analyze-problem

analyze-problem is a skill for Claude Code, Codex from latitude-dev/latitude-llm. It costs 27 tokens per session (320 once invoked), scanned A, original, MIT.

A method for investigating a reported software problem by tracing its code path, data, permissions, side effects, and expected behavior.

In plain words
What is it for?
Use it to analyze bugs and tasks, find their root causes, compare code with tests or specifications, consider edge cases, and define verification steps.
Why use it?
It turns a vague bug report into a supported explanation of what happens, why it happens, how to fix it, and how to check the fix.

Skill for Claude CodeCodex

About the project

Latitude is an open-source platform for monitoring AI agents by collecting execution traces, grouping failures, dispatching coding agents to make fixes, and replaying failures to verify them. Teams use it to observe agent behavior, investigate errors, and monitor whether fixes prevent regressions. The catalogue entries include skills, instructions, and an MCP server for working with Latitude.

latitude-dev/latitude-llm · 4,626 stars · on GitHub

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 skills/latitude-dev/latitude-llm/analyze-problem
Any agent
npx skills add latitude-dev/latitude-llm --skill analyze-problem
Clone the repo
git clone --depth 1 https://github.com/latitude-dev/latitude-llm

Made for: Claude Code, Codex.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for analyze-problem

README.md
[![agentmods](https://agentmods.dev/badge/skills/latitude-dev/latitude-llm/analyze-problem.svg)](https://agentmods.dev/skills/latitude-dev/latitude-llm/analyze-problem)
Your own site
<a href="https://agentmods.dev/skills/latitude-dev/latitude-llm/analyze-problem"><img src="https://agentmods.dev/badge/skills/latitude-dev/latitude-llm/analyze-problem.svg" alt="Measured on agentmods" height="20"></a>
Per session 27 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 320 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 $0.00027 $0.00320
Opus 5 $0.00014 $0.00160
Sonnet 5 $0.00005 $0.00064
Haiku 4.5 $0.00003 $0.00032

Measured 5d ago against content hash 8dcc4dfc7703, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

analyze-problem 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 5d 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.

.agents/skills/analyze-problem/SKILL.md · 30 lines

What it actually says

Read the provided bug report/issue/task.

Then thoroughly investigate the codebase to understand the relevant context:

  • Find the entry points involved (routes/handlers/jobs/commands) and trace the execution path that leads to the reported behavior.
  • Identify the data flow and state involved (inputs, validation, persistence, caching, permissions, feature flags, side effects).
  • Locate the expected behavior (specs/tests/docs) and compare it to the actual behavior in code.
  • If the report includes steps to reproduce, map each step to the corresponding code path and decision points.
  • Consider common edge cases (null/empty values, timing/race conditions, retries, idempotency, pagination, timezones, auth, stale cache).

Deliverable: Explain, in your own words:

  1. What you believe the issue is (observable behavior + when it happens),
  2. Why you believe it is happening (root cause and the specific assumption/logic that breaks),
  3. How you would fix it (proposed approach and what behavior/logic should change),
  4. What you would verify (minimal set of checks/tests/logs to confirm the fix and prevent regressions).

Constraints:

  • Do not apply any changes automatically.
  • Avoid listing files/components/services unless strictly necessary to support the explanation.
  • If the codebase does not contain enough context to be confident, explicitly say what’s missing and what evidence would confirm the hypothesis.
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. 5d ago First seen · 30 lines · 27 tokens per session scan A 8dcc4dfc7703

Subscribe to this mod's changes

analyze-problem is a skill published in the GitHub repository latitude-dev/latitude-llm (4,626 stars, last pushed today), licensed MIT. It adds 27 tokens to every session and 320 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.