v0.3.0

Separate Ollama chat protocol from Ollama-native capabilities.

Architecture:

• Record ADR 0001 in English and Chinese.
• Delegate ollama-cloud chat to the shared pi-ai OpenAI Chat Completions implementation.
• Keep Ollama-native /api/tags, /api/show, /api/web_search, and /api/web_fetch for discovery and Web capabilities.
• Map the configured native /api base URL to /v1 for chat only.

Reliability:

• Use provider-issued OpenAI tool-call IDs and replay tool results with matching tool_call_id.
• Pin Ollama-specific OpenAI compatibility: max_tokens, reasoning_effort, streaming usage, no store or prompt_cache fields.
• Preserve the 15-second Search/Fetch budget and one transient retry from v0.2.3.

Compatibility notes:

• Chat now requires models to be present in the saved catalog.
• GenerateOptions.stop remains unsupported by the shared PiAiAdapter.
• Existing llm-ollama settings, provider id, credentials, picker, and Search/Fetch provider ids remain unchanged.
• Old logs containing duplicate ollama-call-0 values are not migrated.

Validation:

• pnpm run check: 67 tests, build, and pack check passed.
• Live Ollama Cloud chat, tool calls, and Web Search verified.

v0.2.3

Fix Ollama tool-call identity and bound Ollama Cloud Web Search/Fetch requests.

• Use provider-issued tool-call IDs when Ollama returns them, with deterministic process-local fallbacks for older responses.
• Add a 15-second default per-attempt budget for Ollama Cloud Web Search/Fetch, with one transient retry.
• Keep redirect handling fail-closed and document the new webRequestTimeoutMs setting.

v0.2.2

Changes\n\n- Use Ollama /api/tags as the sole model discovery catalog source.\n- Remove unreliable HTML cloud catalog fetching.\n- Keep model-picker selections aligned with the current draft catalog.\n- Apply selections by replacing the catalog instead of appending duplicates.\n- Preserve current-only models in the picker.\n\n验证：105 个测试、构建和 package check 通过。

dsh-llm-ollama v0.2.1

• Discover cloud model cards from Ollama's /search?c=cloud for the default public endpoint.
• Merge and deduplicate cloud cards with /api/tags models.
• Remove HTML cloud suffixes without restoring them.
• Retry one transient cloud catalog connection failure.
• Add discovery coverage and update bilingual documentation.

dsh-llm-ollama v0.2.0

• Register Ollama Cloud /api/web_search and /api/web_fetch as web capability providers under the ollama-cloud id, reusing the chat credential and base URL.
• Switch a profile by pinning searchProvider/fetchProvider to ollama-cloud in its cordis.patch.yml (snippet in the README).
• Credentialed web requests reject redirects, so the API key cannot leak to a redirect target.
• The settings card now stays visible for remote browsers with a loopback-only explanation.

dsh-llm-ollama v0.1.5

• Keep the Ollama Cloud card visible in remote browsers instead of hiding it: the Harness configuration plane is loopback-only, so remote browsers now see an explanatory notice (matching the Codex Connect pattern).
• Document the loopback-only configuration limitation and the SSH-forward workflow in both READMEs.

dsh-llm-ollama v0.1.4

• Render the shell-overlay picker through the same body-level portal composition used by DSH modals.
• Keep the model picker above the Settings dialog instead of hidden behind its stacking context.
• Preserve the immediate loading state and bounded concurrent discovery added in v0.1.3.

dsh-llm-ollama v0.1.3

• Open the model picker immediately with loading and error states.
• Enrich model metadata with bounded concurrency while preserving provider order.
• Clarify that an entered unsaved key is used immediately for discovery.
• Reduce real Ollama Cloud discovery from tens of seconds to a few seconds.

dsh-llm-ollama v0.1.2

• Save the API URL and model catalog atomically so models persist after reopening.
• Select discovered models in a DSH-style frame overlay.
• Use semantic DSH theme tokens for visible Save-button text.
• Report GPT-OSS reasoning efforts as low/medium/high; /api/show only supplies the thinking capability.