Debug Mode logo

Debug Mode

Organization
MadAppGang
debug-mode

Enable, disable, and manage debug mode for agentdev sessions. Records all tool invocations, skill activations, hook triggers, and agent delegations to JSONL. Use when debugging agent behavior, optimizing workflows, or analyzing session performance.

Overview

PublisherMadAppGang
Repositoryclaude-code
Skill namedebug-mode
Stars
281
Forks
26
Bundled files
Instructions only
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by MadAppGang on GitHub. Read the source before you install it.

Installation

Install the Debug Mode AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/MadAppGang/claude-code.git /tmp/claude-code
mkdir -p .claude/skills
cp -r /tmp/claude-code/plugins/agentdev/skills/debug-mode .claude/skills/debug-mode
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Debug Mode in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Debug Mode on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Debug Mode is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

plugin: agentdev updated: 2026-01-20

AgentDev Debug Mode

Debug mode captures detailed session information for analysis, debugging, and optimization. All events are recorded to a JSONL file in claude-code-session-debug/.

Configuration

Debug mode uses per-project configuration stored in .claude/agentdev-debug.json.

Config File Format

Location: .claude/agentdev-debug.json (in project root)

json
{
  "enabled": true,
  "level": "standard",
  "created_at": "2026-01-09T07:00:00Z"
}

Fields:

  • enabled: boolean - Whether debug mode is active
  • level: string - Debug level (minimal, standard, verbose)
  • created_at: string - ISO timestamp when config was created

Enabling Debug Mode

Use the command to create the config file:

/agentdev:debug-enable

This creates .claude/agentdev-debug.json with enabled: true.

Or manually create the file:

bash
mkdir -p .claude
cat > .claude/agentdev-debug.json << 'EOF'
{
  "enabled": true,
  "level": "standard",
  "created_at": "2026-01-09T07:00:00Z"
}
EOF

Debug Levels

LevelCaptured Events
minimalPhase transitions, errors, session start/end
standardAll of minimal + tool invocations, agent delegations
verboseAll of standard + skill activations, hook triggers, full parameters

Default level is standard.

Changing Debug Level

Using jq:

bash
jq '.level = "verbose"' .claude/agentdev-debug.json > tmp.json && mv tmp.json .claude/agentdev-debug.json

Output Location

Debug sessions are saved to:

claude-code-session-debug/agentdev-{slug}-{timestamp}-{id}.jsonl

Example:

claude-code-session-debug/agentdev-graphql-reviewer-20260109-063623-ba71.jsonl

JSONL Format

Each line in the JSONL file is a complete JSON event object. This append-only format is:

  • Crash-resilient (no data loss on unexpected termination)
  • Easy to process with jq
  • Streamable during the session

Event Schema (v1.0.0)

json
{
  "event_id": "550e8400-e29b-41d4-a716-446655440001",
  "correlation_id": null,
  "timestamp": "2026-01-09T06:40:00Z",
  "type": "tool_invocation",
  "data": { ... }
}

Fields:

  • event_id: Unique UUID for this event
  • correlation_id: Links related events (e.g., tool_invocation -> tool_result)
  • timestamp: ISO 8601 timestamp
  • type: Event type (see below)
  • data: Type-specific payload

Event Types

TypeDescription
session_startSession initialization with metadata
session_endSession completion
tool_invocationTool called with parameters
tool_resultTool execution result
skill_activationSkill loaded by agent
hook_triggerPreToolUse/PostToolUse hook fired
agent_delegationTask delegated to sub-agent
agent_responseSub-agent returned result
phase_transitionWorkflow phase changed
user_interactionUser approval/input requested
proxy_mode_requestExternal model request via Claudish
proxy_mode_responseExternal model response
errorError occurred

What Gets Captured

Session Metadata

  • Session ID and path
  • User request
  • Environment (Claudish availability, plugin version)
  • Start/end timestamps

Tool Invocations

  • Tool name
  • Parameters (sanitized - credentials redacted)
  • Execution context (phase, agent)
  • Duration and result size

Agent Delegations

  • Target agent name
  • Prompt preview (first 200 chars)
  • Proxy mode model if used
  • Session path

Proxy Mode

  • Model ID
  • Request/response duration
  • Success/failure status

Phase Transitions

  • From/to phase numbers and names
  • Transition reason (completed, skipped, failed)
  • Quality gate results

Errors

  • Error type (tool_error, hook_error, agent_error, etc.)
  • Message and stack trace
  • Context (phase, agent, tool)
  • Recoverability

Sensitive Data Protection

Debug mode automatically sanitizes sensitive data:

Redacted Patterns:

  • API keys (sk-*, ghp_*, AKIA*, etc.)
  • Tokens (bearer, access, auth)
  • Passwords and secrets
  • AWS credentials
  • Slack tokens (xox*)
  • Google API keys (AIza*)

Analyzing Debug Output

Prerequisites

Install jq for JSON processing:

bash
# macOS
brew install jq

# Linux
apt-get install jq

Quick Statistics

bash
# Count events by type
cat session.jsonl | jq -s 'group_by(.type) | map({type: .[0].type, count: length})'

Tool Usage Analysis

bash
# Tool invocation counts
cat session.jsonl | jq -s '
  [.[] | select(.type == "tool_invocation") | .data.tool_name]
  | group_by(.)
  | map({tool: .[0], count: length})
  | sort_by(-.count)'

Failed Operations

bash
# Find all errors and failed tool results
cat session.jsonl | jq 'select(.type == "error" or (.type == "tool_result" and .data.success == false))'

Timeline View

bash
# Chronological event summary
cat session.jsonl | jq '"\(.timestamp) [\(.type)] \(.data | keys | join(", "))"'

Event Correlation

bash
# Find tool invocation and its result
INVOCATION_ID="550e8400-e29b-41d4-a716-446655440001"
cat session.jsonl | jq "select(.event_id == \"$INVOCATION_ID\" or .correlation_id == \"$INVOCATION_ID\")"

Phase Duration Analysis

bash
# Calculate time between phase transitions
cat session.jsonl | jq -s '
  [.[] | select(.type == "phase_transition")]
  | sort_by(.timestamp)
  | .[]
  | {phase: .data.to_name, timestamp: .timestamp}'

Agent Delegation Timing

bash
# Find slowest agent delegations
cat session.jsonl | jq -s '
  [.[] | select(.type == "agent_response")]
  | sort_by(-.data.duration_ms)
  | .[:5]
  | .[]
  | {agent: .data.agent, duration_sec: (.data.duration_ms / 1000)}'

Proxy Mode Performance

bash
# External model response times
cat session.jsonl | jq -s '
  [.[] | select(.type == "proxy_mode_response")]
  | .[]
  | {model: .data.model_id, success: .data.success, duration_sec: (.data.duration_ms / 1000)}'

Disabling Debug Mode

Use the command:

/agentdev:debug-disable

Or manually update:

bash
jq '.enabled = false' .claude/agentdev-debug.json > tmp.json && mv tmp.json .claude/agentdev-debug.json

Or delete the config file:

bash
rm -f .claude/agentdev-debug.json

Cleaning Up Debug Files

Remove All Debug Files

bash
rm -rf claude-code-session-debug/

Remove Files Older Than 7 Days

bash
find claude-code-session-debug/ -name "*.jsonl" -mtime +7 -delete

Remove Files Larger Than 10MB

bash
find claude-code-session-debug/ -name "*.jsonl" -size +10M -delete

File Permissions

Debug files are created with restrictive permissions:

  • Directory: 0o700 (owner only)
  • Files: 0o600 (owner read/write only)

This prevents other users from reading potentially sensitive session data.

Example Session Output

jsonl
{"event_id":"init-1736408183","timestamp":"2026-01-09T06:36:23Z","type":"session_start","data":{"schema_version":"1.0.0","session_id":"agentdev-graphql-reviewer-20260109-063623-ba71","user_request":"Create an agent that reviews GraphQL schemas","session_path":"ai-docs/sessions/agentdev-graphql-reviewer-20260109-063623-ba71","environment":{"claudish_available":true,"plugin_version":"1.4.0","jq_available":true}}}
{"event_id":"550e8400-e29b-41d4-a716-446655440001","timestamp":"2026-01-09T06:36:25Z","type":"tool_invocation","data":{"tool_name":"TodoWrite","parameters":{"todos":"[REDACTED]"},"context":{"phase":0,"agent":null}}}
{"event_id":"550e8400-e29b-41d4-a716-446655440002","correlation_id":"550e8400-e29b-41d4-a716-446655440001","timestamp":"2026-01-09T06:36:25Z","type":"tool_result","data":{"tool_name":"TodoWrite","success":true,"result_size_bytes":156,"duration_ms":12}}
{"event_id":"550e8400-e29b-41d4-a716-446655440003","timestamp":"2026-01-09T06:36:26Z","type":"phase_transition","data":{"from_phase":null,"to_phase":0,"from_name":null,"to_name":"Init","transition_reason":"completed","quality_gate_result":true}}
{"event_id":"550e8400-e29b-41d4-a716-446655440004","timestamp":"2026-01-09T06:36:30Z","type":"agent_delegation","data":{"target_agent":"agentdev:architect","prompt_preview":"SESSION_PATH: ai-docs/sessions/agentdev-graphql-reviewer...","prompt_length":1456,"proxy_mode":null,"session_path":"ai-docs/sessions/agentdev-graphql-reviewer-20260109-063623-ba71"}}
{"event_id":"end-1736408565","timestamp":"2026-01-09T06:42:45Z","type":"session_end","data":{"success":true}}

Troubleshooting

Debug File Not Created

  1. Check if debug mode is enabled:

    bash
    /agentdev:debug-status
  2. Verify config file:

    bash
    cat .claude/agentdev-debug.json
  3. Verify the directory is writable:

    bash
    ls -la claude-code-session-debug/

jq Commands Not Working

  1. Install jq: brew install jq or apt-get install jq
  2. Verify JSONL format (each line should be valid JSON):
    bash
    head -1 session.jsonl | jq .

Large Debug Files

Debug files can grow large in verbose mode. Use minimal level for lighter capture:

Update config:

bash
jq '.level = "minimal"' .claude/agentdev-debug.json > tmp.json && mv tmp.json .claude/agentdev-debug.json

Or clean up old files regularly:

bash
find claude-code-session-debug/ -name "*.jsonl" -mtime +3 -delete

Integration with Other Tools

Viewing in VS Code

The JSONL format works with JSON syntax highlighting. For better viewing:

  1. Install "JSON Lines" VS Code extension
  2. Use "Format Document" on each line individually

Importing to Analytics

bash
# Convert to CSV for spreadsheet import
cat session.jsonl | jq -rs '
  (.[0] | keys_unsorted) as $keys
  | ($keys | @csv),
  (.[] | [.[$keys[]]] | @csv)' > session.csv

Streaming to External Service

bash
# Tail and send to logging service
tail -f session.jsonl | while read line; do
  curl -X POST -d "$line" https://logging.example.com/ingest
done

Commands Reference

CommandDescription
/agentdev:debug-enableEnable debug mode (creates config file)
/agentdev:debug-disableDisable debug mode (updates config file)
/agentdev:debug-statusCheck current debug mode status

Frequently asked questions

What does the Debug Mode AI skill do?

Enable, disable, and manage debug mode for agentdev sessions. Records all tool invocations, skill activations, hook triggers, and agent delegations to JSONL. Use when debugging agent behavior, optimizing workflows, or analyzing session performance.

Why use Debug Mode on TypingMind?

Because you install it once and use it with any model. Debug Mode is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Debug Mode in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/MadAppGang/claude-code/tree/main/plugins/agentdev/skills/debug-mode. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Debug Mode?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Debug Mode?

As many as you like. As long as a model supports skills, you can use Debug Mode with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Debug Mode AI skill free?

Yes. It is published on GitHub by MadAppGang under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇