sessions, preventing large transcript repaints from being mistaken for stalled output or exceeding the terminal output limit.


• Fixed inline images leaving blank rows in Herdr panes when resuming or rendering sessions in nested terminals.


• Fixed the TUI crashing on reference-style Markdown links whose labels match JavaScript built-in names; these links now render safely as plain text.


• Fixed fatal cleanup leaving the cursor inside a focused input before error output is displayed.


• Fixed resumed sessions showing stale background bands until the next keypress in WSL and Windows Terminal.

@oh-my-pi/pi-utils

Added

• Added postmortem.fatal for terminal-safe top-level failure reporting.

Fixed

• Fixed sub-second duration formatting so it no longer exposes floating-point precision noise.


• Managed Chrome-for-Testing downloads now reject unsupported Linux ARM64 hosts instead of installing an incompatible x86_64 browser.


• Fixed Markdown reference-link labels that match built-in Object.prototype names, such as constructor and __proto__, so they are no longer misidentified as definitions or emitted with an undefined URL ( in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in in made their first contribution in made their first contribution in made their first contribution in made their first contribution in made their first contribution in made their first contribution in
  Vollständiger Original-Bericht

  Ausführliche Details, Code-Beispiele & Hersteller-Stellungnahme auf github.com.

  ↗ Original-Artikel auf github.com lesen