惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

量子位
WordPress大学
WordPress大学
小众软件
小众软件
云风的 BLOG
云风的 BLOG
IT之家
IT之家
人人都是产品经理
人人都是产品经理
奇客Solidot–传递最新科技情报
奇客Solidot–传递最新科技情报
Last Week in AI
Last Week in AI
博客园 - 【当耐特】
T
Tailwind CSS Blog
阮一峰的网络日志
阮一峰的网络日志
V
V2EX
宝玉的分享
宝玉的分享
博客园 - Franky
F
Fortinet All Blogs
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
GbyAI
GbyAI
Hugging Face - Blog
Hugging Face - Blog
Jina AI
Jina AI
D
Docker
博客园 - 聂微东
C
Check Point Blog
H
Help Net Security

Show HN

GitHub - astefanutti/shaderbang: Shebang for Shaders Show HN: Generate Claude Code Workflows using Spec Driven Development approach Show HN: AI agents for UK GDAD PCF roles and their skills The Two Pillars: Mixer Mode and Meta-Software in the Reorganization of Software Work After AI GitHub - JaiCode08/teleport-env What 1,000+ Harness Experiments Taught Me About Self-Improving Agents Show HN: Liiists, a Markdown-first, iOS and CLI list app SwiperTab – Get this Extension for 🦊 Firefox (en-US) GitHub - kouhxp/fftext: Summarize, explain, fact-check, or translate any text, URL, or file. No GPU. No cloud. One command GitHub - sweetpad-dev/sweetpad: Develop Swift/iOS projects using VSCode GitHub - dogmaticdev/IRON: IRON a.k.a. Intermediate Representation Object Notation is a Interpreter/Database that is used to create Programming Languages. GitHub - sjhalani7/vaen: Package your AI coding harness into a portable .agent file, and share it across repos, teams, & the community without ever having to copy-paste instructions, skills, MCP config, or secrets. Show HN: Gandalf the Grader Show HN: Citadeld – replay any CI failure locally from a single file GitHub - tdortman/cuSBF: High-Performance GPU Super Bloom Filter coral-ai/claude-code-token-xray at main · Coral-Bricks-AI/coral-ai GitHub - ulyssestenn/funes: Funes is a Git-based framework for LLM-managed knowledge work: an AI Librarian ingests raw sources, builds an interlinked Markdown knowledge base, and uses it to produce cited reports, analyses, and other outputs. GitHub - ThatXliner/gah: Git Add Hunk, built for agents to use GitHub - harmont-dev/harmont-cli: Command-line client for the Harmont CI platform GitHub - brooksmcmillin/mcp-authflow: OAuth 2.0 Authorization Server framework for MCP servers GitHub - javaid-codes/audit-supply-chain-agents GitHub - amorey/gochan: A small library of common channel architectures for Go, inspired by Rust GitHub - arifozgun/OpenGem: Free, Open-Source AI API Gateway with Gemini, OpenAI & Anthropic Compatibility in 1 file GitHub - Pranesh950/BioPetals: 🌸 Run BIOxAI models at home, BitTorrent-style. Fine-tuning and inference up to 10x faster than offloading GitHub - cnguyen14/bounty-doctor: Diagnose a GitHub bounty issue before you waste hours: detects honeypot scam repos, AI-bot attempt swarms, and stale contests. Show HN: CoreMCP – MCP Server for On-Prem DBs Show HN: KittyHTML – Render HTML/CSS as an inline image in your terminal GitHub - bingud/filemat: Web-based file manager Show HN: TruthLens – Free multi-signal deepfake image detector GitHub - apexlocal-jz/claude-usage-tray: Windows system-tray app showing your Claude Code rate-limit usage at a glance. Zero deps, ~300 lines of PowerShell. Cross-IDE (works regardless of VS Code, Cursor, plain terminal).
GitHub - harihkk/Debug-Brief: Local-first CLI that turns ...
itshkrishna · 2026-06-22 · via Show HN

DebugBrief

CI PyPI

Record what you do while debugging and turn it into an evidence-backed Markdown report for a pull request, a handoff, or an incident note.

A failing test streams live, the fix lands, redo passes, and the generated report appears

DebugBrief captures the notes you write and the commands you run, then builds a report from what actually happened: what you tried, what failed, what passed, and which files changed in between. It does not use AI and does not infer a root cause or report a test result you did not get.

Install

pipx install debugbrief
# or
uv tool install debugbrief

Plain pip install debugbrief works too. DebugBrief needs Python 3.9 or newer. Native Git is used when available to capture repository metadata and changed files. The project you debug does not need to be Python; only DebugBrief runs on Python.

Quickstart

debugbrief start "Fix add() returning wrong result"
debugbrief note "add() subtracts instead of adds; the test expects 5."
debugbrief run -- python -m pytest -q test_calc.py   # fails
# make your fix
debugbrief redo                                      # same test, now passes
debugbrief end                                       # writes the PR report

Everything after -- runs exactly as you typed it, with its output streaming live to your terminal. DebugBrief flags (--timeout, --shell, --no-redact, --verify) go before the --. redo re-runs the last captured command, and end defaults to the pr report mode.

Tip: a one-line alias makes the capture prefix disappear in daily use.

alias db="debugbrief run --"
db pytest -q

What you get

A report built only from recorded evidence. A short excerpt from a real run:

## Summary

Failing check `python -m pytest -q test_calc.py` passed after 2 attempts over 2s, changes touched calc.py.

## Red to green

A check failed at 12:02:09 and `python -m pytest -q test_calc.py` passed at 12:02:10 (window 1s).

Between the failing and passing checks, these files changed (correlation, not proven cause):
- `calc.py`

Full samples: PR, handoff, incident.

How it works

  • run executes a command under a pseudo-terminal so its output streams live, then records the real exit code, a bounded output preview, the duration, and a per-command git snapshot.
  • Pass or fail comes only from the exit code. A command counts as verified only when a recognized test, build, lint, or typecheck runner actually exits 0.
  • It works with any language. A recognized runner (pytest, jest, vitest, go test, cargo test, dotnet test, make check, and more) is classified automatically. Any other command is still captured, and you mark it a check with --verify.
  • end derives the report from those events: the red-to-green window, the reproduce and verify commands, a timeline, the observed error, and the failed attempts. Empty sections are omitted.
  • Secrets are redacted before anything is written to disk. Redaction is best effort; --no-redact opts out for a single command.

Full command reference and the complete recognized-runner list: docs/COMMANDS.md. Security model and redaction details: SECURITY.md.

Post a report straight to a pull request (GitHub CLI optional):

debugbrief end --stdout | gh pr comment --body-file -

Commands

The five you use most:

Command What it does
start "<title>" Start a session
note <text> Record an observation
run -- <command> Run and capture a command
redo Re-run the last captured command
end [--mode pr|handoff|incident] Finalize and write a report (default pr)

The rest:

Command What it does
init Set up the project and show the workflow
status Show the active session
preview [--mode ...] Print the report without ending the session
cancel [--yes] Discard the active session, no report
doctor [--fix] Health-check the project and state
recover Repair a broken session pointer after a crash
last Show the most recent report path
open Open the most recent report
list List recorded sessions
show <id> Show a recorded session

Full flags and behavior for each: docs/COMMANDS.md.

Dependencies

DebugBrief uses the Python standard library and native git. On Python 3.11 and newer it needs nothing else. On Python 3.9 and 3.10 it uses the small tomli package to read an optional .debugbrief.toml. DebugBrief itself makes no network requests, uses no AI, and collects no telemetry.

Supported platforms

Linux and macOS are tested in CI across Python 3.9 through 3.14. Other Unix-like systems may work but are not currently tested. Native Windows and PowerShell are not supported.

Limitations

  • Capture is explicit through debugbrief run. Output streams live while a bounded preview is stored for the report, so there is no full transcript.
  • Full-screen TUIs (a vim session, htop) are not meaningfully captured. Run those directly and record the outcome with note.
  • Redaction is conservative and best effort; it does not catch every secret.
  • Git sections need native git; outside a repository they are omitted.

Development

pip install -e ".[dev]"
pytest

See CONTRIBUTING.md for the full setup and contribution guidelines.

License

MIT. See LICENSE.