@tangle-network/agent-runtime 0.104.0 → 0.105.0

package/dist/{worktree-fanout-WsOzt8qu.d.ts → worktree-fanout-DxhAWr5Z.d.ts}

@@ -3,8 +3,8 @@ import { AnalystFinding, DefaultVerdict } from '@tangle-network/agent-eval';
 import { N as NodeId, a as Settled, g as Spend, h as AgentSpec, S as Scope, c as SpawnJournal, b as ResultBlobStore, i as ExecutorRegistry, B as Budget, A as Agent, j as RootHandle, e as SupervisedResult, f as Executor } from './types-CvNwMzQt.js';
 import { I as Iteration, R as RuntimeHooks } from './types-BCoemcXU.js';
 import { BackendType } from '@tangle-network/sandbox';
-import { W as WorktreeHarnessResult, G as GitRunner, a as WorktreeCheckRunner, D as DeliverableSpec } from './completion-gate-Bv47cOnn.js';
-import { L as LocalHarness, r as runLocalHarness } from './local-harness-CtK9dSny.js';
+import { W as WorktreeHarnessResult, G as GitRunner, a as WorktreeCheckRunner, D as DeliverableSpec } from './completion-gate-BMy5LGoP.js';
+import { L as LocalHarness, r as runLocalHarness } from './local-harness-BDNcl6jI.js';
 
 /**
  *

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@tangle-network/agent-runtime",
-  "version": "0.104.0",
+  "version": "0.105.0",
   "description": "Shared task-lifecycle skeleton for agents: a recursive loop kernel for chat turns, one-shot tasks, and multi-attempt loops, with trace capture and eval-gated self-improvement. Domain behavior lives in adapters; scoring and ship-gates in @tangle-network/agent-eval.",
   "homepage": "https://github.com/tangle-network/agent-runtime#readme",
   "repository": {
@@ -114,7 +114,8 @@
     "verify:bench": "pnpm build && pnpm --filter @tangle-network/agent-bench run typecheck:public && pnpm --filter @tangle-network/agent-bench test && pnpm --filter @tangle-network/agent-bench run verify:package:local-runtime",
     "verify:bench:published": "pnpm build && pnpm --filter @tangle-network/agent-bench run typecheck:public && pnpm --filter @tangle-network/agent-bench test && pnpm --filter @tangle-network/agent-bench run verify:package",
     "verify:package": "pnpm run check:testing-fixture && pnpm run check:skills && node scripts/verify-package-exports.mjs",
-    "verify:primeintellect": "pnpm build && node scripts/verify-primeintellect-v1.mjs",
+    "verify:official-optimizers": "node scripts/verify-official-optimizers.mjs",
+    "verify:primeintellect": "pnpm build && node scripts/verify-primeintellect.mjs",
     "verify:primeintellect:live": "pnpm build && node scripts/verify-primeintellect-live.mjs",
     "docs:api": "pnpm run build && typedoc && node scripts/gen-primitive-catalog.mjs",
     "docs:freshness": "node scripts/check-docs-freshness.mjs",
@@ -122,7 +123,7 @@
   },
   "devDependencies": {
     "@biomejs/biome": "^2.4.15",
-    "@tangle-network/agent-eval": "0.125.0",
+    "@tangle-network/agent-eval": "0.126.6",
     "@tangle-network/agent-interface": "0.32.0",
     "@tangle-network/sandbox": "^0.12.0",
     "@types/node": "^25.9.3",
@@ -154,7 +155,7 @@
   "license": "MIT",
   "packageManager": "pnpm@10.28.0",
   "peerDependencies": {
-    "@tangle-network/agent-eval": ">=0.122.8 <0.126.0",
+    "@tangle-network/agent-eval": ">=0.126.6 <0.127.0",
     "@tangle-network/agent-interface": ">=0.32.0 <0.33.0",
     "@tangle-network/sandbox": ">=0.12.0 <1.0.0",
     "playwright": "^1.40.0"
@@ -168,7 +169,7 @@
     }
   },
   "dependencies": {
-    "@tangle-network/agent-knowledge": "^4.1.0",
+    "@tangle-network/agent-knowledge": "5.0.1",
     "@tangle-network/agent-profile-materialize": "0.5.1",
     "tar-stream": "3.2.0"
   }

package/skills/build-with-agent-runtime/SKILL.md

@@ -34,7 +34,7 @@ Do not move product storage transactions into a provider-neutral package.
 |---|---|
 | One product chat turn | `handleChatTurn(...)` |
 | One normalized streamed agent turn | `streamAgentTurn(...)` and `collectAgentTurn(...)` |
-| One task or multi-turn loop | `runAgentTask(...)`, `runAgentTaskStream(...)`, or `runLoop(...)` |
+| One task or multi-turn loop | `runAgentTask(...)`, `runAgentTaskStream(...)`, or `runAgentRounds(...)` |
 | Supervisor and workers | `supervise(...)` or `superviseSurface(...)` |
 | Parallel work with a shared budget | `fanout(...)` |
 | Fixed composition | `pipeline(...)`, `panel(...)`, or `verify(...)` |
@@ -50,11 +50,16 @@ Do not move product storage transactions into a provider-neutral package.
 
 ## Improvement flow
 
-`improve(profile, findings, options)` searches one surface and returns a detached winner.
+`improve(profile, options)` searches one surface and returns a detached winner.
 It never changes a profile, document, repository, memory store, or knowledge base.
 
-Supported profile surfaces are prompt, one named inline skill, tools, MCP, hooks, subagents, whole profile, and curated memory in `profile.resources.instructions`.
-Code uses isolated worktrees and returns a sealed patch candidate.
+For a profile field, pass one complete agent-eval `OptimizationMethod`, explicit train, selection, and final-test partitions, judges, and the candidate execution function.
+Use `officialGepa(...)` with an explicit recipe when upstream GEPA should own search.
+Use `officialSkillOpt(...)` when Microsoft's SkillOpt should own search.
+Both require `evaluationId`; change it whenever dispatch, judges, models, or scoring behavior changes.
+Resumable runs accept `never`, `if-compatible`, or `required` and reuse state only when agent-eval derives the same run identity.
+Runtime has no local prompt, skill, memory, or profile optimizer fallback.
+Code uses Runtime's isolated worktrees and returns a sealed patch candidate.
 Knowledge uses `runKnowledgeImprovementJob(...)` and returns paired snapshots.
 
 Use `proposeAgentImprovement(...)` for a production proposal.
@@ -80,12 +85,12 @@ Never treat a lost response as a failed write without reconciling it.
 
 ## Surface rules
 
-- Prompt changes `profile.prompt` only.
-- Skill optimization selects one inline skill by `skills.resourceName`; profile resources must fail closed.
+- Prompt changes `profile.prompt` only and requires a complete method.
+- Skill optimization selects one inline skill by `skills.resourceName`, requires a complete method, and requires profile resources to fail closed.
 - Curated memory changes `profile.resources.instructions`; retrieval stores and memory databases belong in the knowledge flow.
-- Tools, MCP, hooks, subagents, and whole-profile changes require an explicit proposer because Runtime cannot invent domain capabilities safely.
+- Tools, MCP, hooks, subagents, curated memory, rollout policy, and whole-profile changes require a complete method.
 - Code candidates must come from the Runtime worktree path so patch identity and cleanup stay intact.
-- Workflow and policy files are code surfaces; parameter sweeps come from agent-eval.
+- Workflow files are code surfaces. Parameter sweeps belong in a complete agent-eval method.
 - Knowledge candidates remain detached until the shared activation path applies or restores their frozen snapshots.
 
 ## Product integration
@@ -104,7 +109,8 @@ The product must not recreate candidate hashing, paired comparison, confidence i
 ## Do not duplicate
 
 - Do not write a provider-specific profile wrapper; extend `AgentProfile` and its materializer.
-- Do not write a second optimizer loop; compose Eval proposers through `improve(...)`.
+- Do not write a second optimizer loop; pass a complete agent-eval method to `improve(...)`.
+- Do not use Runtime's code generator to approximate GEPA, SkillOpt, or another upstream profile optimizer.
 - Do not write a second candidate catalog; persist the immutable proposal records.
 - Do not let an analyst or adapter commit, push, open a pull request, or edit a live store.
 - Do not hand-roll SSE parsing, usage totals, profile matrices, bootstrap statistics, sandbox acquisition, or worktree cleanup.

package/skills/loop-writer/SKILL.md

@@ -21,7 +21,7 @@ Do not copy signatures from this skill.
 | Parallel repository workers with isolated branches | `worktreeFanout(...)` |
 | Repeated work in a graded tool environment | `runAgentic(...)` |
 | Equal-budget comparison over that environment | `runBenchmark(...)` |
-| Low-level round policy with custom planning and stopping | `runLoop(...)` |
+| Low-level round policy with custom planning and stopping | `runAgentRounds(...)` |
 
 If an existing row fits, use it and stop.
 Do not create another wrapper solely to rename inputs or results.