Commit Graph

14 Commits

Author SHA1 Message Date
Peter Steinberger
c903ca8158 feat(agents): default code mode to per-model auto activation (#114906) 2026-07-28 03:00:22 -04:00
Peter Steinberger
cf4cb0ac85 feat(agents): report per-run stats (code-mode engagement, round trips, cost) in agent JSON envelopes (#114688)
* feat(agents): add per-run stats to embedded agent run meta

Adds codeModeEngaged, assistantTurns, bridgeCalls, and costUsd to
EmbeddedAgentRunMeta.agentMeta and mirrors them on the agent exec --json
envelope. Code-mode engagement is stamped from the tool-surface truth,
round trips accumulate across attempts beside usage, bridge counts come
from the run's tool-search catalog counters, and cost reuses the shared
model pricing helpers (cache tiers included, omitted without cost data).

* fix(agents): accumulate bridge call counts across run attempts

Attempt cleanup clears the per-attempt tool-search catalog, so retries and
fallbacks discarded earlier bridge counts. Fold each attempt's bridgeCalls
into the run accumulator beside assistantTurns and stamp the cumulative
totals into agentMeta, matching the documented per-run contract.
2026-07-27 23:48:01 -04:00
Peter Steinberger
ea707382fd feat(code-mode): first-class nodes API (#114877)
* feat(agents): add nodes to code mode

* fix(agents): add code mode nodes discovery hint
2026-07-27 23:03:23 -04:00
Peter Steinberger
7394262cd1 feat: per-model code-mode capability flags with an "auto" master-switch tier (#114695)
* feat(tools): per-model code-mode capability flags and auto master-switch tier

* feat(anthropic): flag claude-sonnet-4-6 as code-mode preferred

* revert(anthropic): drop claude-sonnet-4-6 from the code-mode preferred set

* test(anthropic): type the manifest compat field in the catalog contract test

* fix(anthropic): carry catalog compat onto hand-built forward-compat model rows
2026-07-27 16:47:41 -04:00
Sasan
7599bbe636 fix(agents): stop code-mode shell-command retry storms (#113070)
Reject recognizable POSIX and Windows shell source before QuickJS execution while preserving valid JavaScript, TypeScript, syntax errors, standard globals, and real hoisted bindings. Add real-worker regression coverage and adversarial cross-platform stress proof. Fixes #113069.
2026-07-27 08:10:37 -04:00
Peter Steinberger
55d66fbf98 chore(scripts): remove resolved investigation tools and orphans (#113532)
* chore(scripts): remove orphaned wrappers

* chore(scripts): remove resolved investigation tools

* refactor(code-mode): remove orphaned plugin namespaces

* test(code-mode): remove stale namespace import
2026-07-25 02:47:40 -07:00
Peter Steinberger
46b9a3e12d docs: Swarm user guide (#110325) (#111384) 2026-07-19 05:33:00 -07:00
Peter Steinberger
746d257f05 feat: contract sessions-family tool outputs (#110424)
* feat(agents): contract sessions-family tool outputs

* refactor(agents): remove unused fast-mode exports
2026-07-18 06:43:12 +01:00
Peter Steinberger
b9c5c7c9fc feat(agents): contract filesystem tools for code mode (#110395) 2026-07-18 06:26:04 +01:00
Peter Steinberger
19296b003b improve(agents): normalized web_search output contract with boundary-owned wrapping (#110308)
* feat: normalize web search output contract

* refactor(agents): promote web_search contract from a dedicated output module

* fix(agents): wrap unwrapped web_search text and pass unknown provider payloads through as raw

* fix(agents): gate web_search results branch on conforming rows and cover metadata fields

* fix(agents): gate every provider text path at the web_search boundary

* test(agents): align web_search fixtures with boundary-owned wrapping

* fix(agents): make the web_search boundary own the untrusted-content envelope

* fix(agents): report declared web_search errors first and align the documented contract

* fix(agents): bind envelope stripping to real markers and densify result rows

* fix(agents): emit canonical urls and a closed error code from web_search

* fix(agents): keep structured provider error diagnostics in the wrapped message

* fix(agents): satisfy production export and lint gates for the web_search contract
2026-07-18 03:39:29 +01:00
Peter Steinberger
53b53f1977 improve(agents): exact web_fetch output contract with cleaned result shape (#110223)
* feat: add web fetch output contract

* fix(agents): keep web_fetch contract schema module-local

* docs: refresh docs map for the web_fetch result section
2026-07-18 03:28:29 +01:00
Peter Steinberger
d371ea1f01 perf(agents): wave-1 tool output contracts for code mode (#110215)
* feat(agents): keep output contracts complete across opaque schema leaves

* test(agents): add code mode schema-hint hot-path micro-benchmark

* test(agents): count raw-first inspection execs per tool in live bench

* feat(agents): expand compact output contract hints

* feat(agents): declare wave one tool output contracts

* docs(tools): list built-in output contracts

* fix(agents): bound union scans before literal-union analysis

* fix(agents): keep wave-1 contract schemas module-local and type-exact
2026-07-18 01:02:18 +01:00
Peter Steinberger
3d1e8f9475 perf(agents): compose conversation tools in code mode (#110098) 2026-07-17 20:54:32 +01:00
Peter Steinberger
b319493a52 improve: let Code Mode use declared tool result shapes (#109813)
* feat(agents): add Code Mode output contracts

* perf(agents): defer untrusted tool schemas

* chore(plugin-sdk): refresh API baseline

* fix(agents): preserve deferred schema markers

* fix(agents): preserve policy-blocked tool results

* fix(agents): preserve nullable output contracts

* fix(agents): reject policy blocks from success contracts

* fix(agents): preserve output contracts through normalization

* test(agents): type normalized tool fixtures

* fix(agents): validate catalog results before projection

* fix(agents): keep result snapshots internal

* fix(agents): fail closed on unrenderable unions

* fix(agents): preserve empty result markers

* test(agents): keep result fixture discriminator literal

* test(agents): use native own-property check
2026-07-17 14:46:37 +01:00