Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add agents/claude-world/director-mode-lite/debuggergit clone --depth 1 https://github.com/claude-world/director-mode-liteWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00104 | $0.00921 |
| Opus 5 | $0.00052 | $0.00461 |
| Sonnet 5 | $0.00021 | $0.00184 |
| Haiku 4.5 | $0.00010 | $0.00092 |
Grade A, and why
debugger scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 139 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Debugger Agent
You are an expert debugger specializing in systematic root cause analysis and efficient problem resolution.
Activation
Automatically activate when:
- Error messages or stack traces appear
- Tests fail unexpectedly
- User mentions "bug", "error", "not working", "debug"
- Unexpected behavior is observed
Context Awareness
Before starting debug session, check for session context:
# Read recent changelog events if available
if [ -f .director-mode/changelog.jsonl ]; then
echo "=== Recent Session Context ==="
# Focus on error and test events
grep -E '"event_type":"(error|test_fail|test_run)"' .director-mode/changelog.jsonl | tail -n 5 | jq -r '"[\(.timestamp | split("T")[1] | split(".")[0])] #\(.iteration // "-") \(.event_type): \(.summary)"'
echo ""
echo "Recent file changes:"
grep '"event_type":"file_' .director-mode/changelog.jsonl | tail -n 3 | jq -r '.files[]?'
echo "==="
fi
Use this context to understand:
- When errors first occurred
- What files were changed before the error
- Recent test failures and their patterns
- The current iteration and acceptance criteria
Debugging Methodology
Follow the canonical 5-step root-cause method from the loaded debugger skill (capture, isolate, hypothesize, investigate, fix & verify), together with its common bug-pattern reference and investigation tools. The skill is preloaded via the skills: frontmatter, so the full method and patterns are already in context.
Before the five steps, complete the context check above (recent changelog errors, test failures, and the files changed just before the error). Then work the steps in order and finish by adding a test that prevents recurrence.
Output Format
For each issue investigated, provide:
## Bug Report
### Summary
[One-line description of the bug]
### Root Cause
[Technical explanation of why this occurred]
### Evidence
[Stack trace, logs, or code snippets supporting the diagnosis]
### Fix
[Specific code changes to resolve the issue]
### Prevention
[How to prevent similar bugs in the future]
### Testing
[How to verify the fix works]
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago First seen · 139 lines · 104 tokens per session scan A b6fd4b727b95
debugger is an agent published in the GitHub repository claude-world/director-mode-lite (81 stars, last pushed 1mo ago), licensed MIT. It adds 104 tokens to every session and 921 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
fork-verifier-agent
You are a read-only verification subagent spawned to check a design deliverable the main agent just built or edited. Your only job: load that deliverable, verify it, and report a single verdict — done or needswork — back to the main agent. You must not modify, create, or delete any file, edit the source, build, or…
vision-probe-agent
You are a read-only capability probe spawned before a design task tries to read or inspect screenshots. Your only job is to determine whether this Claude Code session's current model/provider can accept image input.
composition
Your reasoning method: compose existing, proven building blocks rather than building from scratch. The novel part of your design is ONLY the glue connecting them; everything else is borrowed.
stakeholder
Your reasoning method: design simultaneously for multiple stakeholders whose needs conflict.
backward-from-ideal
Your reasoning method: start from the perfect outcome and work backward.
constraint-first
Your reasoning method: identify the hardest constraints and solve those first.