log-agent-issue

Appends structured JSONL issue entries from subagents to a session log file.

6|1|Updated Nov 26, 2025
One-click install
npx skills add https://github.com/nathanvale/side-quest-marketplace-old --skill log-agent-issue-nathanvale
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: log-agent-issue
Source: https://github.com/nathanvale/side-quest-marketplace-old/tree/main/plugins/log-agent-issue/skills/log-agent-issue
Command: npx skills add https://github.com/nathanvale/side-quest-marketplace-old --skill log-agent-issue-nathanvale

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Subagents often encounter errors, warnings, and notable events that disappear without a trace, making it hard to debug multi-agent workflows after the fact. This Skill provides a fire-and-forget way to record those issues in a structured, centralized log. ## Core Features & Use Cases - Structured JSONL Logging: Appends issue entries with type, message, tool name, context, and suggestions to ~/.claude/logs/agent-issues.jsonl. - Non-Blocking Execution: Runs in the background with silent failure so logging never interrupts the agent's main workflow. - Automatic Metadata: The script injects timestamp and sessionId automatically for traceability. - Use Case: A triage subagent hits a tool timeout mid-task; it logs an error entry with the tool name and a remediation suggestion, then continues working without delay. ## Quick Start Ask the agent to log any error or warning it encounters during its task using the log-agent-issue skill with a JSON payload describing the issue.

Frequently Asked Questions about log-agent-issue

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

FAQPage Schema
How do I log errors from a subagent without blocking its workflow?

Pipe a JSON payload to the log-issue script via Bash with run_in_background set to true and 2>/dev/null appended. This makes logging fire-and-forget so the agent never waits on or fails because of the logging call.

What is the JSON payload format for logging agent issues?

The payload contains an agentId and an issues array. Each issue has a type (error, warning, or info), a message, and optional fields for toolName, filePath, context, and suggestion. Timestamp and sessionId are added automatically.

Where are agent issue logs stored?

Entries are appended as JSONL to ~/.claude/logs/agent-issues.jsonl. Each line is a self-contained JSON object including the session ID passed as a script argument.

When should I log an info-level issue versus an error?

Use error for tool failures and exceptions, warning for skipped steps or fallback paths, and info only sparingly for notable events. Batch related issues into a single call rather than making multiple logging invocations.

What happens if the logging script fails?

Nothing visible to the workflow. The 2>/dev/null redirect suppresses stderr, so logging failures are silent by design and never interrupt the agent's primary task.