io.ReadAll errors were silently discarded with `body, _ := io.ReadAll(...)`,
which could cause empty or partial data to be used for JSON unmarshaling
or error messages. This adds proper error checks for all instances.
* feat: add extended thinking support for Anthropic models
Support configurable thinking levels (off/low/medium/high/xhigh/adaptive)
via `agents.defaults.thinking_level` config field.
- "adaptive": uses Anthropic's adaptive thinking API (Claude 4.6+)
- "low/medium/high/xhigh": uses budget_tokens (all thinking-capable models)
- "off": disables thinking (default)
API constraints handled:
- Temperature cleared when thinking is enabled
- budget_tokens clamped to max_tokens-1
- Thinking response blocks parsed into Reasoning field
Relates to #645, #966
* fix: address PR review feedback for thinking support
- Add ThinkingCapable interface for provider capability detection
- Warn when thinking_level is set but provider doesn't support it
- Warn when temperature is cleared due to thinking enabled
- Adjust budget values per Anthropic best practices (medium=16K, xhigh=64K)
- Add budget clamp warning and 80% threshold warning
- Add parseResponse thinking block tests
- Add thinking_level field to config.example.json
* refactor: move ThinkingLevel from AgentDefaults to ModelConfig
Thinking is a model-level capability, not a global agent property.
Per-model config avoids silent ignoring on non-Anthropic providers
and eliminates spurious warning logs in multi-provider setups.
Addresses PR #1076 review feedback from @yinwm.
Add Avian (https://avian.io) as an OpenAI-compatible provider with
API base https://api.avian.io/v1 and AVIAN_API_KEY env var support.
Models: deepseek/deepseek-v3.2, moonshotai/kimi-k2.5, z-ai/glm-5,
minimax/minimax-m2.5. Supports chat completions, streaming, and
function calling.
Changes:
- Add Avian to ProvidersConfig struct, IsEmpty(), HasProvidersConfig()
- Add avian protocol to factory provider and default API base
- Add avian case to legacy provider selection (factory.go)
- Add avian migration rule for old config format
- Add default model entries to ModelList (deepseek-v3.2, kimi-k2.5)
- Add avian to example config
- Update AllProviders test count from 18 to 19
- Separate third-party imports from local module imports (gci)
- Fix byte slice literal formatting (gofumpt)
- Rename shadowed err variable to ftErr (govet)
- Remove trailing blank lines in test files
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- serializeMessages: preserve ToolCallID/ToolCalls when Media is present
- resolveMediaRefs: add 20MB file size limit to prevent OOM
- mimeFromExtension: return empty string for unknown extensions
- Add 11 unit tests for serializeMessages, resolveMediaRefs, mimeFromExtension
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
The serializeMessages() function was not preserving the reasoning_content
field when serializing messages for vision API calls. This caused the
TestProviderChat_PreservesReasoningContentInHistory test to fail.
This fix ensures reasoning_content is included in both text-only messages
and vision messages with media attachments.
Co-authored-by: Zachary Guerrero <zack.grrr@gmail.com>
- Add Media []string field to Message struct for image/media URLs
- Implement serializeMessages() to format messages with image_url content parts
- Enables OpenAI-compatible vision APIs to receive image attachments
* fix(pkg/providers):do regex precompile insteadd on the fly
* fix(providers): replace HTTP-specific regex with standalone status code matcher
The precompiled HTTP regex used uppercase "HTTP" which never matched
because ClassifyError lowercases the input. Replace it with a
case-insensitive word-boundary pattern that matches any standalone
3-digit status code (300-599), which also subsumes the HTTP/x.x case.
Add test case for standalone status code extraction.
* fix(providers): restore http regex and add standalone status code matcher
Restore the http-prefixed regex (without unnecessary (?i) flag since
input is already lowercased by ClassifyError) as a mid-priority pattern
to reduce false positives. Add a standalone word-boundary matcher as a
fallback for bare status codes like "429". Fix test to use lowercased
input matching the actual calling convention.
* perf(tools): move path regex compilation from per-call to package init
The path regex in guardCommand was compiled on every call. Hoist it
to a package-level var (absolutePathPattern) alongside defaultDenyPatterns
in a single var block, so it is compiled once at init time.
* style(tools): move inline comment to fix golines formatting error
- Fix ignored error from SendAndWait call
- Improve error message for unimplemented stdio mode with helpful guidance
- Add TODO comment with reference link for future stdio implementation
The openaiMessage struct and stripSystemParts() were not carrying over
the ReasoningContent field when serializing conversation history for
API requests. This caused thinking models (e.g. kimi-k2.5) to receive
incomplete assistant messages on subsequent turns, resulting in 400
errors from the Moonshot API.
Add the ReasoningContent field to openaiMessage and copy it in
stripSystemParts(). Also add a test to verify reasoning_content is
preserved when sending conversation history.
Fixes#588
Related: #876
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Whitespace is a linter that checks for unnecessary newlines at the start and end of functions, if, for, etc.
Signed-off-by: Kai Xia <kaix+github@fastmail.com>
Avoid rebuilding the entire system prompt on every BuildMessages() call
by caching the static portion (identity, bootstrap, skills summary,
memory) and only recomputing it when workspace source files change.
Key changes:
- ContextBuilder caches the static prompt behind an RWMutex with
double-checked locking. Source file changes are detected via cheap
os.Stat mtime checks so no explicit invalidation is needed.
- Track file existence at cache time (existedAtCache map) so that
newly created or deleted bootstrap/memory files also trigger a
rebuild — the old modifiedSince() silently returned false on
os.IsNotExist.
- Walk the skills directory recursively with filepath.WalkDir to
catch content-only edits at any nesting depth; directory mtime
alone misses in-place file modifications on most filesystems.
- ToolRegistry.sortedToolNames() sorts tool names before iteration,
ensuring deterministic tool definition order across calls — a
prerequisite for LLM-side prefix/KV cache reuse.
- Merge all context (static + dynamic + summary) into a single
system message for provider compatibility: the Anthropic adapter
extracts messages[0] as the top-level system parameter, and Codex
reads only the first system message as instructions.
- Fix a data race in BuildMessages() where cachedSystemPrompt was
read without holding the lock in a debug log statement.
- Add tests: single system message invariant, mtime auto-invalidation,
new-file creation detection, skill file content change, explicit
InvalidateCache, cache stability, concurrent access (20 goroutines
x 50 iterations, passes go test -race), and a benchmark.
* cleanup dead code.
Signed-off-by: Kai Xia <kaix+github@fastmail.com>
* add these two back with flag.
Signed-off-by: Kai Xia <kaix+github@fastmail.com>
* fix ci
Signed-off-by: Kai Xia <kaix+github@fastmail.com>
* remove this confusing line
Signed-off-by: Kai Xia <kaix+github@fastmail.com>
* make fmt
Signed-off-by: Kai Xia <kaix+github@fastmail.com>
* remove unused method.
picked up by golangci-lint run
Signed-off-by: Kai Xia <kaix+github@fastmail.com>
---------
Signed-off-by: Kai Xia <kaix+github@fastmail.com>
The configuration field for specifying the model has been renamed from
"model" to "model_name" for better clarity and consistency with the
model_list configuration.
A GetModelName() accessor method has been added to maintain backward
compatibility. Existing configurations using the old "model" field will
continue to work correctly.
This change affects:
- Configuration structure (AgentDefaults struct)
- All references across the codebase
- Documentation in all language variants
- Example configuration files
Mistral's API strictly validates tool_calls in assistant messages and
rejects non-standard fields. The ToolCall struct had Name and Arguments
as top-level JSON fields, duplicating data already in Function.Name
and Function.Arguments. OpenAI silently ignored these extras but
Mistral returns 422.
Change json tags to "-" so these internal fields are no longer
serialized to API payloads while remaining available in Go code.
Add Mistral as a first-class provider alongside the 17 existing ones.
Mistral uses the OpenAI-compatible API at https://api.mistral.ai/v1
with provider-specific model prefix stripping (mistral/model → model).
Changes:
- Add Mistral to ProvidersConfig, IsEmpty(), HasProvidersConfig()
- Add mistral entry in default model_list (defaults.go)
- Add mistral protocol in factory_provider.go and getDefaultAPIBase()
- Add mistral prefix stripping in openai_compat normalizeModel()
- Add mistral case in legacy factory.go resolveProviderSelection()
- Add mistral migration entry in ConvertProvidersToModelList()
- Add mistral to supported providers in migrate/config.go
- Add mistral section in config.example.json
- Update AllProviders test (17 → 18 providers)
Tested end-to-end with mistral-small-latest model.
Models like Moonshot kimi-k2.5 and DeepSeek-R1 return a
reasoning_content field in assistant messages. When thinking is enabled,
the API requires this field to be echoed back in subsequent requests.
PicoClaw was silently dropping it, causing 400 errors on tool-call
round-trips.
- Add ReasoningContent to Message and LLMResponse types
- Parse reasoning_content in openai_compat parseResponse()
- Carry reasoning_content through assistant tool-call messages
- Add unit test for reasoning_content parsing
Fixes#588
Replace all claude-sonnet-4 references with claude-sonnet-4.6 across
codebase including documentation, tests, and configuration examples.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Remove sync.RWMutex and rrCounters from Config struct
- Simplify GetModelConfig to use global atomic counter for load balancing
- Remove unnecessary locks from HasProvidersConfig, SaveConfig, etc.
- Add buildModelWithProtocol helper to handle models with existing prefix
- Fix TestCreateProviderReturnsHTTPProviderForOpenRouter to use model_list
- Upgrade all Claude 3 references to Claude 4 across documentation
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Auth fixes:
- Fix OpenAI/Anthropic OAuth and token login to update ModelList
- Fix logout to clear AuthMethod in ModelList
- Add helper functions: isOpenAIModel, isAnthropicModel, isAntigravityModel
- Fix slice bounds panic in isAntigravityModel using strings.HasPrefix
- All auth operations now preserve existing model_list configuration
Factory provider fixes:
- Add OAuth support for openai protocol in CreateProviderFromConfig
- CodexAuthProvider is now used when auth_method is oauth/token
Default model updates:
- OpenAI login: set default model to gpt-5.2
- Anthropic login: set default model to claude-sonnet-4
- Antigravity login: set default model to gemini-flash (remove provider field)
Model changes:
- Change default OpenAI model from gpt-4o to gpt-5.2
- gpt-5.2 is compatible with Codex API (chatgpt.com backend)
- Update all README files, config examples, and migration code
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add support for persisting thought_signature metadata from Google/Gemini 3
models. This introduces ExtraContent and GoogleExtra types to handle
provider-specific metadata, and ensures thought signatures are properly
preserved through the tool call lifecycle.
- Move provider creation logic to factory_provider.go with protocol-based approach
- Add OpenAIProviderConfig with WebSearch support and embedded ProviderConfig
- Add maxTokensField to OpenAI-compatible provider for configurable token field
- Introduce new providers: Ollama, DeepSeek, GitHubCopilot, Antigravity, Qwen
- Remove redundant CreateProvider function from factory.go
- Add ThoughtSignature field to FunctionCall for tool response handling
- Remove duplicate Name field assignment in tool loop
- Update tests to reflect new provider configuration structure
- Preserve user's configured model during config migration (issue #5)
- Simplify ExtractProtocol using strings.Cut
- Extract NormalizeToolCall to shared utility, removing ~70 lines of duplicate code
- Clean up unused fields in providerMigrationConfig struct
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
1. Add VLLM default API base (http://localhost:8000/v1)
- Previously returned empty string, causing provider creation to fail
2. Implement MaxTokensField configuration
- Add maxTokensField field to HTTPProvider
- Add NewHTTPProviderWithMaxTokensField constructor
- Use configured field name for max_tokens parameter
- Fallback to model-based detection for backward compatibility
3. Add tests for VLLM, deepseek, ollama default API bases
Example config usage:
{
"model_name": "glm-4",
"model": "openai/glm-4",
"max_tokens_field": "max_completion_tokens"
}
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
- Move OAuth helper functions to factory_provider.go
- Add auto-migration in LoadConfig: old providers -> model_list
- Add Workspace field to ModelConfig for CLI-based providers
- Fix OAuth handling to use auth store instead of raw APIKey
- Update tests to use new model_list configuration format
This eliminates the giant switch-case in legacy_provider.go,
achieving the goal of "zero-code provider addition" from the
design document (issue #283).
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Refactor command handlers into separate files to improve code organization
and maintainability. Each command (agent, auth, cron, gateway, migrate,
onboard, skills, status) now has its own dedicated file.
Restructure provider creation to support new model_list configuration
system that enables zero-code addition of OpenAI-compatible providers.
Move legacy provider logic to separate file for backward compatibility.
Move configuration functions from config.go to separate files
(defaults.go, migration.go) for better organization.
Resolve conflicts in pkg/providers/types.go and pkg/agent/loop.go:
- types.go: use protocoltypes aliases from PR #213, keep fallback types
- loop.go: drop old single-agent createToolRegistry (replaced by multi-agent pattern)
Refactor to align with PR #213 patterns:
- instance.go: use NewExecToolWithConfig (accept full config for deny patterns)
- registry.go: pass full config to NewAgentInstance
- loop.go: add Perplexity web search options to registerSharedTools
Merge upstream/main into refactor/provider-protocol-122.
Resolve http_provider.go conflict (keep thin delegate).
Wire OpenAIProviderConfig.WebSearch through providerSelection
and into CodexProvider for codex-auth and codex-cli-token paths.
Phase 1: centralize protocol message/tool/response types in protocoltypes and keep compatibility aliases in providers and protocol packages.
Phase 1: preserve HTTPProvider constructor compatibility and route Anthropic api_base through factory auth/provider constructors with base URL normalization.
Phase 2: expand provider routing/auth tests (deepseek/nvidia/shengsuanyun, codex/claude oauth/codex-cli) and add openai_compat + anthropic coverage for proxy transport, model normalization, numeric option coercion, token-source refresh, and base URL behavior.
Phase 3: apply gofmt and validate with Dockerized tests (go test ./pkg/providers/... ./pkg/migrate and go test ./...).
Resolve conflicts:
- pkg/agent/loop.go: integrate context compression, command handling,
utf8 token estimation, and summarization notification into
multi-agent routing architecture
- pkg/config/config_test.go: merge imports from both branches
- pkg/agent/loop_test.go: update test to use registry-based sessions
* feat: add Codex CLI provider for OpenAI subprocess integration
Add CodexCliProvider that wraps `codex exec --json` as a subprocess,
analogous to the existing ClaudeCliProvider pattern. This enables using
OpenAI's Codex CLI tool as a local LLM backend.
- CodexCliProvider: subprocess wrapper parsing JSONL event stream
- Credential reader for ~/.codex/auth.json with token expiry detection
- Factory integration: provider "codex-cli" and auth_method "codex-cli"
- Fix tilde expansion in workspace path for CLI providers
- 37 unit tests covering parsing, prompt building, credentials, and mocks
* fix: add tool call extraction to Codex CLI provider
- Extract shared tool call parsing into tool_call_extract.go
(extractToolCallsFromText, stripToolCallsFromText, findMatchingBrace)
- Both ClaudeCliProvider and CodexCliProvider now share the same
tool call extraction logic for PicoClaw-specific tools
- Fix cache token accounting: include cached_input_tokens in total
- Add 2 new tests for tool call extraction from JSONL events
- Update existing tests for corrected token calculations
* fix(docker): update Go version to match go.mod requirement
Dockerfile used golang:1.24-alpine but go.mod requires go >= 1.25.7.
This caused Docker builds to fail on all branches with:
"go: go.mod requires go >= 1.25.7 (running go 1.24.13)"
Update to golang:1.25-alpine to match the project requirement.
* fix: handle codex CLI stderr noise without losing valid stdout
Codex writes diagnostic messages to stderr (e.g. rollout errors) which
cause non-zero exit codes even when valid JSONL output exists on stdout.
Parse stdout first before checking exit code to avoid false errors.
* style: fix gofmt formatting and update web search API in tests
- Remove trailing whitespace in web.go and base_test.go
- Update config_test.go and web_test.go for WebSearchToolOptions API
Fixes four issues identified in the community code review:
- Session persistence broken on Windows: session keys like
"telegram:123456" contain ':', which is illegal in Windows
filenames. filepath.Base() strips drive-letter prefixes on Windows,
causing Save() to silently fail. Added sanitizeFilename() to
replace invalid chars in the filename while keeping the original
key in the JSON payload.
- HTTP client with no timeout: HTTPProvider used Timeout: 0 (infinite
wait), which can hang the entire agent if an API endpoint becomes
unresponsive. Set a 120s safety timeout.
- Slack AllowFrom type mismatch: SlackConfig used plain []string
while every other channel uses FlexibleStringSlice, so numeric
user IDs in Slack config would fail to parse.
- Token estimation wrong for CJK: estimateTokens() divided byte
length by 4, but CJK characters are 3 bytes each, causing ~3x
overestimation and premature summarization. Switched to
utf8.RuneCountInString() / 3 for better cross-language accuracy.
Also added unit tests for the session filename sanitization.
Ref #116
Update registerSharedTools to use new WebSearchToolOptions API and
add hardware tools (I2C, SPI) from upstream. Accept upstream's
new web tools config test.
Resolve conflicts in loop.go, config.go, config_test.go,
spawn.go, and subagent.go. Integrate upstream ToolResult/AsyncTool
pattern with multi-agent routing features. Rename mockProvider
to mockRegistryProvider in registry_test.go to avoid redeclaration
with upstream's loop_test.go.
- Add ChannelSender interface for sending heartbeat results to users
- Add sendResponse() to automatically deliver results to last channel
- Add createDefaultHeartbeatTemplate() for first-run setup
- Support HEARTBEAT_OK silent response (legacy compatibility)
- Add structured logging with INFO/ERROR levels
- Move integration tests to separate file with build tag
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add ClaudeCliProvider that executes the local CLI as a subprocess,
enabling PicoClaw to leverage advanced capabilities (MCP tools,
workspace awareness, session management) through any messaging channel.
- Implement LLMProvider interface via subprocess execution
- Support --system-prompt, --model, --output-format json flags
- Parse real v2.x JSON response format including usage tokens
- Handle error responses, stderr, context cancellation
- Register "claude-cli", "claude-code", "claudecode" aliases in CreateProvider
- 56 unit tests with mock scripts + 3 integration tests against real binary
- 100% coverage on all functions except stripToolCallsJSON (85.7%)
- Add Provider field to AgentDefaults struct
- Modify CreateProvider to use explicit provider field first, fallback to model name detection
- Allows using models without provider prefix (e.g., llama-3.1-8b-instant instead of groq/llama-3.1-8b-instant)
- Supports all providers: groq, openai, anthropic, openrouter, zhipu, gemini, vllm
- Backward compatible with existing configs
Fixes issue where models without provider prefix could not use configured API keys.
Update to latest major version of the official OpenAI Go SDK.
Fix breaking change: FunctionCallOutput.Output is now a union type
(ResponseInputItemFunctionCallOutputOutputUnionParam) instead of string.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Add ClaudeProvider (anthropic-sdk-go) and CodexProvider (openai-go) that
use the correct subscription endpoints and API formats:
- CodexProvider: chatgpt.com/backend-api/codex/responses (Responses API)
with OAuth Bearer auth and Chatgpt-Account-Id header
- ClaudeProvider: api.anthropic.com/v1/messages (Messages API) with
Authorization: Bearer token auth
Update CreateProvider() routing to use new SDK-based providers when
auth_method is "oauth" or "token", removing the stopgap that sent
subscription tokens to pay-per-token endpoints.
Closes#18
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
