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 skills/matteing/opal/debuggingnpx skills add matteing/opal --skill debugginggit clone --depth 1 https://github.com/matteing/opalWhat 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 | $0.00050 | $0.02010 |
| Opus 5 | $0.00025 | $0.01005 |
| Sonnet 5 | $0.00010 | $0.00402 |
| Haiku 4.5 | $0.00005 | $0.00201 |
Grade A, and why
debugging 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 — 201 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Debugging Skill
You use the debug_state tool and Opal's runtime introspection to diagnose problems during a session. This skill teaches you when and how to self-diagnose.
When to act
- A tool call fails unexpectedly or returns surprising results.
- The agent seems stuck in a loop or keeps retrying the same action.
- You suspect context is getting large and may be near the token limit.
- A feature (sub-agents, skills) doesn't seem to be working.
- The user asks you to debug yourself or inspect your own state.
The debug feature is disabled by default
The debug_state tool and the in-memory event log are gated behind the debug feature flag, which is off by default. If you try to call debug_state and the tool isn't available, tell the user they need to enable it first.
How the user enables debug mode
From the TUI menu: Press Ctrl+\ (or the configured menu hotkey) to open the Opal Menu, then toggle Debug introspection on. This takes effect immediately for the current session.
From Elixir code:
Opal.start_session(%{
features: %{debug: %{enabled: true}}
})
Via application config:
config :opal,
features: %{debug: %{enabled: true}}
When debug is enabled, Opal keeps an in-memory event log (bounded ring buffer of the last 400 events) and exposes the debug_state tool. When debug is disabled, the event log is cleared and the tool is filtered out of the active tool set.
Using debug_state
Basic snapshot (no messages, no events)
debug_state({})
Returns: session ID, model info, provider, working directory, token usage, tool availability, and queue state.
With recent events
debug_state({"event_limit": 100})
The event_limit parameter controls how many recent events to include (default: 50, max: 500). Events are returned newest-first and include a timestamp, event type, and a truncated data preview.
With conversation messages
debug_state({"include_messages": true, "message_limit": 10})
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 · 201 lines · 50 tokens per session scan A f80cb9aa9dcd
debugging is a skill published in the GitHub repository matteing/opal (58 stars, last pushed 20d ago), licensed MIT. It adds 50 tokens to every session and 2,010 once invoked, about $0.0003 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 skills, from other repositories
agent-pattern-verification
Verify AI agent code for dangerous patterns including infinite loops, unbounded retries, tool hallucinations, and context overflow. Use before deploying agent workflows or when reviewing agent code.
cloudflare-agent
Enable ControlKeel governance for Cloudflare Agents with policy gates, budget enforcement, PII detection, and secure execution.
controlkeel-governance
Operate inside a ControlKeel-governed session. Use this before code edits, shell execution, delegation, deploy work, or any task that needs CK validation, findings, budget, proof, or routing context.
bounded-loop
Run objective, repeatable improvement work under an immutable verifier and hard iteration, cost, time, and no-progress limits. Use for approved experiments with an automated metric and sandboxed worker.
challenge
Adversarially challenge the current plan or finding. Assume the direction is wrong and build the strongest case against it. Surfaces blind spots, false assumptions, and unexplored alternatives before execution starts.
continuity
Learn, record, audit, and apply codebase patterns consistently across a repo by comparing current code to canonical local examples stored in CK memory. Use when asked to preserve continuity, learn a pattern, check drift, fix inconsistent implementations, or enforce local conventions.