cs.CVApr 6, 2026
SaveFileGram: Grounding Agent Personalization in File-System Behavioral Traces
Organizations: S-Lab, Nanyang Technological University · Synvo AI
Abstract
Coworking AI agents operating within local file systems are rapidly emerging as a paradigm in human-AI interaction; however, effective personalization remains limited by severe data constraints, as strict privacy barriers and the difficulty of jointly collecting multimodal real-world traces prevent scalable training and evaluation, and existing methods remain interaction-centric while overlooking dense behavioral traces in file-system operations; to address this gap, we propose FileGram, a comprehensive framework that grounds agent memory and personalization in file-system behavioral traces, comprising three core components: (1) FileGramEngine, a scalable persona-driven data engine that simulates realistic workflows and generates fine-grained multimodal action sequences at scale; (2) FileGramBench, a diagnostic benchmark grounded in file-system behavioral traces for evaluating memory systems on profile reconstruction, trace disentanglement, persona drift detection, and multimodal grounding; and (3) FileGramOS, a bottom-up memory architecture that builds user profiles directly from atomic actions and content deltas rather than dialogue summaries, encoding these traces into procedural, semantic, and episodic channels with query-time abstraction; extensive experiments show that FileGramBench remains challenging for state-of-the-art memory systems and that FileGramEngine and FileGramOS are effective, and by open-sourcing the framework, we hope to support future research on personalized memory-centric file-system agents.
Figures & tables
Figure 1: Overview of the FileGram Project. FileGram introduces a personalized AI coworker natively integrated into the user file system. By consolidating cross-session activities and file outputs into long-term behavioral memory, the agent infers intent and proactively synchronizes workspaces, establishing a new paradigm for real-world interactive coworking.
Figure 2: Data generation pipeline. FileGramEngine generates one trajectory per profile–task pair. Agents execute in profile-isolated workspaces for each task; raw tool traces are filtered and canonicalized to retain real action signals while removing simulation artifacts, and outputs are materialized as standardized behavioral traces with aligned text/document/visual views for cross-modal evaluation.
| Dim | Name | Left (L) | Middle (M) | Right (R) |
| A | Consumption Pattern | Sequential deep reading | Targeted search-first | Breadth-first browsing |
| B | Production Style | Comprehensive & detailed | Balanced | Minimal & concise |
| C | Organization Preference | Deeply nested (3+ levels) | Adaptive (1–2 levels) | Flat (root only) |
| D | Iteration Strategy | Incremental small edits | Balanced refinement | Bulk rewrite |
| E | Curation | Selective (active cleanup) | Pragmatic (moderate cleanup) | Preservative (accumulative) |
| F | Cross-Modal Behavior | Visual-heavy (charts, figures) | Balanced (tables) | Text-only |
Table 2: Six behavioral dimensions with L/M/R tiers used for profile construction in FileGramEngine .
Figure 3: Data distribution. 20 profiles 32 tasks yield 640 trajectories comprising 10K output files and 20,028 atomic actions.
Figure 4: FileGramQA distribution. 4.6K questions by track (inner) and sub-task (outer).
Figure 5: QA examples from FileGramBench . Representative questions from the four tracks, including both MCQ and open-ended formats.
Figure 6: FileGramOS architecture. Three-stage pipeline: (1) per-trajectory encoding of traces via parallel extraction streams into an Engram; (2) cross-engram consolidation routing data into procedural, semantic, and episodic channels, including an LLM verifier for variation vs. outlier ; and (3) query-adaptive retrieval.
Figure 7: Qualitative comparison. Left: A BehavFP question where FileGramOS ’s three-channel architecture—procedural statistics, semantic narration, and episodic clustering—jointly recover the correct profile, while baselines each miss different signals. Right: A TraceDis question involving multimodal artifacts, where cross-format output gaps and parsing losses cause widespread failures.
Appendix figures & tables10 assets
Supplementary material from the paper’s appendix.
Appendix
| Retained: Atomic Actions | |||
| Event Type | Count | Category | Key Fields |
| file_read | 4,541 | Read | path, type, depth, view_count, view_range, length, revisit_ms |
| file_browse | 1,649 | Read | dir_path, files_listed, depth |
| file_search | 294 | Read | search_type, query, files_matched, files_opened |
| file_write | 3,024 | Write | path, type, operation, length, before/after_hash, media_ref |
| file_edit | 1,057 | Write | path, tool, lines_added/deleted/modified, diff, before/after_hash |
Table 5: Event types in raw trajectories. The upper block lists the 12 atomic action types retained after cleaning with their total counts across 640 trajectories. The lower block lists the 10 simulation metadata types removed, which account for 74.3% of all raw events.
| Group | Key | Source Events | Computation | Interpretation |
| reading_strategy | search_ratio | file_search, file_read, file_browse | Targeted search vs. sequential browsing | |
| browse_ratio | file_browse, file_read, file_search | Directory-level exploration | ||
| revisit_ratio | file_read | Re-reading previously viewed files | ||
| output_detail | avg_output_length | file_write: create | Average verbosity of created files | |
| files_created | file_write: create | Number of output files produced | ||
| total_output_chars | file_write: create | Total production volume |
Table 6: Procedural fingerprint specification. All 17 features are computed deterministically from cleaned atomic actions, grouped by the six behavioral dimensions. Source event types follow table 5 .
| A: Consump. | B: Product. | C: Organiz. | D: Iteration | E: Curation | F: Cross-M. | |
| UC1: Proactive | Predict next read | Predict format | Pre-create dirs | — | Predict cleanup | Predict chart need |
| UC2: Defaults | Set read mode | Set length & tone | Set folder depth | Set edit granularity | — | Set output modality |
| UC3: Smart Org. | — | — | Maintain hierarchy | — | Predict retention | — |
| UC4: Recovery | Reconstruct reads | Reconstruct drafts | Navigate folders | Reconstruct edits | — | — |
| UC5: Continuity | Consistent reading | Consistent style | Consistent structure | Consistent editing | Consistent curation | Consistent modality |
| UC6: Conflict | Detect read drift | Detect style change | Detect reorganiz. | Detect edit shift | Detect curation chg. | — |
Table 7: Dimension derivation. Each row is an OS-agent use case; each column is a behavioral dimension. Cells indicate the specific capability the dimension enables for that use case.
| Attribute | Dim. | L | M | R |
| name | — | Free-form display name | ||
| role | — | Professional role | ||
| language | — | Primary output language | ||
| reading_strategy | A | Sequential deep | Search-first | Breadth-first |
| thoroughness | A | Exhaustive | Selective | Minimal |
| tone | B | Formal, academic | Professional | Casual |
Table 8: Profile attribute schema. Three identity attributes define the user; 16 behavioral attributes span dimensions A–F, each discretized into L/M/R tiers. version_strategy is shared by dimensions C and D. Profile Reconstruction evaluates all 16 behavioral attributes per profile, yielding scored items.
| ID | Name | Role | A | B | C | D | E | F | ID | Name | Role | A | B | C | D | E | F |
| p1 | Chen Wei | Research Analyst | L | L | L | L | L | M | p11 | Priya Sharma | Supply Chain Ana. | M | L | L | L | L | R |
| p2 | Liu Jing | Policy Analyst | L | L | R | R | L | M | p12 | Wang Fang | Journalism Editor | R | L | R | L | M | M |
| p3 | Sam Taylor | Ops Manager | M | R | R | R | M | R | p13 | Zhao Ming | Landscape Arch. | L | M | L | M | L | L |
| p4 | Nakamura Yuki | Finance Consultant | M | L | M | L | R | L | p14 | Daniel Osei | Compliance Officer | M | R | L | L | M | R |
| p5 | Maria Santos | Marketing Coord. | R | M | M | M | R | M | p15 | Sophie Laurent | Project Manager | R | L | M | M | R | M |
| p6 | Alex Kim | Event Planner | R | M | L | R | M | R | p16 | Marcus Chen | Data Analyst | M | M | R | M | L | R |
Table 9: Profile instances. L/M/R tier assignments across dimensions A–F for all 20 profiles. Each tier appears in at least 5 profiles per dimension to prevent evaluation bias.
| Task | Type | MM | Description | Dims | In | Input File Types |
| T-01 | Understand | ✗ | Investment analyst work overview summary | A, B, E, F | 26 | .md(7), .pdf(6), .eml(5), .txt(5), .png(2), .csv(1) |
| T-02 | Understand | ✓ | Legal case materials review and timeline | A, B, E, F | 24 | .eml(6), .pdf(5), .docx(5), .txt(3), .png(2), .xlsx(1), .ics(1), .mp3(1) |
| T-03 | Create | ✗ | Personal knowledge base creation | B, (C), (E) | 0 | empty workspace |
| T-04 | Create | ✗ | Meeting minutes and follow-up document creation | B, (E) | 0 | empty workspace |
| T-05 | Organize | ✗ | Messy folder cleanup and reorganization | A, B, C, E, F | 30 | .eml(8), .png(4), .md(4), .txt(4), .pdf(4), .jpg(2), .ics(2), .csv(1), .docx(1) |
| T-06 | Synthesize | ✗ | Multi-source synthesis research report | A, B, E, F | 21 | .pdf(8), .eml(6), .txt(3), .md(3), .docx(1) |
Table 10: Task pool overview. 32 tasks across 6 types with their activated dimensions and input file composition. A dimension is listed when of profiles show non-trivial signal; parenthesized dimensions indicate 30–69% partial activation. MM marks multimodal tasks.
| QA Backbone | Proc. | Sem. | Avg. |
| Gemini 2.5-Flash | 84.4 | 80.0 | 82.8 |
| GPT-4.1 | 82.5 | 78.3 | 80.9 |
| Claude Sonnet 4 | 83.8 | 80.0 | 82.2 |
Table 11: Cross-backbone trace validation. Per-attribute reconstruction accuracy (%) across three QA backbones, all receiving the same FileGramOS memory context. Inter-backbone variance stays below 2.0 pp, confirming model-agnostic signal.
| Method | Category | Memory Mechanism | Key Config |
| No Context | Context | None | Lower bound |
| Full Context | Context | Full concatenation | 625.2K tok avg |
| Naive RAG | Context | Chunk embed + top-5 retrieval | 512-tok chunks, overlap 64 |
| VisRAG | Context | ColPali vision embed + top-5 | Page images + text fallback |
| Eager Summ. | Text | Per-trajectory LLM summary | Concatenated summaries |
| Mem0 | Text | Flat key–value store | Official SDK defaults |
Table 12: Baseline implementation summary. All 12 baselines plus FileGramOS share Gemini 2.5-Flash as QA backbone and receive identical cleaned event logs and output files per profile.
| Variant | T1: Underst. | T2: Reason. | T3: Detect. | Channel | Avg | |||||
| Attr Rec | Behav FP | Behav Inf | Trace Dis | Anom Det | Shift Ana | Proc | Sem | Epi | ||
| FileGramOS | 50.6 | 35.2 | 42.1 | 80.9 | 70.2 | 37.8 | 60.1 | 55.0 | 58.9 | 59.6 |
| Proc. | 5.5 | 1.1 | 2.3 | 27.8 | 3.9 | 8.3 | 12.2 | 7.3 | 6.8 | 11.1 |
| Sem. | 10.6 | 4.2 | 6.1 | 2.9 | 6.7 | 7.8 | 3.3 | 8.8 | 4.2 | 5.5 |
| Epi. | 5.1 | 4.2 | 5.1 | 1.9 | 6.2 | 5.8 | 2.3 | 2.5 | 6.1 | 4.2 |
Table 13: Channel ablation. First row: absolute accuracy (%); ablation rows: per-cell relative to the full model. The procedural channel contributes the largest overall drop; all three channels carry distinct, complementary signal.
| Configuration | T1 | T2 | T3 | Avg | |
| Retriever display length | |||||
| 300 chars (default) | 46.0 | 70.7 | 48.7 | 53.5 | — |
| 500 chars | 46.7 | 70.0 | 49.3 | 53.8 | 0.3 |
| 800 chars | 48.0 | 71.3 | 49.3 | 54.5 | 1.0 |
| 1000 chars | 46.7 | 72.0 | 49.3 | 54.3 | 0.8 |
| Context presentation | |||||
Table 14: Parameter sensitivity. Per-track accuracy (%) on Tracks 1–3. : relative to the 300-char default for retriever rows, and to the 800-char optimum for context rows. Ingest-time parameters have zero effect and are omitted.
More in this topic
Agent Memory
Agent memory lets AI systems retain, update, and use facts, experiences, and skills across sessions and long-horizon tasks.
See all 192 papers