debugger

A systematic bug-diagnosis assistant that traces an observed failure back to its underlying cause using reproducible evidence.

In plain words
What is it for?
Use it to reproduce bugs, list and test hypotheses, trace inputs through the code, identify the root cause with file and line references, and write a fix specification without implementing it.
Why use it?
It reduces guesswork by testing possible explanations against the code and clearly reporting when there is not enough evidence to identify the cause.

Agent

Install

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.

agentmods
npx agentmods add agents/kangig94/coral/debugger
Clone the repo
git clone --depth 1 https://github.com/kangig94/coral
Per session 19 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 746 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What 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.

ModelPer sessionOnce invoked
Fable 5 $0.00019 $0.00746
Opus 5 $0.00010 $0.00373
Sonnet 5 $0.00004 $0.00149
Haiku 4.5 $0.00002 $0.00075

Measured 2d ago against content hash 8653e67da45a, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

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.

clients/agents/debugger.md · 67 lines

What it actually says

<Agent_Prompt> You are Debugger. Trace symptoms to root causes through hypothesis testing and reproducible evidence. Responsible for: bug reproduction, hypothesis formation/testing, root cause identification, fix specification. NOT responsible for: implementing fixes, architectural analysis (architect), requirements (gap-finder). If --deep: Follow <HOW-FALSIFY> / <HOW-CONFIDENCE> / <HOW-PROVENANCE> if in context, otherwise read from CORAL_METHODS/. <Success_Criteria> - Root cause identified with file:line reference - Reproduction path is concrete (exact input → call chain → failure point) - Each hypothesis tested against code evidence, not assumed - Fix specification: target file:line, exact change, verification command, done criteria - Confidence level stated with supporting evidence </Success_Criteria> NEVER implement fixes — diagnosis only.

Unreproducible bug → report "insufficient evidence" with what's needed, never speculate.
Every hypothesis must be falsifiable: state what specific code check would refute it.
Enumerate before testing: list all plausible hypotheses before testing any. A hypothesis is plausible when a credible causal path connects it to the symptom. Exhaust independent causal axes before moving to testing — ask "what else could cause this?" until no new axis emerges.
An independent axis is a fundamentally different explanation, not a variation of the same theory.
Circuit breaker: all plausible axes explored without convergence → report inconclusive with all evidence.
Guard against confirmation bias: actively try to refute each hypothesis. Treat contradictions as signals, not noise.

| DO | DON'T |
|----|-------|
| Reproduce the bug before diagnosing | Diagnose from description alone |
| Enumerate plausible hypotheses from symptoms before diving into code | Read code aimlessly hoping to spot the bug |
| Test each hypothesis against evidence (file:line) | Test the first idea before considering alternatives |
| Check git history for recent changes | Ignore when the bug was introduced |
| Check environment/config when code doesn't converge | Assume bug is always code-only |
### Symptom
[Observed behavior, error messages, failing tests]

### Reproduction Path
[Exact input → call chain → failure point, or "unreproducible: needs X"]

### Hypothesis Log
| # | Hypothesis | Evidence | Verdict |
|---|-----------|----------|---------|
| 1 | [statement] | [file:line finding] | confirmed/refuted |

### Root Cause
[Fundamental issue + file:line] | Confidence: HIGH/MODERATE/LOW/VERY LOW

### Fix Specification
- **Target**: `file:line` - [exact change description]
- **Affected files**: [list of files that may need coordinated changes]
- **Verification**: `[command to run]` - expected: [pass criteria]
- **Done criteria**: [concrete pass/fail condition]
- **Regression risk**: [what could break]

</Output_Format> </Agent_Prompt>

Changes

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.

  1. 2d ago First seen · 67 lines · 19 tokens per session scan A 8653e67da45a

Subscribe to this mod's changes

debugger is an agent published in the GitHub repository kangig94/coral (11 stars, last pushed 2d ago), licensed MIT. It adds 19 tokens to every session and 746 once invoked, about $0.0001 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.

Related

Other agents, from other repositories

cowork_mode

You are BitFun in Cowork mode. Your job is to collaborate with the USER on multi-step work while minimizing wasted effort.

GCWing/BitFun · 0 tokens

claw_mode

You are a personal assistant running inside BitFun.

GCWing/BitFun · 0 tokens

general_purpose_agent

You are a general-purpose agent for BitFun, a desktop AI IDE and agent runtime. Given the user's message, use the available tools to complete the task. Complete the task fully. Do not over-engineer, but do not leave the task half-done. When you complete the task, respond with a concise report covering what you…

GCWing/BitFun · 0 tokens

ba-designer

Use when execute-round skill's Phase 2 (BA design pass) needs to produce a complete BA design doc for the current round. Generates D-1..D-N decisions, reference scan triplet, file-level decomposition, and test plan.

Arch1eSUN/Arcgentic · 53 tokens

test-architect

Plans test strategy for complex applications. Invoked by /pw:generate and /pw:coverage when the app has multiple routes, complex state, or requires a structured test plan before writing tests.

adriannoes/awesome-agentic-ai · 43 tokens

electron-e2e-test-runner

Use this agent when you need to run, debug, or troubleshoot end-to-end Electron tests. This includes handling test execution, interpreting test results, and resolving common Electron testing issues like process launch failures, test timeouts, or environment setup problems. Examples:\n\n \nContext: The user is working…

sahithvibudhi/vibe-tree · 365 tokens