teddy-cli 0.1.17.dev1360__tar.gz → 0.1.17.dev1361__tar.gz

{teddy_cli-0.1.17.dev1360 → teddy_cli-0.1.17.dev1361}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: teddy-cli
-Version: 0.1.17.dev1360
+Version: 0.1.17.dev1361
 Summary: The local-first, test-driven agentic engineering harness. Because you should own the means of development.
 Author: Raphael Atteritano
 License: AGPL-3.0-only
@@ -208,10 +208,6 @@ Use subcommands to overwrite specific defaults:
 
 The first execution of `teddy start` runs an interactive setup to configure your API key and preferred editor. For deeper customization, edit `.teddy/config.yaml`. All `llm.*` settings pass directly through to [LiteLLM](https://docs.litellm.ai/docs/completion/input), allowing you to route to cloud providers or local endpoints (Ollama, llama.cpp, or any OpenAI-compatible server via `api_base`). The config file also lets you adjust your external editor, command execution timeouts, diff similarity thresholds, and automated context pruning budgets.
 
-#### OpenRouter App Attribution
-
-The bundled config attributes OpenRouter usage to TeDDy through the documented `HTTP-Referer` and `X-OpenRouter-Title` headers, so the usage appears on [TeDDy's app page](https://openrouter.ai/apps?url=https://github.com/atte500/teddy). Attribution is provider-side visibility only: the headers travel to the endpoint you already configured and the TeDDy project receives nothing. The block ships in the bundled baseline, so set `llm.extra_headers: null` in your own `.teddy/config.yaml` to opt out. LiteLLM then falls back to its own defaults, which credit `liteLLM` instead. Other providers ignore these headers.
-
 ### Environment Variables
 
 TeDDy checks the following process environment variables at startup and during a session.
@@ -324,7 +320,9 @@ TeDDy packages your workspace context into a single clipboard payload, and the a
 
 <br />
 
-No. TeDDy collects zero telemetry, zero analytics, and runs no background tracking. All requests go directly to whichever LLM provider or local endpoint you configure in your local config file. The bundled config sends OpenRouter's published app-attribution headers, so OpenRouter credits your usage to TeDDy on its public leaderboard, and that visibility reaches only the provider you chose. Set `llm.extra_headers: null` in `.teddy/config.yaml` to opt out, which makes LiteLLM fall back to its own defaults.
+No. TeDDy collects zero telemetry, zero analytics, and runs no background tracking. All requests go directly to whichever LLM provider or local endpoint you configure in your local config file.
+
+By default, OpenRouter requests include attribution headers (`HTTP-Referer` and `X-OpenRouter-Title`) to credit usage to TeDDy on OpenRouter rankings. Set `llm.extra_headers: null` in `.teddy/config.yaml` to opt out. Other providers ignore these headers.
 
 </details>
 
@@ -372,9 +370,9 @@ The tool itself is not monetized. If your team needs custom agent architectures,
 
 ## Learn More
 
-- [Project Roadmap & Vision](/docs/project/PROJECT.md)
-- [System Architecture](/docs/architecture/ARCHITECTURE.md)
-- [Agent Prompt Templates](/src/teddy_executor/resources/config/prompts/)
+- [Project Roadmap & Vision](./docs/project/PROJECT.md)
+- [System Architecture](./docs/architecture/ARCHITECTURE.md)
+- [Agent Prompt Templates](./src/teddy_executor/resources/config/prompts/)
 
 ---
 

{teddy_cli-0.1.17.dev1360 → teddy_cli-0.1.17.dev1361}/README.md

@@ -174,10 +174,6 @@ Use subcommands to overwrite specific defaults:
 
 The first execution of `teddy start` runs an interactive setup to configure your API key and preferred editor. For deeper customization, edit `.teddy/config.yaml`. All `llm.*` settings pass directly through to [LiteLLM](https://docs.litellm.ai/docs/completion/input), allowing you to route to cloud providers or local endpoints (Ollama, llama.cpp, or any OpenAI-compatible server via `api_base`). The config file also lets you adjust your external editor, command execution timeouts, diff similarity thresholds, and automated context pruning budgets.
 
-#### OpenRouter App Attribution
-
-The bundled config attributes OpenRouter usage to TeDDy through the documented `HTTP-Referer` and `X-OpenRouter-Title` headers, so the usage appears on [TeDDy's app page](https://openrouter.ai/apps?url=https://github.com/atte500/teddy). Attribution is provider-side visibility only: the headers travel to the endpoint you already configured and the TeDDy project receives nothing. The block ships in the bundled baseline, so set `llm.extra_headers: null` in your own `.teddy/config.yaml` to opt out. LiteLLM then falls back to its own defaults, which credit `liteLLM` instead. Other providers ignore these headers.
-
 ### Environment Variables
 
 TeDDy checks the following process environment variables at startup and during a session.
@@ -290,7 +286,9 @@ TeDDy packages your workspace context into a single clipboard payload, and the a
 
 <br />
 
-No. TeDDy collects zero telemetry, zero analytics, and runs no background tracking. All requests go directly to whichever LLM provider or local endpoint you configure in your local config file. The bundled config sends OpenRouter's published app-attribution headers, so OpenRouter credits your usage to TeDDy on its public leaderboard, and that visibility reaches only the provider you chose. Set `llm.extra_headers: null` in `.teddy/config.yaml` to opt out, which makes LiteLLM fall back to its own defaults.
+No. TeDDy collects zero telemetry, zero analytics, and runs no background tracking. All requests go directly to whichever LLM provider or local endpoint you configure in your local config file.
+
+By default, OpenRouter requests include attribution headers (`HTTP-Referer` and `X-OpenRouter-Title`) to credit usage to TeDDy on OpenRouter rankings. Set `llm.extra_headers: null` in `.teddy/config.yaml` to opt out. Other providers ignore these headers.
 
 </details>
 
@@ -338,9 +336,9 @@ The tool itself is not monetized. If your team needs custom agent architectures,
 
 ## Learn More
 
-- [Project Roadmap & Vision](/docs/project/PROJECT.md)
-- [System Architecture](/docs/architecture/ARCHITECTURE.md)
-- [Agent Prompt Templates](/src/teddy_executor/resources/config/prompts/)
+- [Project Roadmap & Vision](./docs/project/PROJECT.md)
+- [System Architecture](./docs/architecture/ARCHITECTURE.md)
+- [Agent Prompt Templates](./src/teddy_executor/resources/config/prompts/)
 
 ---
 

{teddy_cli-0.1.17.dev1360 → teddy_cli-0.1.17.dev1361}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "teddy-cli"
-version = "0.1.17.dev1360"
+version = "0.1.17.dev1361"
 description = "The local-first, test-driven agentic engineering harness. Because you should own the means of development."
 readme = "README.md"
 keywords = [

teddy_cli-0.1.17.dev1361/src/teddy_executor/resources/config/config.yaml

@@ -0,0 +1,56 @@
+# TeDDy Configuration
+
+editor: ""
+
+execution:
+  default_timeout_seconds: 60       # EXECUTE timeout.
+  similarity_threshold: 0.95        # EDIT match threshold (1.0 = exact match; ignores relative indentation).
+  max_output_lines: 100             # Caps EXECUTE stdout/stderr output lines.
+
+read:
+  max_lines: 1000                   # Caps READ action output lines.
+
+research:
+  max_results: 5                    # Number of search results to fetch and scrape.
+
+checks:
+  prompts: true                     # Warn when local .teddy/prompts/ differ from bundled defaults.
+  templates: true                   # Warn when local docs/templates/ differ from bundled defaults.
+  update: true                      # Warn when a newer TeDDy release is available on PyPI.
+
+yolo_default: false                 # When true, runs non-interactively by default; overridden per-command by -n / --no-yolo.
+
+guardrails:                         # Non-interactive limits (--yolo / --pipeline). Counters reset on 'teddy resume' and on message turns if message_reset: true.
+  enabled: true
+  max_turns: 150
+  max_session_cost: 5.00
+  message_reset: true
+
+auto_pruning:
+  enabled: true
+  turn_context_threshold: 50000     # Token budget for turn-scoped files only (excludes session.context & system prompts).
+  prune_failure_history: true      # Prunes non-green turns once recovery succeeds.
+  prune_validation_failures: true   # Prunes malformed turns that failed validation.
+  preserve_message_turns: true      # Preserves user-communicating turns from pruning.
+  max_turns_retention: 10           # Max historical turns retained in turn context.
+
+# LiteLLM reference: https://docs.litellm.ai/docs/completion/input (all keys under llm pass directly to litellm).
+llm:
+  model: "openrouter/~deepseek/deepseek-flash-latest:nitro"  # OpenRouter syntax: "openrouter/<slug>" (:nitro=fast, :floor=cheap).
+  api_key: "${TEDDY_LLM_API_KEY}"                           # Resolves from .teddy/.env or shell environment.
+  max_retries: 3
+  timeout: 300
+  extra_headers:
+    HTTP-Referer: "https://github.com/atte500/teddy"
+    X-Title: "TeDDy"
+    X-OpenRouter-Title: "TeDDy"
+    X-OpenRouter-Categories: "cli-agent"
+
+# Local models (Ollama, LM Studio):
+#   model: "ollama/qwen3-coder:30b"
+#   model: "lm_studio/qwen3-coder-30b"
+
+# OpenAI-compatible server (vLLM, llama.cpp):
+#   model: "openai/qwen3-coder-30b"
+#   api_base: "http://localhost:8080/v1"
+#   api_key: "dummy-key-custom-endpoints"

teddy_cli-0.1.17.dev1360/src/teddy_executor/resources/config/config.yaml

@@ -1,96 +0,0 @@
-# TeDDy Configuration
-
-# The preferred external editor for reviewing and modifying plans/messages.
-# Supports arbitrary command strings (e.g., "code --wait", "nvim -R", "zed").
-# Fallback chain: Config -> VISUAL/EDITOR env vars -> discovery prompt.
-editor: ""
-
-# Optional diff viewer flags override. If set, these flags are used instead
-# of the built-in translation table for this editor.
-# Example: ["--diff", "--wait"]
-# diff_flags: []
-
-# Execution Settings
-execution:
-  default_timeout_seconds: 60
-  similarity_threshold: 0.95 # 1.00 means exact match required (ignoring relative indentation).
-  max_output_lines: 100 # Caps EXECUTE output to the last X lines.
-
-# File Read Settings
-read:
-  max_lines: 1000 # Caps READ action output to the first X lines.
-
-# Research Settings
-research:
-  max_results: 5 # Number of SERP results to retrieve and automatically scrape.
-
-# Notification Checks
-# Independent toggles for user-facing notifications. All three default to
-# enabled (true) when absent, so existing user configs keep working.
-checks:
-  prompts: true    # Notify when .teddy/prompts/ files differ from the bundled defaults.
-  templates: true  # Notify when docs/templates/ files differ from the bundled defaults.
-  update: true     # Notify when a newer TeDDy version is available (background fetch still runs).
-
-# Default for the --yolo / -y flag. When true, sessions auto-approve all
-# actions (non-interactive) without needing to pass --yolo. The explicit
-# --no-yolo / -n flag always overrides this setting for a single invocation.
-yolo_default: false
-
-# Safety limits enforced in non-interactive runs (--yolo and --pipeline).
-# Limits are per-run: `teddy resume` starts counting from 0 again.
-guardrails:
-  enabled: true
-  max_turns: 150
-  max_session_cost: 5.00
-  message_reset: true
-
-# Auto-Pruning Settings
-# Intelligently pre-deselects context files to save tokens and manage complexity.
-auto_pruning:
-  enabled: true
-  turn_context_threshold: 50000 # Token budget for Turn-scope files only (excludes session.context and system prompts). Triggers pruning of largest Turn-scope files if exceeded.
-  prune_failure_history: true # Prune non-green turns once recovery is achieved.
-  prune_validation_failures: true # Prune plans and reports that failed validation.
-  preserve_message_turns: true # Ensure successful communicating turns are not pruned.
-  max_turns_retention: 10 # Max number of turns persisted in session history (excl. initial request & preserved message turns).
-
-# LLM Settings
-# LiteLLM Configuration Reference: https://docs.litellm.ai/docs/completion/input
-# All keys under 'llm' are passed directly to litellm.completion().
-llm:
-  model: "openrouter/~deepseek/deepseek-flash-latest:nitro"  # Syntax: "openrouter/<model-path>" (model paths: https://openrouter.ai/models); ":nitro"=fastest, ":floor"=cheapest.
-  api_key: "${TEDDY_LLM_API_KEY}"  # Paste your key into .teddy/.env, or replace this with a literal value to override.
-  max_retries: 3
-  timeout: 300
-  # OpenRouter app attribution (https://openrouter.ai/docs/app-attribution).
-  # Usage from this app appears at https://openrouter.ai/apps?url=https://github.com/atte500/teddy.
-  # LiteLLM also sends a legacy X-Title default, so both title headers are pinned here.
-  # This block ships in the baseline. Set `llm.extra_headers: null` in .teddy/config.yaml to opt out.
-  # Non-OpenRouter providers ignore these headers.
-  extra_headers:
-    HTTP-Referer: "https://github.com/atte500/teddy"
-    X-Title: "TeDDy"
-    X-OpenRouter-Title: "TeDDy"
-    X-OpenRouter-Categories: "cli-agent"
-
-# Local models (no API key needed — LiteLLM supplies a keyless default):
-#   model: "ollama/llama3.2"      # Ollama (default endpoint http://localhost:11434)
-#   model: "lm_studio/qwen2.5"    # LM Studio
-#   model: "hosted_vllm/mistral"  # vLLM (set api_base to your server, e.g. http://localhost:8000/v1)
-
-# Any OpenAI-compatible server (e.g. llama.cpp's llama-server) works via the
-# openai/ prefix with an explicit endpoint and a non-empty api_key (a dummy
-# value is fine). A provider prefix is required — an endpoint alone is rejected.
-#   model: "openai/<model-name>"
-#   api_base: "http://localhost:8080/v1"
-#   api_key: "sk-noop"
-
-# Any other LiteLLM/OpenAI-compatible parameters are forwarded unchanged
-# (e.g. temperature, top_p, max_tokens). Omit them to keep provider defaults.
-
-# OpenRouter provider routing (optional): prefer specific upstream providers,
-# in order.
-#   extra_body:
-#     provider:
-#       order: ["<provider-slug>"]  # Preference-ordered list of OpenRouter provider slugs.