{"version":"2.1.307","anchor":"malformed-tool-use-retry-is-now-a-count","canonical_anchor":"malformed-tool-use-retry-is-now-a-count","heading":"Claude Code can retry a badly formed tool call more than once","tier":"notice","area":"Elsewhere","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.307\/e\/malformed-tool-use-retry-is-now-a-count","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.307","markdown":"### Claude Code can retry a badly formed tool call more than once\n\nWhen the model sends a badly formed tool call, Claude Code now counts its retries instead of allowing just one\n\n**What**\n\nA tool call is a request from the model to run one of Claude Code's tools, such as reading a file. When the model sends one that is badly formed, Claude Code tries again. It used to track only whether it had already retried. It now keeps a count of retries, starting at zero.\n\n**Why**\n\nThis makes room for more than one retry when a tool call is badly formed, so a single bad attempt is less likely to end the work.\n\n- Area: Elsewhere\n- Tier: You'll notice\n- Useful: 2\/5\n- Signal: 1\/5\n- Scope: individual\n- Heads-up: no"}