Write a viral-but-honest X/Twitter thread for a benchmark or eval result (e.g. DeepSWE, ProgramBench, ponytail, pi-observational-memory runs). Use when the user asks to write, draft, rewrite, iterate on, or render a tweet thread about benchmark results, eval…
Fetch a full X/Twitter thread — all posts with untruncated text plus images — without auth or the paid API. Use when the user wants to retrieve, extract, scrape, pull, save, or download an X/Twitter thread, tweet text, or tweet images, mentions a status URL…
Use before preparing or executing a confirmed deep-swe-bench launch, especially when configs use advisor, observational-memory workers, subagents, local-vLLM shims, or any model beyond the main executor; use before claiming a launch is working.
Paired trajectory analysis for benchmark churn. Use when comparing two configs on matched task/rep cells, explaining solve flips, diagnosing a local model against a frontier reference, separating net score from churn, or preparing evidence to improve a skill,…
Use before adding or changing a deep-swe-bench config release, model leaf, provider/model API path, config lock, role declaration, usage parser, smoke contract, or extension/subagent worker usage accounting.
Prompt embedding analysis. Use when clustering benchmark prompt/config text, comparing semantic neighbors, or separating prompt-shaped effects from behavioral wrappers in deep-swe-bench results.
Coverage-guided fuzzing for implementation and test work on input-processing and memory-safety boundaries. Use when a feature or bug touches parsers, lexers, decoders, deserializers, codecs, file formats, protocol handlers, malformed, chunked, or adversarial…
Property-based testing for implementation and test work with broad structured behavior. Use when a feature or bug spans input combinations, round trips, encoding and decoding, serialization, normalization or idempotence, ordering or pagination, schema…