* feat: Add OpenRouter chat completion provider adapter with custom headers. (#6436)
* chore: update astrbot.service configuration
* fix(core): use original version constraints instead of locking to installed version
Fixes#6420
The core constraints mechanism was using the currently installed version as
an exact constraint (e.g., `aiosqlite==0.21.0`), preventing plugins from
installing higher versions even when they satisfy the original constraint.
Changes:
- Preserve original version specifier from pyproject.toml (e.g., `>=0.21.0`)
- Allow plugins to require higher versions as long as they satisfy core constraint
- Prevent downgrade by using `>=installed` for packages without version constraint
Example:
- Before: Core constraint `aiosqlite==0.21.0`, plugin requires `>=0.22.1` → BLOCKED
- After: Core constraint `aiosqlite>=0.21.0`, plugin requires `>=0.22.1` → ALLOWED
This enables better dependency management while still protecting core dependencies
from incompatible downgrades.
---------
Co-authored-by: Futureppo <luominzhi2005@qq.com>
Co-authored-by: LIghtJUNction <lightjunction.me@gmail.com>
Co-authored-by: ccsang <ccsang@users.noreply.github.com>
- Introduce 'astrbot bk' command with GPG signing, encryption, and digest support
- Add import/export functionality using core backup modules
- Refactor path management to use 'AstrbotPaths' singleton across CLI commands
- Replace blocking subprocess calls with asyncio.create_subprocess_exec in backup command
- Add comprehensive tests for uninstall and backup commands
- Improve module resource handling for bundled dashboard assets
* fix: only pass dimensions param when explicitly configured
Models like bge-m3 don't support the dimensions parameter in the
embedding API, causing HTTP 400 errors. Previously dimensions was
always sent with a default value of 1024, even when the user never
configured it. Now dimensions is only included in the request when
embedding_dimensions is explicitly set in provider config.
Closes#6421
Signed-off-by: JiangNan <1394485448@qq.com>
* fix: handle invalid dimensions config and align get_dim return
- Add try-except around int() conversion in _embedding_kwargs to
gracefully handle invalid embedding_dimensions config values
- Update get_dim() to return 0 when embedding_dimensions is not
explicitly configured, so callers know dimensions weren't specified
and can handle it accordingly
- Both methods now share consistent logic for reading the config
Signed-off-by: JiangNan <1394485448@qq.com>
* fix: improve logging for invalid embedding_dimensions configuration
---------
Signed-off-by: JiangNan <1394485448@qq.com>
Co-authored-by: Soulter <905617992@qq.com>
* fix: 修改 register_agent 以避免运行时导入 AstrAgentContext
* test: improve register_agent test robustness
- Add fixture for llm_tools cleanup to avoid test interference
- Use multiple import patterns to make guard more robust to refactors
- Add assertion to verify decorated coroutine is wired as handoff handler
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* 删除测试文件: 移除 register_agent 装饰器的运行时行为测试
---------
Co-authored-by: whatevertogo <whatevertogo@users.noreply.github.com>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: Soulter <905617992@qq.com>
* fix: resolve subagent persona lookup for 'default' and unify resolution logic
- Add PersonaManager.get_persona_v3_by_id() to centralize v3 persona resolution
- Handle 'default' persona_id mapping to DEFAULT_PERSONALITY in subagent orchestrator
- Fix HandoffTool.default_description using agent_name parameter correctly
- Add tests for default persona in subagent config and tool deduplication
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
* refactor: simplify get_default_persona_v3 using get_persona_v3_by_id
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---------
Co-authored-by: whatevertogo <whatevertogo@users.noreply.github.com>
Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>
Co-authored-by: Soulter <37870767+Soulter@users.noreply.github.com>
There is an issue with reading the skill directory on the Windows system, which results in a high probability of files under the skill directory being unrecognizable, now fix it.
The async engine is created without a busy timeout, so concurrent
writes (agent responses, metrics, session updates) fail instantly
with 'database is locked' instead of waiting for the lock.
Add connect_args={'timeout': 30} for SQLite engines so the driver
waits up to 30 seconds for the write lock. Combined with the existing
WAL journal mode, this handles the typical concurrent write bursts
from agent + metrics + session operations.
Fixes#6443
- Implement 'astrbot uninstall' to remove systemd service and data files
- Add '--log-level' option to 'astrbot run' (default: INFO)
- Pass log level config to core logger via env var
Fixes#6283
When adding a new embedding provider, the knowledge base creation page
did not show the new provider until restart.
Root cause: create_provider() did not update self.providers_config,
which is used by get_provider_config_list() to return provider lists.
This fix syncs the in-memory config after loading the new provider,
consistent with how reload() handles config updates.
Co-authored-by: ccsang <ccsang@users.noreply.github.com>
Address Sourcery AI review feedback: the image-handling logic was
duplicated for ImageContent and EmbeddedResource cases.
Changes:
- Extract _handle_image_content() helper function
- Consolidate image caching, result appending, and yielding logic
- Reduce code duplication and improve maintainability
Fixes#6140
When a tool returns CallToolResult with multiple content items (e.g.,
both TextContent and ImageContent), the agent was only processing
content[0], ignoring the rest.
Changes:
- Replace direct content[0] access with enumerate(res.content) loop
- Process all content items: TextContent, ImageContent, EmbeddedResource
- Use content_index for image caching to distinguish multiple images
This fixes the issue where tools like Bilibili plugin return both
text descriptions and screenshots, but LLM only received one of them.
* Fix CreateSkillPayloadTool array schema missing items field
The payload parameter's anyOf array variant lacked an items field,
causing Gemini API to reject the tool declaration with 400 Bad Request:
'parameters.properties[payload].any_of[1].items: missing field.'
Add items: {type: object} to the array variant to satisfy the Gemini
API requirement for array type schemas.
Fixes#6279
* Fix TypeError when OpenAI-compatible API returns null choices
Some providers (e.g. OpenRouter) may return a completion where
choices is None rather than an empty list — for instance on rate
limiting, content filtering, or transient errors. The existing code
used len(completion.choices) which throws TypeError on None.
Replace all len(...choices) == 0 checks with 'not ... .choices' which
handles both None and empty list. Affects _query_stream, _parse_openai_completion,
and _extract_reasoning_content.
Fixes#6252
Fixes#6294
QQ official bot receives emoji/sticker messages as raw XML-like tags:
`<faceType=4,faceId="",ext="eyJ0ZXh0IjoiW+a7oeWktOmXruWPt10ifQ==">`
This made the LLM unable to understand the emoji content.
Changes:
- Added `_parse_face_message()` method to parse face message format
- Decode base64 `ext` field to get emoji description text
- Replace face tags with `[表情:描述]` format for readability
Example:
- Input: `<faceType=4,faceId="",ext="eyJ0ZXh0IjoiW+a7oeWktOmXruWPt10ifQ==">`
- Output: `[表情:[满头问号]]`
Co-authored-by: ccsang <ccsang@users.noreply.github.com>
- Add local file caching for dashboard downloads with version validation
- Implement fallback to 'latest' version if specific version download fails
- Add robust error handling in CLI check_dashboard to prevent crashes
- Remove dashboard caching from smoke tests (backend-only mode)
qingyun
LIghtJUNction
shuiping233
jnMetaCode
whatevertogo
linzhengtian
Rhonin Wang
Yufeng He
Soulter
Futureppo
Chen
Chen
xkeyC
lppsuixn
Rin
rin
stevessr
ccsang
Stable Genius
時壹
Rin
洛薇Lovie
xwsjjctz
Trainingcqy
邹永赫