ff_debug

Investigate failing tasks using git state, logs, and local commands to produce structured debug reports.

1|1|Updated Feb 3, 2026
One-click install
npx skills add https://github.com/redblacktree/fastflow --skill ff-debug-redblacktree
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ff_debug
Source: https://github.com/redblacktree/fastflow/tree/main/internal/templates/files/.codex/skills/ff_debug
Command: npx skills add https://github.com/redblacktree/fastflow --skill ff-debug-redblacktree

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? When a Fastflow task or implementation fails, developers need a disciplined way to gather evidence from the local repository before proposing fixes, instead of guessing at root causes. ## Core Features & Use Cases - Evidence-First Investigation: Reads goals and plans, inspects git status, recent commits, and diffs, then searches logs and error text before drawing conclusions. - Structured Debug Reports: Produces a consistent report covering the problem, evidence with file and line references, likely root cause with confidence level, recommended fix, verification commands, and unknowns. - Safe Read-Only Operation: Does not edit files during debugging unless explicitly asked, and prefers reproducible commands over speculation. - Use Case: A validation stage fails in a Fastflow workflow. Run this Skill to inspect the failing tests, trace the error through recent commits, and receive a report pinpointing the likely cause and the exact command to verify the fix. ## Quick Start Ask the agent to debug the failing task described in the current goal file and produce a debug report with evidence and a recommended fix.

Frequently Asked Questions about ff_debug

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I debug a failing Fastflow workflow task?

Run the ff_debug skill with the goal or plan file for the failing task. It inspects git status, recent commits, logs, and error text, then produces a structured debug report with the likely root cause and recommended fix.

What evidence does an AI debugging agent collect before proposing a fix?

This skill gathers git status, the last ten commits, relevant diffs, error text, failing tests, and logs from local paths. It runs the narrowest safe reproduction command and clearly separates confirmed facts from hypotheses.

Can the debug skill modify my code while investigating?

No. The skill operates read-only by default and does not edit files while debugging unless you explicitly ask it to. It focuses on evidence gathering and reporting, leaving fixes as recommendations.

What happens when logs or external services are unavailable during debugging?

The skill states exactly what could not be checked in the Unknowns section of the debug report. It does not assume project-specific external services unless the repository itself documents them.

What format does the debug report output use?

The report is Markdown with sections for Problem, Evidence with file and line references, Likely Root Cause with a confidence level, Recommended Fix, Verification commands, and Unknowns.