{"id":"62f4efa5-c90a-4e7e-b3de-8568dc06d6bb","arxiv_id":"2607.23942","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Coding each mechanism for evidence of lineage and implementation depth, the review closes one candidate gap (GraSP) and isolates five residual control bundles for language agents.","lead":"This review compares ten classic cognitive architectures with dozens of modern language-agent systems at the level of control mechanisms rather than feature labels. It identifies five untested 'bundles' of coupled behaviors (memory, failure recovery, workspace competition, commitment, uncertainty-resource control) as the next open research targets.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The five-bundle partition rests on provisional single-coder E/D codes and an absence search with no recall estimate; a re-coding or a missed system could close/open a bundle, and B5's closure depends on one unverified preprint.","rationale":"The reader's weakest assumption — that the five-bundle partition depends on provisional single-coder E/D codes — is exactly the load-bearing point. I agree with it and extend it: the same provisionality applies to the absence inference that closes B5, and the GraSP closure is a single-preprint judgment not backed by independent implementation inspection in the paper. However, the paper itself repeatedly discloses these limits, presents the residual bundles as falsifiable hypotheses rather than established findings, and gives concrete re-coding and experimental protocols. Therefore the correct verdict remains CONDITIONAL, not ACCEPT or REJECT. The central scientific contribution — a mechanism-level framework and an auditable agenda — is not undermined, but the specific 'five residual bundles' list should be treated as provisional until independent adjudication and GraSP verification are performed.","tokens_in":30235,"tokens_out":5425,"duration_ms":61512,"concrete_test":"Run an independent re-coding of all Appendix C D3/D4 packets by a second coder blind to provisional codes, then recompute the residual bundles with the adjudicated ledger. In the same pass, inspect GraSP's pinned repository and verify the four claimed edges (calibrated routing, typed verification, bounded repair, replanning/ReAct fallback). If any D3/D4 code changes or GraSP lacks a claimed edge, the five-bundle partition fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim — that five residual bundles remain — is an absence claim. It holds only if (a) the D3/D4 codes that separate 'implemented' from 'residual' are correct, and (b) no system outside the frozen corpus already composes a bundle. Both links are currently conditional. Section II-E and Appendix C state that scores are single-coder and provisional until disagreements are adjudicated; a shift from D3 to D4 for one memory, failure, workspace, commitment, or uncertainty system changes the migration ledger and can close a bundle exactly as GraSP closed B5. Section II-B/Appendix F provide replayable query templates but no recall estimate, so absence is bounded by the search's recall, not established. The B5 closure itself is a single-source inference from GraSP [63]; the paper does not report inspecting its repository for the typed DAG verification, bounded repair, and replanning/ReAct fallback edges. If any claimed edge is not actually implemented, the residual count reverts to six. The paper discloses all this as validity limits, but the abstract and Section VI present the five bundles as the headline result, making the provisionality load-bearing rather than peripheral.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a mechanism-level review connecting ten historical cognitive architectures, eight language-agent runtime families, and forty-two mechanism-focused modern systems. Each mechanism is reconstructed via a seven-field tuple (state, control, transition, persistence, failure, learning, resource), and correspondences are coded on two independent axes: evidence relation (E1–E4) and migration depth (D0–D4). The central finding is that modern agents have operationalized many individual control mechanisms, while five residual control bundles remain open: activation–latency–utility (B1), typed impasse–substate–compilation (B2), content competition–workspace–broadcast learning (B3), intention–reconsideration–method authority (B4), and uncertainty–resources–interruption–stopping (B6). A sixth candidate, B5 (skill governance), is reported as closed by GraSP. The paper also contributes a catalog, an auditable coding framework, and preregistration-ready intervention protocols. It explicitly disclaims independent inter-rater reliability, systematic recall estimates, and completed experiments, and repeatedly states that the five-bundle partition is conditional on the current provisional codes.","tokens_in":30523,"tokens_out":4213,"duration_ms":46320,"significance":"If the residual-bundle claim holds, this is a substantial contribution: it moves the comparison of cognitive architectures and language agents from feature labels to control semantics and provides a falsifiable, corpus-relative research agenda. The paper's methodological strengths are real and should be credited: E/D codes are separated, coding rules are operationalized in Table II, runtime evidence is version-pinned in Appendix E, query templates are documented in Appendix F, and the intervention protocols in Appendix D specify ablations and rejection conditions. The closed B5 case is a useful demonstration that the framework can eliminate, not merely generate, research gaps. The main risk is not internal inconsistency but the gap between the headline 'five residual bundles remain' and the evidence base: the codes are provisional and single-coder, and the absence claims are bounded by an unmeasured search recall.","major_comments":[{"comment":"The central result 'Five residual bundles remain' is generated from the E/D codes, but Section II-E states the manuscript 'does not claim independent inter-rater reliability' and Appendix C says scores 'must remain provisional until disagreements are recorded and adjudicated.' Every D3/D4 row in Table IV and Fig. 6 is single-coder. A shift in one D3/D4 grading can close or open a bundle, as the paper itself notes for GraSP/B5. Because the abstract and Section VI present the five-bundle partition as the headline outcome rather than as a provisional output of one coder, this is not a peripheral validity caveat. Please either complete the promised second-author recoding and report disagreements, or rephrase the headline and Section VI throughout as conditional on the current provisional codes.","section":"§II-E, Appendix C, Abstract"},{"comment":"The closure of B5 rests entirely on one source, GraSP [63], described in Appendix C/O as combining calibrated multi-skill routing, typed DAG verification, bounded repair, and replanning/ReAct fallback. The review's own protocol (§II-C) requires official implementation evidence when a mechanism depends on executable behavior, and Appendix E provides version-pinned repository evidence for runtime families but not for GraSP. Without inspecting the GraSP release or otherwise verifying the four claimed edges against the paper's artifacts, the statement that 'B5 is therefore an implemented E3/D4 convergence case' is a single-source inference. Since B5's closure changes the residual count from six to five, this point is load-bearing. To fix: inspect and pin the GraSP implementation, or mark B5 as 'reported but not independently verified.'","section":"§VI-B, §VI-C(e), Appendix C R7e"},{"comment":"The paper states that result counts were not frozen and that no search-recall estimate is made. The five residual bundles are absence claims: they assert that no reviewed system composes the edges. Without a recall estimate for the six query blocks in Table XIV, the abstract's 'Five residual bundles remain' overstates what the search can certify relative to the broader literature. The paper already says it makes no PRISMA completeness claim, so the fix is largely presentation-level if the claims are consistently limited to 'within the frozen corpus and the documented search.' But the abstract and Section VI-C state the bundles unconditionally. Please align the wording throughout with the actual evidence boundary.","section":"§II-B, Appendix F"}],"minor_comments":[{"comment":"The name 'Voyager' appears as 'V oyager' in several places, apparently a rendering artifact. Please fix.","section":"Throughout (e.g., §IV-B, Fig. 7)"},{"comment":"The row 'This review' marks 'Primary' in all five columns. Consider a heading or note clarifying that these labels are self-assessments of analytical focus, not independent evaluations.","section":"Table I"},{"comment":"The paper says a dated research log supplies replayable query templates and record-level dispositions, but Appendix F only summarizes the blocks. If space permits, include the template list or a stable link, since the auditability claim depends on access to those templates.","section":"Appendix F"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is unusually transparent about its own limitations, and the protocol design is strong. My recommendation is driven by the gap between the headline five-bundle result and the evidence base: the codes are provisional and single-coder, the B5 closure is a single-source inference, and the absence claims lack a recall estimate. None of these issues appears unfixable within the manuscript's scope; the authors' own appendices already promise the missing recoding and protocol details. I would not require completed experiments for acceptance of a review, but the central claim should either be independently coded or be presented as conditional in all headline statements."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The one thing you should know: this is a useful survey, and the field would be slightly better off if its coding vocabulary became common. The E1–E4 and D0–D4 axes, the seven-field mechanism tuple, and the merge/split residual-extraction rules are not present in the five comparator reviews it positions against. That is a real contribution. The catalog of historical mechanisms and the modern systems that approximate them is careful, and the paper is unusually honest about its own evidentiary limits: it states flat out that no inter-rater reliability has been established, that its corpus has no recall estimate, and that no experiments have been run. The GraSP closure is a nice touch — an absence claim being falsified by a closest-baseline screening shows the framework can do more than generate wish lists.\n\nThe soft spots are exactly where the reader put them, and the stress-test note lands. The five residual bundles are the headline result in the abstract and conclusion, but they rest on provisional single-coder D3/D4 codes. If another coder recodes a memory or failure system differently, the migration ledger changes and a bundle opens or closes. The absence of any bundle outside the frozen corpus is an absence claim bounded by the search, and the search recall is not estimated. And the B5 closure — the one concrete win — depends on reading GraSP's paper rather than inspecting its repository. That is a lot of load-bearing scaffolding under a short list of open problems. The paper discloses all of this in Section VII and Appendices C and F, so it is a confidence problem more than an integrity problem. But the abstract and conclusion do not carry that hedging, and the reader who stops at the abstract will over-read the result.\n\nIs the paper worth a serious referee? Yes. It is a well-structured, reproducible-framework review that should be sent out despite the provisionality. The right outcome of peer review is not rejection but a demand for independent recoding and for softening the headline claims — or, better, a companion note reporting the second-author recoding, which the paper already plans. I would not cite it for the five bundles as established gaps; I would cite it for the coding axes and the residual-extraction method.\n\nTake it to reading group if your group likes discussing review methodology. Otherwise, it is worth a skim plus Appendix C if you are building agent runtimes and want a checklist of couplings to test.","headline":"A genuinely reusable synthesis framework whose headline map of five residual bundles is a provisional research agenda, not an established finding — and the paper mostly knows this.","tokens_in":31037,"tokens_out":1986,"would_cite":true,"duration_ms":25835,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This review argues that modern language agents have already migrated most individual control mechanisms from classical cognitive architectures, and that the open research target is five residual couplings — runtime invariants connecting sta","keywords":["mechanism-level review","cognitive architectures","language agents","migration depth","runtime invariants","residual couplings","control semantics","evidence coding"],"falsifier":"Have a second coder, blind to the provisional codes, re-classify the D3/D4 evidence packets the paper provides: if any reviewed system already composes one of the five missing couplings, that bundle closes, just as the skill-governance candidate closed. Alternatively, run the paper's own preregistered protocol for the memory bundle against a recency-based baseline; if the simpler selector is non-inferior on recall and overhead, the bundle dissolves empirically.","tokens_in":30086,"feed_emoji":"🧩","tokens_out":5535,"duration_ms":59469,"temperature":0.7,"pith_summary":"The paper's bounded synthesis claim: modern language agents have operationalized most of the cognitive substrate — adaptive memory policies, typed failure recovery, dynamic team selection, workflow search, skill induction, resource scheduling, and uncertainty-conditioned action — but often by independent convergence rather than documented inheritance. What remains are not missing mechanisms but missing couplings: the paper derives, via a seven-field mechanism representation, an evidence-depth coding, and explicit merge/split rules, exactly five residual bundles worth testing as composable runtime invariants, plus one candidate it closes to an existing system. A sympathetic reader should care because this reframes cognitive-architecture research as a source of precise, falsifiable control hypotheses rather than feature lists.","feed_headline":"Five control couplings still missing from agent runtimes","feed_subtitle":"A mechanism-level ledger shows adaptive memory, failure recovery, and skill governance exist — the gap is composing them into one runtime la","key_machinery":"The central object is the mechanism tuple K = ⟨S, C, T, P, F, L, R⟩ — state substrate, control locus, transition trigger, persistence boundary, failure semantics, learning operator, and resource/governance policy — used to reconstruct every historical and modern system, plus two independent coding axes: E1–E4 for whether a correspondence is documented lineage, structural migration, functional approximation, or convergence, and D0–D4 for how deeply the mapped mechanism is implemented, from a conceptual label to an experience-adapted control law. Residual extraction merges atomic missing invariants only when they share an insertion boundary and one supplies state or authority for the other; cl","core_discovery":"On the paper's own terms, the discovery is uneven migration depth: for each of ten historical cognitive architectures, the strongest reviewed modern system achieves D3 or D4 on the individual control edge (autonomous trigger plus outcome-dependent learning), yet no reviewed system composes the historical couplings — and the one composition candidate that did emerge, skill governance, is already implemented by a system combining calibrated routing, typed verification, bounded repair, and replanning/fallback. The five residual bundles are therefore not feature requests; they are specific state-trigger-authority-transition-learning packets, each with insertion points, closest baselines, and fal","pith_inferences":["If the governor semantics is right, it gives a cheap diagnostic: any control decision left to a prompt rather than an explicit state-and-trigger law is likely still at D2–D3, so migration depth can be estimated from interface design alone.","The pattern of independent convergence suggests a prediction: as these bundles are implemented, agent runtimes will increasingly resemble the classical architectures' couplings — a convergence that would validate the historical link without any documented lineage.","The closed skill-governance case implies a methodological extension: every residual bundle should be periodically re-screened against closest baselines as new systems appear, so some of the five bundles may already be closed or narrowed by systems published after this review's cutoff.","A testable extension the paper does not pursue is applying the same E/D coding to commercial agent platforms with persistent memory, interruption, and approval hooks, whose released interfaces may compose parts of the commitment and resource bundles faster than research prototypes."],"forward_implications":["Future agent evaluations should measure internal transition integrity — state integrity, interruption behavior, repeated failure, drift, transfer, and cost — alongside task success, not just final outputs.","Adaptive memory research must now be compared against systems that already learn memory operations and admission under value-per-byte governance, so the incremental contribution is the activation–latency–utility coupling, not adaptive admission itself.","Skill governance as a composed chain is already implemented by a reviewed system combining calibrated routing, typed verification, bounded repair, and replanning/fallback; any proposed extension must use that system as the mandatory baseline.","Each residual bundle yields a preregistered intervention with a runtime insertion point, an added control law, a closest baseline, and a primary falsifier — not an open-ended architecture proposal."],"fun_headline_variants":["Five control couplings still missing from agent runtimes","Agent runtimes miss five control couplings, review finds","Five control-coupling gaps persist in language agents","Review pinpoints five missing control couplings in agents","Five agent control couplings remain uncomposed: review"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The partition into five residual bundles rests on a single coder's provisional classifications of every D3/D4 claim; the paper states that independent inter-rater reliability has not been established, and any reclassification could open or close a bundle.","fun_headline_variants_meta":{"raw":{"variants":["Five control couplings still missing from agent runtimes","Agent runtimes miss five control couplings, review finds","Five control-coupling gaps persist in language agents","Review pinpoints five missing control couplings in agents","Five agent control couplings remain uncomposed: review"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000108,"raw_usage":{"total_tokens":896,"prompt_tokens":771,"completion_tokens":125,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":515,"completion_tokens_details":{"reasoning_tokens":52}},"tokens_in":515,"tokens_out":125,"duration_ms":2575,"temperature":1.0,"reasoning_tokens":52,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-31T23:26:52.012087+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Have a second coder, blind to the provisional codes, re-classify the D3/D4 evidence packets the paper provides: if any reviewed system already composes one of the five missing couplings, that bundle closes, just as the skill-governance candidate closed. Alternatively, run the paper's own preregistered protocol for the memory bundle against a recency-based baseline; if the simpler selector is non-inferior on recall and overhead, the bundle dissolves empirically.","supporting_citations":[],"review_version":1}