Solution for long term memory for agent coding CLIs and to facilitate handoff between different agent vendors
Rust
fix(auto-improve): keep eval policy server-owned
AkitaOnRails5ed815a5c6db
diff --git a/CHANGELOG.md b/CHANGELOG.mdindex 6b9f8d75..1eb4e308 100644--- a/CHANGELOG.md+++ b/CHANGELOG.md@@ -12,6 +12,10 @@ version in [`docs/changelog/`](docs/changelog/):## [Unreleased]### Fixed+- Made the executable auto-improvement eval policy server-owned. Manual CLI and+ HTTP reviews now use `[auto_improve.eval]` from the server process, and legacy+ requests that try to supply an `eval` object fail closed with an upgrade+ message.- Made managed cross-harness continuation survive quota exhaustion without avoluntary handoff. The startup packet now places its server-authored resumedirective outside the untrusted transcript fence, and refresheddiff --git a/crates/ai-memory-cli/src/commands/auto_improve.rs b/crates/ai-memory-cli/src/commands/auto_improve.rsindex 633a3f50..1389207c 100644--- a/crates/ai-memory-cli/src/commands/auto_improve.rs+++ b/crates/ai-memory-cli/src/commands/auto_improve.rs@@ -34,7 +34,6 @@ struct AutoImproveRequest {include_raw_fallback: bool,proposal_actor: String,pending_path: String,- eval: ai_memory_consolidate::AutoImproveEvalConfig,}#[derive(Debug, Deserialize, Serialize)]@@ -98,13 +97,6 @@ pub async fn run(config: &Config, args: AutoImproveArgs) -> Result<()> {include_raw_fallback: args.include_raw_fallback || settings.include_raw_fallback,proposal_actor: settings.proposal_actor.clone(),pending_path: settings.pending_path.clone(),- eval: ai_memory_consolidate::AutoImproveEvalConfig {- enabled: settings.eval.enabled,- command: settings.eval.command.clone(),- timeout_secs: settings.eval.timeout_secs,- targets: settings.eval.targets.clone(),- min_delta: settings.eval.min_delta,- },};let response: StageResponse = post_json(&endpoint, "/admin/auto-improve", &request).await?;diff --git a/crates/ai-memory-mcp/src/admin.rs b/crates/ai-memory-mcp/src/admin.rsindex 89436b27..df6f3453 100644--- a/crates/ai-memory-mcp/src/admin.rs+++ b/crates/ai-memory-mcp/src/admin.rs@@ -323,11 +323,10 @@ struct AutoImproveRequest {/// Pending proposal sidecar folder.#[serde(default = "default_auto_improve_pending_path")]pending_path: String,- /// Optional executable proposal eval gate. When omitted, direct admin- /// requests inherit the server's configured eval policy; CLI requests send- /// their local config explicitly.+ /// Removed compatibility field. Executable eval policy is server-owned;+ /// requests that still send this field fail closed.#[serde(default)]- eval: Option<ai_memory_consolidate::AutoImproveEvalConfig>,+ eval: Option<serde_json::Value>,/// Removed compatibility field. Requests that still send it fail closed so/// old preview callers cannot accidentally write pages.#[serde(default)]@@ -2384,6 +2383,14 @@ async fn handle_auto_improve(headers: HeaderMap,Json(req): Json<AutoImproveRequest>,) -> Result<impl IntoResponse, (StatusCode, Json<serde_json::Value>)> {+ if req.eval.is_some() {+ return Err((+ StatusCode::UNPROCESSABLE_ENTITY,+ Json(serde_json::json!({+ "error": "auto-improve eval request field was removed; configure [auto_improve.eval] on the server"+ })),+ ));+ }if req.dry_run.is_some() || req.stage.is_some() || req.mode.is_some() {return Err((StatusCode::UNPROCESSABLE_ENTITY,@@ -2430,10 +2437,7 @@ async fn handle_auto_improve(max_final_body_chars: req.max_final_body_chars,max_rule_page_tokens: req.max_rule_page_tokens,max_procedure_page_tokens: req.max_procedure_page_tokens,- eval: req- .eval- .clone()- .unwrap_or_else(|| state.auto_improve_review_config.eval.clone()),+ eval: state.auto_improve_review_config.eval.clone(),};let report =@@ -11711,7 +11715,7 @@ mod tests {}#[tokio::test]- async fn auto_improve_admin_omitted_eval_inherits_server_defaults() {+ async fn auto_improve_admin_refuses_request_eval_and_inherits_server_defaults() {let tmp = TempDir::new().unwrap();let store = Store::open(tmp.path()).unwrap();let wiki = Wiki::new(tmp.path(), store.writer.clone())@@ -11771,6 +11775,44 @@ mod tests {};let router = admin_router(state);+ let rejected = router+ .clone()+ .oneshot(+ Request::builder()+ .method("POST")+ .uri("/admin/auto-improve")+ .header("content-type", "application/json")+ .body(Body::from(+ serde_json::to_vec(&serde_json::json!({+ "workspace": "default",+ "project": "scratch",+ "session_id": session_id.to_string(),+ "min_observations": 1,+ "min_session_duration_secs": 0,+ "min_confidence": 0.75,+ "max_proposals_per_run": 5,+ "eval": {+ "enabled": true,+ "command": tmp.path().join("request-controlled-program").display().to_string(),+ "timeout_secs": 1,+ "targets": ["notes"],+ "min_delta": 0.0+ }+ }))+ .unwrap(),+ ))+ .unwrap(),+ )+ .await+ .unwrap();+ assert_eq!(rejected.status(), StatusCode::UNPROCESSABLE_ENTITY);+ let body = to_bytes(rejected.into_body(), usize::MAX).await.unwrap();+ let json: serde_json::Value = serde_json::from_slice(&body).unwrap();+ assert_eq!(+ json["error"],+ "auto-improve eval request field was removed; configure [auto_improve.eval] on the server"+ );+let resp = router.oneshot(Request::builder()diff --git a/docs/auto-improve-eval-gates.md b/docs/auto-improve-eval-gates.mdindex cf73c612..5609cc56 100644--- a/docs/auto-improve-eval-gates.md+++ b/docs/auto-improve-eval-gates.md@@ -19,6 +19,12 @@ The command is split on whitespace and executed directly, not through a shell.Use a wrapper script if you need quoting, environment setup, or multiplecommands.+The eval policy is owned by the server process. `ai-memory auto-improve` and+direct HTTP requests cannot replace the configured command, targets, timeout,+or score threshold. When a CLI talks to a remote server, configure+`[auto_improve.eval]` in that server's config; older clients that send an+`eval` request field receive `422 Unprocessable Entity` and must be upgraded.+## Request contractThe scorer receives one JSON object on stdin:diff --git a/docs/auto-improvement-loop.md b/docs/auto-improvement-loop.mdindex 465ef1be..aa8d14ce 100644--- a/docs/auto-improvement-loop.md+++ b/docs/auto-improvement-loop.md@@ -235,7 +235,8 @@ timeouts, invalid JSON, `passed = false`, missing `passed`, or score deltas belo`min_delta` reject only that targeted proposal. Non-targeted proposals bypass thegate. If all proposals fail eval, the run is still staged with zero proposals andthe rejected candidates so the rejection buffer can remember the failed attempt.-Hooks never run the eval command. See+The server owns this executable policy: neither the CLI nor a direct admin+request can replace it. Hooks never run the eval command. See[`auto-improve-eval-gates.md`](auto-improve-eval-gates.md) for the full stdin /stdout contract and example scorer scripts.diff --git a/docs/security-boundaries.md b/docs/security-boundaries.mdindex 1782316a..512c4114 100644--- a/docs/security-boundaries.md+++ b/docs/security-boundaries.md@@ -65,6 +65,7 @@ means the boundary is not built yet.| 7b | New public page metadata is bounded and cannot supply scope, author or admission controls; admin kind retains its legacy adapter | `ai-memory-core/src/page.rs` `PageWriteMetadata::into_frontmatter`; MCP validates all four fields; admin consumes legacy kind explicitly and shares entities/abstract/relations validation before scope creation; both use `Wiki::write_page` for sanitization/admission/authenticated attribution | `admin_write_page.rs` `write_page_metadata_rejects_invalid_payload_before_scope_creation` (invalid shapes, unknown relation kinds, raw-input over-limit values, untrimmed relation scopes and escaping paths refused; legitimate control admitted), `write_page_metadata_preserves_admin_auth_and_sanitization` (anonymous/DB-user refused, root control admitted with scrubbed metadata and authenticated attribution); `server.rs` `memory_write_page_metadata_rejects_invalid_before_scope_creation`, `memory_write_page_metadata_preserves_scope_author_and_sanitization` (foreign user denied; writer control admitted, forged frontmatter/author ignored, secrets scrubbed, invalid rewrite leaves disk and SQL latest unchanged); core `page_write_metadata_scope_components_refuse_surrounding_whitespace` and `page_write_metadata_limits_apply_before_trimming_and_normalization`; admin `write_page_metadata_admin_legacy_kind_matches_base` compares long/control-character kind behavior with the base adapter; entity-bound refusal bite-checked with its validation removed | STRONG || 7c | A compaction checkpoint cannot bypass `consolidate` admission by deferring its provider rewrite (#1229) | `ai-memory-hooks/src/router.rs` `write_checkpoint_page` gives the PreCompact / PostCompaction rule-based write the `consolidate` admission operation whenever a consolidator is configured, and `process_authorized` queues the durable rewrite only after that write succeeds; the worker's rewrite runs admission again through `Consolidator::consolidate_session` | `router.rs` `checkpoint_admission_refusal_writes_no_page_and_queues_no_rewrite` (a policy refusing `consolidate` leaves no page and no queued job; control: a policy gating only `handoff_begin` writes the page and queues the job), `checkpoint_preserves_the_consolidate_admission_operation` (the hook path sends exactly one `consolidate` admission call and never calls the provider); both bite-checked with the admission context removed | STRONG || 7d | The Cursor Agent CLI child, which reads captured session text, never sees the server's own credentials | `ai-memory-llm/src/cursor_agent.rs` `complete_text` spawns `agent` with `env_clear()` plus `CursorAuth::child_env`, which `CursorAuth::new` filters through `cursor_child_env` (an allowlist of `PATH`/`HOME`/user, locale, temp/XDG, session bus, proxy/CA and Windows basics, plus `LC_*`/`CURSOR_*`); the environment and the workspace root are captured once by `ai-memory-cli/src/config.rs` `Config::cursor_auth`, so the provider reads no environment itself; a relative workspace root is refused | `cursor_provider.rs` `the_child_gets_only_the_allowlisted_environment` (`AI_MEMORY_AUTH_TOKEN` and `OPENAI_API_KEY` canaries and the test process's own `CARGO_*` variables never reach the child; control: `PATH`, `HOME` and `CURSOR_API_KEY` do, and the workspace sits under the injected root; bite-checked by dropping `env_clear()`), `a_relative_workspace_root_is_refused`; `config.rs` `cursor_provider_hands_the_child_only_allowlisted_env` | STRONG |+| 7e | Executable auto-improvement eval policy is server-owned | `ai-memory-mcp/src/admin.rs` retains the legacy request `eval` key only as a rejected compatibility field and builds `AutoImproveReviewConfig.eval` exclusively from `AdminState`; `ai-memory-cli/src/commands/auto_improve.rs` sends no executable policy over HTTP | MCP `auto_improve_admin_refuses_request_eval_and_inherits_server_defaults` sends a request-controlled program path and receives 422, then omits the field and proves the configured server gate still runs | STRONG || 8a | Messaging: recipient-only visibility | `ai-memory-store/src/ops.rs` pop target-select + `reader.rs` `list_messages` (`to_*` predicate) | `agent_messages.rs` `a_message_is_only_visible_to_its_recipient`; `agent_messages_tools.rs` non-recipient cannot pop | STRONG || 8b | Messaging: cancel-own-only | `ai-memory-store/src/ops.rs` `cancel_messages` (AND-gated on `from_*` sender coordinate) | `agent_messages.rs` — whole-outbox **and** a foreign **specific-id** cancel refused | STRONG || 8c | Messaging: pop-exactly-once | `ai-memory-store/src/ops.rs` `pop_message_in_transaction` atomic CAS `WHERE state='pending'` | `agent_messages.rs` — sequential **and** `tokio::join!` concurrent double-pop yields exactly one `Some` | STRONG |