Model report, last 90 days
Claude
93%
Still working
0
Verified by people
178
Automated checks
By version
| Version | Prompts | People | Broke | Auto |
|---|---|---|---|---|
| Haiku 4.5 | 178 | 0 | 0 | 166/178 |
Stopped working
- Security-focused code review against a specific threat model
- Manager/worker multi-agent orchestration design for a complex task
- Prompt injection defense design for an agent that reads untrusted content
- Compress a bloated prompt without silently losing behavior
- Compare how different models handle the same prompt and why they diverge
- Community challenge: golf this prompt to the fewest tokens without losing behavior
- Edit a prompt for clarity without changing what it asks for
- Reverse-engineer a likely prompt from an output you want to reproduce
- Turning an interview transcript into a polished profile piece
- Diagnosing a specific student error pattern from a set of graded work
- Two-minute personalized roast based on five weird facts about me
- Write my life as a movie trailer voiceover, based on real recent events
Holding up
- Course outline design working backward from assessments
- Pricing page copy that handles objections honestly
- Post-mortem writer that separates facts from blame
- Adapt a prompt's tone for a different audience without losing its function
- Show-don't-tell audit that avoids the generic fix
- Postmortem prompt for analyzing a production agent failure
- Worldbuilding consistency and depth pass
- Draft a difficult status update that names risk without spin
- Zero-downtime database schema migration plan
- Personal operating system setup: roles, systems, and weekly rhythm
- Convert a free-form chat prompt into a strict JSON-output prompt
- Prompt an agent to ask clarifying questions instead of guessing on ambiguous requests
- Red-team an agent's system prompt for failure modes before deployment
- Compare two job or project offers with a real weighted framework
- Check whether a prompt's approach actually generalizes to a new use case
- Prioritize a task list with an Eisenhower plus effort overlay
- State machine design for a multi-turn conversational agent with recovery
- Write a tool description that stops an LLM from misusing the function
- Root-cause a flaky test using a structured hypothesis funnel
- How verification works here, and why a failed report is worth as much as a good one
- Design a two-tier memory architecture for a long-running assistant agent
- Debug a confusing stack trace by working backward from the error
- Decompose a long-horizon task into checkpointed subtasks an agent can resume
- Study guide from a syllabus or set of course materials
- REST/GraphQL API design review against real client use cases