Semgrep

Public discussion around Semgrep over the past four weeks was dominated by a persistent undercurrent of frustration, with bug reports and reliability concerns accounting for the largest share of complaint-side mentions by a considerable margin. Commenters surfaced a recurring class of path-handling errors, with several mentions describing hook scripts that incorrectly prepend working directory paths to already-absolute file paths, producing doubled paths and silent scan failures across Windows, WSL, and git worktree environments. This pattern of quiet, hard-to-detect breakdowns seemed to heighten irritation, as discussion suggested that a tool failing silently in a security context is treated as a more serious offense than one that fails loudly.

Sentiment trajectory across the window was unsteady. Early weeks showed very low scores on minimal mention volume, then climbed into the low-to-mid fifties around mid-May before sliding back as mention counts surged in early and mid-June. The largest bursts of activity coincided with score dips, which discussion suggested reflects moments when a wider audience encountered reproducible problems and brought them into public view at once. By the final tracked week sentiment had settled into a narrow, middling band without meaningful recovery.

On the praise side, feature appreciation and integration quality drew the most positive attention, and competitor comparison threads occasionally framed Semgrep favorably, though those threads were a small slice of overall volume. Divided opinion was clearest around dependency management, where commenters expressed frustration with tight version pinning on transitive dependencies while others appeared neutral or sympathetic to the maintenance rationale. Feature-bucketing and workflow classification decisions also drew split reactions, with some discussion questioning whether current categorization choices matched user mental models. Overall tone leaned skeptical, with reliability concerns outweighing positive signals across the window.