Conversation
Deferred tools kept every unused schema out of the pinned prefix, but a deferred tool's first call never ran: it returned the schema and told the model to retry, so ordinary investigation (list_dir, grep_files, Web fetch, Run) spent a turn plus a tool-surface re-pin before doing anything. Children and Workflow leaves also started with none of the tools their assignment named explicitly. - Parent and child: a first call that already matches the unseen schema (object, every required field, no undeclared field) executes and activates at the catalog tail. Malformed calls still get the schema. Hydration runs after every authority gate (deny/allow lists, Plan, repo law, approval), and dispatch still requires a registry spec, so this changes turn cost, not authority. The executed path emits a `tool.deferred_first_use_executed` audit record. - Children warm their explicit `allowed_tools` onto the first request through the existing bounded activation cache (8 names / 16 KiB). Visibility only: grants, role posture and envelopes are unchanged. Known limitation: argument types are left to each tool's own validation. Evidence: 60 passed, 0 failed (13,192 skipped) across deferred, small_surface, tool_catalog, hydrat and tool_search selections, including a Scout whose well-formed first list_dir runs, whose malformed call gets the schema, and whose write/edit stay refused. Two tests that encoded the old never-execute behavior now use malformed input to keep covering the schema-hint path. TUI all-target/all-feature Clippy with CI flags and fmt passed. Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
This branch has not been deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
No-Issue: 0.10.1 release-readiness recovery and first-call tool work, found in dogfooding; no public issue tracks it.
Deferred tools kept every unused schema out of the pinned prefix, but a
deferred tool's first call never ran: it returned the schema and told
the model to retry, so ordinary investigation (list_dir, grep_files,
Web fetch, Run) spent a turn plus a tool-surface re-pin before doing
anything. Children and Workflow leaves also started with none of the
tools their assignment named explicitly.
(object, every required field, no undeclared field) executes and
activates at the catalog tail. Malformed calls still get the schema.
Hydration runs after every authority gate (deny/allow lists, Plan,
repo law, approval), and dispatch still requires a registry spec, so
this changes turn cost, not authority. The executed path emits a
tool.deferred_first_use_executedaudit record.allowed_toolsonto the first requestthrough the existing bounded activation cache (8 names / 16 KiB).
Visibility only: grants, role posture and envelopes are unchanged.
Known limitation: argument types are left to each tool's own validation.
Evidence: 60 passed, 0 failed (13,192 skipped) across deferred,
small_surface, tool_catalog, hydrat and tool_search selections,
including a Scout whose well-formed first list_dir runs, whose
malformed call gets the schema, and whose write/edit stay refused. Two
tests that encoded the old never-execute behavior now use malformed
input to keep covering the schema-hint path. TUI all-target/all-feature
Clippy with CI flags and fmt passed.
🤖 Generated with Claude Code