feat: improve configuration, memory management, and docs

[3210448723]

Summary

This pull request improves DojoAgents documentation, provider configuration,
generated-memory management, and runtime reliability.

Changes

• Add bilingual agent internals documentation covering the runtime, agent loop,
  tools, events, memory, plugins, and multi-agent planning.
• Add bilingual known-limitations documentation with explicit behavioral
  boundaries and validation status.
• Extend OpenAI-compatible provider configuration with token and context-window
  settings, including dashboard configuration support.
• Add a dashboard API for deleting generated memory with path validation and
  post-deletion verification.
• Improve portfolio migration compatibility and market-data fallback behavior.
• Honor both provider and request token limits for MCP sampling.
• Avoid unnecessary turn-intent classification when no conversation history is
  available.
• Update affected tests and stale fixtures to match current runtime contracts.

Validation

• 141 targeted regression tests passed.
• Ruff checks passed.
• MkDocs strict build passed.
• Dashboard frontend build passed.

The full test suite was not reported as completed because the single-process run
encountered prolonged filesystem I/O stalls.

Known dependency issue

npm audit currently reports one high-severity issue in the transitive
postcss@8.5.15 dependency. This pull request does not modify the dependency
lockfile and records the issue as a known limitation.

[Copilot]

Pull request overview

This PR improves DojoAgents’ configurability and operational ergonomics by extending OpenAI-compatible provider limits, adding a Dashboard API to clear generated skill-memory safely, improving market-data fallback behavior, and significantly expanding bilingual internal documentation (including a “known limitations” register).

Changes:

• Extend provider configuration to support context_window and max_tokens, propagate limits through runtime/dashboard/MCP sampling, and add frontend form controls.
• Add a Dashboard DELETE /api/v1/memory/generated-skills endpoint backed by a new generated-memory service with deletion verification and safety checks.
• Improve runtime reliability/performance (skip turn-intent classification when no history; bulk kline snapshot indexing + per-symbol fallback) and update tests/docs accordingly.

Reviewed changes

Copilot reviewed 63 out of 63 changed files in this pull request and generated 2 comments.

Show a summary per file

File	Description
tests/test_turn_intent.py	Adds regression coverage for skipping turn-intent classification without history.
tests/test_runtime_multi_agent_plan.py	Ensures runtime wires OpenAI-compatible provider routing and output limits correctly.
tests/test_openai_provider_usage.py	Expands provider behavior tests (minimal request shape, author-prefixing, streaming tool-call metadata, reasoning field parsing).
tests/test_mcp_advanced.py	Updates MCP tests and adds coverage for sampling output-limit min(config, request).
tests/test_gateway_state.py	Updates integrated loop expectations to account for additional LLM calls (e.g., intent classifier).
tests/test_dashboard_config_update.py	Ensures dashboard config updates sync agent model when default provider changes.
tests/test_dashboard_chat_openai.py	Verifies runtime sync preserves OpenAI-compatible provider limits and context window.
tests/test_config_multi_agent_plan.py	Adds config parsing coverage for provider max_tokens.
tests/test_cli_gateway_setup.py	Adjusts gateway setup tests to capture log output instead of stdout where appropriate.
tests/test_built_in_plugins.py	Removes the example-plugin loading test (likely reflecting plugin set/behavior changes).
tests/test_agent_loop_integrated.py	Updates integrated loop assertions to account for additional LLM call(s).
tests/dashboard/stores/test_portfolio_migration.py	Updates portfolio migration tests for current schema normalization and backup behavior.
tests/dashboard/routers/test_memory.py	New tests for generated-memory deletion endpoint behavior and safety edge cases.
tests/dashboard/routers/conftest.py	Extends test doubles to match updated router/service expectations.
mkdocs.yml	Adds new bilingual nav entries for “Agent Internals” and “Known Limitations”.
dojoagents/tools/mcp_tool.py	Honors both provider and request token limits for MCP sampling; aligns SSE import for patchability in tests.
dojoagents/dashboard/web/src/types/settings.ts	Extends provider form types to include context_window and max_tokens.
dojoagents/dashboard/web/src/components/settings/SettingsModal.tsx	Adds UI fields and patch logic for provider context_window / max_tokens.
dojoagents/dashboard/services/portfolio_store.py	Adjusts migration logic to preserve an explicit v2 intermediate state before v3 conversion.
dojoagents/dashboard/services/generated_memory_service.py	Introduces a generated-memory clearing service with path validation and deletion verification.
dojoagents/dashboard/services/dojo_data_gateway.py	Adds bulk kline indexing (warm/cache) with correct fallback behavior for windows and missing symbols.
dojoagents/dashboard/server.py	Propagates provider max_tokens into runtime provider wiring; registers new memory router.
dojoagents/dashboard/schemas/memory.py	Adds response schema for generated-memory clearing endpoint.
dojoagents/dashboard/routers/memory.py	Adds DELETE /api/v1/memory/generated-skills endpoint using threadpool execution and structured errors.
dojoagents/config/models.py	Adds LLMProviderConfig.max_tokens; allows explicit null for web tool backends.
dojoagents/config/loader.py	Parses provider max_tokens; preserves explicit null for web tool backends (no silent defaulting).
dojoagents/agent/turn_intent.py	Skips intent classification when no session history is present.
dojoagents/agent/runtime.py	Passes provider max_tokens into OpenAI-compatible provider construction.
dojoagents/agent/providers.py	Adds provider-level max_tokens support and improves reasoning extraction across response variants.
docs/site/zh/reference/dashboard-api.md	Documents dashboard security boundary and new memory deletion API.
docs/site/zh/reference/configuration.md	Documents provider max_tokens and explicit null disabling for web tools; references known limitations.
docs/site/zh/getting-started/model-configuration.md	Adds guidance for custom OpenAI-compatible providers including limits.
docs/site/zh/development/repository-map.md	Updates repository map with links to new “Agent Internals” doc and extension guidance.
docs/site/zh/development/known-limitations.md	Adds bilingual “known limitations” register (ZH).
docs/site/zh/development/index.md	Adds “Known Limitations” link in development index (ZH).
docs/site/zh/development/adding-tools.md	Adds further reading link to “Agent Internals” (ZH).
docs/site/zh/architecture/tools-and-sandbox.md	Adds “Agent Internals” cross-link (ZH).
docs/site/zh/architecture/runtime.md	Adds “Agent Internals” cross-link (ZH).
docs/site/zh/architecture/plugins.md	Adds “Agent Internals” cross-link (ZH).
docs/site/zh/architecture/overview.md	Updates architecture overview to reflect Strands kernel + bridges; adds further reading links (ZH).
docs/site/zh/architecture/multi-agent-planning.md	Adds “Agent Internals” cross-link (ZH).
docs/site/zh/architecture/memory.md	Adds “Agent Internals” cross-link (ZH).
docs/site/zh/architecture/index.md	Adds “Agent Internals” entry (ZH).
docs/site/zh/architecture/event-driven.md	Adds “Agent Internals” cross-link (ZH).
docs/site/zh/architecture/agent-loop.md	Adds pointer to “Agent Internals” for real implementation (ZH).
docs/site/zh/architecture/agent-internals.md	New deep-dive “Agent Internals” documentation (ZH).
docs/site/en/reference/dashboard-api.md	Documents dashboard security boundary and new memory deletion API.
docs/site/en/reference/configuration.md	Documents provider max_tokens and explicit null disabling for web tools; references known limitations.
docs/site/en/getting-started/model-configuration.md	Adds guidance for custom OpenAI-compatible providers including limits.
docs/site/en/development/repository-map.md	Updates repository map with “Agent Internals” links and extension guidance.
docs/site/en/development/known-limitations.md	Adds bilingual “known limitations” register (EN).
docs/site/en/development/index.md	Adds “Known Limitations” link in development index (EN).
docs/site/en/development/adding-tools.md	Adds further reading link to “Agent Internals” (EN).
docs/site/en/architecture/tools-and-sandbox.md	Adds “Agent Internals” cross-link (EN).
docs/site/en/architecture/runtime.md	Adds “Agent Internals” cross-link (EN).
docs/site/en/architecture/plugins.md	Adds “Agent Internals” cross-link (EN).
docs/site/en/architecture/overview.md	Updates architecture overview to reflect Strands kernel + bridges; adds further reading links (EN).
docs/site/en/architecture/multi-agent-planning.md	Adds “Agent Internals” cross-link (EN).
docs/site/en/architecture/memory.md	Adds “Agent Internals” cross-link (EN).
docs/site/en/architecture/index.md	Adds “Agent Internals” entry (EN).
docs/site/en/architecture/event-driven.md	Adds “Agent Internals” cross-link (EN).
docs/site/en/architecture/agent-loop.md	Adds pointer to “Agent Internals” for real implementation (EN).
docs/site/en/architecture/agent-internals.md	New deep-dive “Agent Internals” documentation (EN).

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[kyokagong-alphadojo]

Hi, we are currently in the process of migrating new features to the feat/harness-v2 branch. Please submit your PR against that branch instead. We plan to merge feat/harness-v2 into the main branch in the next 1–2 weeks. Thank you for your contribution!