Releases·Tool Release
Show HN: Gage – Rust-based CLI scans Claude Code session transcripts for bugs with cited evidence
A developer released Gage, an open source Apache 2 CLI and TUI written in Rust that scans Claude Code session transcripts and files issues with evidence. Unlike diff-based review, it uses sessions' code and reasoning, and the author reports it caught a bug he had approved in the original session. Every issue must cite lines in the session record so claims trace to their source.
6 engagement1 sourceFri, Sep 4, 2026, 04:13 PM
AI Overview
- Gage, an Apache 2 Rust CLI + TUI, scans Claude Code session transcripts and files issues that each cite lines in the session record
- Author says Gage caught a bug he had approved after Claude flagged the risk in-session; the commit alone didn't show it
- Scans run under your Claude login: counted against subscription limits or about $0.50–$1.00 per session on usage billing, roughly $2 per resolved issue daily
- Gage integrates with Claude Code to review, fix, or close issues as skipped
Entities
Related
News·Incident
Developer discovers Claude Code silently overcharged users 100x due to billing bug
A developer using Claude Code to ship a product discovered a bug where users were overcharged by 100x due to an incorrect assumption made by the AI during a coding session. The bug was nearly impossible to spot in code review, leading the developer to create a hook that extracts decisions and assumptions from session transcripts before PR creation.
A developer using Claude Code to ship a product discovered a bug where users were overcharged by 100x due to an incorrect assumption made by the AI during a coding session. The bug was nearly impossible to spot in code review, leading the developer to create a hook that extracts decisions and assumptions from session transcripts before PR creation.
Updated Tue, Jul 21, 2026, 10:23 PM — Developer creates a hook to extract decisions from Claude Code transcripts to prevent similar bugs.
13 engagement1 sourceredditTue, Jul 21, 2026, 10:23 PM
Products·launched
Autonomous audit loop for Claude Code with verified task backlog
A Claude Code skill that autonomously audits a codebase, writes a backlog of tasks each with runnable acceptance checks, and iterates until a clean audit and adversarial sign-off. It produces a git log of checkpoints, enabling trust in the changes. Found and fixed bugs in bat, chalk, and fasthttp that were merged upstream.
A Claude Code skill that autonomously audits a codebase, writes a backlog of tasks each with runnable acceptance checks, and iterates until a clean audit and adversarial sign-off. It produces a git log of checkpoints, enabling trust in the changes. Found and fixed bugs in bat, chalk, and fasthttp that were merged upstream.
7 engagement1 sourceredditMon, Aug 3, 2026, 01:47 PM
Products·building
Open-source agent that watches user sessions, finds and fixes bugs
Opslane is an open-source agent that monitors user sessions to identify user-facing issues, investigates them, and automatically creates pull requests with verified fixes. It uses LLMs to analyze session data and code, prioritizing bugs that matter to customers. It solves the problem of overwhelming bug backlogs by focusing on real user impact.
Opslane is an open-source agent that monitors user sessions to identify user-facing issues, investigates them, and automatically creates pull requests with verified fixes. It uses LLMs to analyze session data and code, prioritizing bugs that matter to customers. It solves the problem of overwhelming bug backlogs by focusing on real user impact.
7 engagement1 sourcehackernewsThu, Aug 27, 2026, 03:45 PM
Products·launched
RippleCheck: catch side effects from AI code changes via AST analysis
RippleCheck is an open-source tool that scans a codebase using AST analysis (ts-morph) to detect functions and dependencies that might break when an AI tool makes changes. It does not auto-fix anything but highlights what to check and generates a copyable prompt for the AI tool. It works as an MCP server and integrates with Claude Code hooks.
RippleCheck is an open-source tool that scans a codebase using AST analysis (ts-morph) to detect functions and dependencies that might break when an AI tool makes changes. It does not auto-fix anything but highlights what to check and generates a copyable prompt for the AI tool. It works as an MCP server and integrates with Claude Code hooks.
3 engagement1 sourceredditThu, Jul 30, 2026, 12:12 PM
Products·launched
Hubo: Claude Code skill with implementer and reviewer roles
Hubo is an open-source, MIT-licensed skill for Claude Code that sets up a two-role workflow: an implementer writes and tests code, while an independent, read-only reviewer challenges the diff and evidence. The implementer fixes findings or pushes back with evidence, repeating until the reviewer clears the work, addressing the problem of coding agents reviewing their own work with blind spots.
Hubo is an open-source, MIT-licensed skill for Claude Code that sets up a two-role workflow: an implementer writes and tests code, while an independent, read-only reviewer challenges the diff and evidence. The implementer fixes findings or pushes back with evidence, repeating until the reviewer clears the work, addressing the problem of coding agents reviewing their own work with blind spots.
9 engagement1 sourceredditMon, Jul 27, 2026, 09:25 PM