Organizations: Agency for Science, Technology and Research (A*STAR), Singapore · Tencent Research Institute, China · Shanghai Dongfang Yiteng Technology Co., Ltd., China · Chengdu Xuansu Technology, China · Wanlian Yida Holding Group Co., Ltd., China · Centre for Frontier AI Research (CFAR), A*STAR, Singapore
Agent pattern catalogs describe architectural alternatives, but engineers also need evidence for selecting and composing them. We extend a cognitive-function--execution-topology framework with an evidence-driven selection procedure and explicit composition contracts. The study combines two industrial case accounts, a contributor's collaboration design, seven practitioner workshops, and executable fault-injection experiments. Three local reference systems use SQLite state, shared artifact files, and a file-backed map-publication surrogate. A 2 x 2 ablation separates function-specific contracts from execution controls across 36 scenario families, with ten input variants per family. Under equal family weighting, composition correctly completes or safely refuses 11 of 12 families in each system, compared with 2, 2, and 1 for the simple baselines. These are constructed challenge-suite results, not production success rates. Recovery without consumer-side verification increases false-success cases in the map surrogate. All composed systems still fail a boundary case in which trusted evidence is wrong. An audit of a frozen six-arm language-model pilot reproduces its scores but confirms an evaluator-dependent ranking reversal. The contribution connects industrial failure, pattern selection, executable treatment, external acceptance, and revision. Selection is demonstrated rather than evaluated against an alternative procedure; catalog admission rests on separate criteria. The results support bounded composition mechanisms, not a claim that the framework improves engineers' design decisions over an equally informative conventional method.
Figures & tables
Source
Available evidence
Role in this study
Dongfang Yiteng
Contributor account of prototype failures, state architecture, task dependencies, and recovery design
Industrial motivation for exact parameter binding and durable execution
Xuansu GIS
Failure accounts, pipeline and lifecycle descriptions, console and map-output artifacts
Industrial motivation for consumer acceptance and bounded recovery
Shared artifacts
Fan’s two design notes and interaction diagram; no supplied deployment measurements
Attributed design input for event and artifact experiments
Seven workshops
Dated discussion records, transcripts, questions, disagreements, and proposed revisions
Requirements for measurement and selection
RegTech pilot
Six frozen model outputs, one source bundle, 12 checklist items, original scorer
Historical outcome and evaluator audit
Reference study
Executed SQLite/file systems, injected faults, per-run traces and artifact readback
Controlled evidence for the implemented mechanisms
Table 1: Source types and the claims they support. Existing field material and newly executed reference experiments retain separate provenance.
Figure 1: Evidence links a reported failure to an executable treatment and a decision that can be revisited. The external oracle remains fixed when the runtime verifier is ablated.
System
F : function-specific contract
T : execution control
Example pattern coordinates
Business
Bind the action to a trusted record; check task authority
Fix dependency order; commit effect and completion together; resume from checkpoints
Action/Memory responsibilities; Chain with durable orchestration and Governance admission
Artifacts
Validate allowed scope, version, and bytes against a trusted digest
Atomic effect/receipt; deduplicated consumption and completion; collection after all receipts
Decode consumer response; inspect status, type, body, and image uniformity
Route known observed errors to repair; bound retries; stop on unknown errors
Perception and verification; Reasoning/Reflection with Route and Loop
Table 2: Two treatment bundles in each reference system. All four combinations use the same environment constructor and task oracle.
System
B00
B10
B01
B11
Business execution
2 / 0 / 8
4 / 2 / 6
6 / 0 / 4
9 / 2 / 1
Shared artifacts
2 / 0 / 5
2 / 4 / 2
7 / 0 / 5
7 / 4 / 1
Map publication
1 / 0 / 8
1 / 3 / 1
2 / 0 / 10
8 / 3 / 1
Table 3: Main results as scenario families out of 12 per system. Every categorical cell is unchanged across the ten input variants. Each cell is accepted completion / safe refusal or stop / false success. Failed runs without a false claim account for the remainder.
Figure 2: Correct resolution includes separately specified refusals and bounded stops. It is not a task-completion rate. The fixed, deliberately fault-heavy suite gives equal weight to each family.
Arm
Predicted items
Matches
Precision
Recall
F1
Single pass
16
6
.3750
.5000
.4286
Critic and repair
16
7
.4375
.5833
.5000
Self-refinement
16
7
.4375
.5833
.5000
Candidate-guided review
16
9
.5625
.7500
.6429
Union queue
29
10
.3448
.8333
.4878
Consensus refinement
16
8
.5000
.6667
.5714
Table 4: Recomputed historical pilot scores under its original matcher. These are lexical/structural matching scores, not independently adjudicated semantic coverage.
Appendix figures & tables1 asset
Supplementary material from the paper’s appendix.
Appendix
Business execution
Shared artifacts
Map publication
1
Clean update
Clean two-consumer delivery
Clean publication
2
Altered proposal identifier
Duplicate delivery
HTTP 200 with XML error
3
Wrong valid identifier
Crash before effect
Blank rendered image
4
Reordered steps
Crash after effect
Zero metatile error
5
Crash before effect
Corrupt artifact bytes
Transient publication failure
6
Crash after effect
Stale artifact version
Transient consumer failure
Appendix
Table 5: Twelve families per reference system. The final row is an explicit boundary test of the composed implementation.
Large language model (LLM) agent reliability concerns dependable, safe, and verifiable behavior across tools, multi-step tasks, and changing environments.