trace-claude-code
Automatically trace Claude Code conversations to Braintrust for observability. Captures sessions, conversation turns, and tool calls as hierarchical traces.
Full 5-layer analysis of a specific function. Use when debugging or deeply understanding code.
$ npx -y skills add parcadei/Continuous-Claude-v3 --skill tldr-deep --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
/tldr-deepContext preview
The summary Claude sees to decide when to auto-load this skill.
Full 5-layer analysis of a specific function. Use when debugging or deeply understanding code.
name: tldr-deep description: Full 5-layer analysis of a specific function. Use when debugging or deeply understanding code.
Full 5-layer analysis of a specific function. Use when debugging or deeply understanding code.
| Layer | Purpose | Command | |-------|---------|---------| | L1: AST | Structure | `tldr extract <file>` | | L2: Call Graph | Navigation | `tldr context <func> --depth 2` | | L3: CFG | Complexity | `tldr cfg <file> <func>` | | L4: DFG | Data flow | `tldr dfg <file> <func>` | | L5: Slice | Dependencies | `tldr slice <file> <func> <line>` |
Given a function name, run all layers:
# First find the file tldr search "def <function_name>" . # Then run each layer tldr extract <found_file> # L1: Full file structure tldr context <function_name> --project . --depth 2 # L2: Call graph tldr cfg <found_file> <function_name> # L3: Control flow tldr dfg <found_file> <function_name> # L4: Data flow tldr slice <found_file> <function_name> <target_line> # L5: Slice
## Deep Analysis: {function_name}
### L1: Structure (AST)
File: {file_path}
Signature: {signature}
Docstring: {docstring}
### L2: Call Graph
Calls: {list of functions this calls}
Called by: {list of functions that call this}
### L3: Control Flow (CFG)
Blocks: {N}
Cyclomatic Complexity: {M}
[Hot if M > 10]
Branches:
- if: line X
- for: line Y
- ...
### L4: Data Flow (DFG)
Variables defined:
- {var1} @ line X
- {var2} @ line Y
Variables used:
- {var1} @ lines [A, B, C]
- {var2} @ lines [D, E]
### L5: Program Slice (affecting line {target})
Lines in slice: {N}
Key dependencies:
- line X → line Y (data)
- line A → line B (control)
---
Total: ~{tokens} tokens (95% savings vs raw file)1. **Debugging** - Need to understand all paths through a function 2. **Refactoring** - Need to know what depends on what 3. **Code review** - Analyzing complex functions 4. **Performance** - Finding hot spots (high cyclomatic complexity)
from tldr.api import (
extract_file,
get_relevant_context,
get_cfg_context,
get_dfg_context,
get_slice
)
# All layers for one function
file_info = extract_file("src/processor.py")
context = get_relevant_context("src/", "process_data", depth=2)
cfg = get_cfg_context("src/processor.py", "process_data")
dfg = get_dfg_context("src/processor.py", "process_data")
slice_lines = get_slice("src/processor.py", "process_data", target_line=42)A persistent, learning, multi-agent development environment built on Claude Code Continuous Claude transforms Claude Code into a continuously learning system that maintains context across sessions, orchestrates specialized agents, and eliminates wasting
Repo: parcadei/Continuous-Claude-v3
Automatically trace Claude Code conversations to Braintrust for observability. Captures sessions, conversation turns, and tool calls as hierarchical traces.
Guide for integrating Agentica SDK with Claude Code CLI proxy
Reference guide for Agentica multi-agent infrastructure APIs