Ccf Literature Monitor

CCF Literature Monitor

Core Rule

Monitor arXiv, OpenReview, conference/proceedings feeds, project pages, labs, and named competitors for new papers that could overlap with the user's idea or paper. Report findings factually. Do not exaggerate novelty threats, dismiss real overlap, or infer priority from weak evidence. Provide actionable signals: RELAX, RESEARCH, FOLLOW-UP.

Modes

• arxiv-watch: scan recent papers in target categories or keyword clusters.
• venue-watch: scan OpenReview, proceedings, accepted-paper lists, or official venue pages.
• novelty-check: compare a given idea with recent papers and flag overlap.
• trend-scouting: summarize emerging directions, crowded areas, and under-tested gaps.
• competitor-tracking: monitor named research groups, labs, repositories, datasets, benchmarks, or authors.
• paper-alert-digest: produce a recurring watch report suitable for saving into a project folder.

Skill Linkage

• @READS ccfa.yaml for idea state.
• @SHARES findings with ccf-literature-searcher.
• @SHARES novelty implications with ccf-idea-reviewer.
• @SHARES rescue or differentiation opportunities with ccf-idea-optimizer.
• @SHARES citation and positioning candidates with ccf-paper-writer.
• @FLAGS conflicting papers for ccf-integrity-auditor.

Invocation Controls

CCFA Handoff Mode: PARTIAL (Recommended). Follow metadata.ccf_skill_controls.handoff_question_mode and ../ccf-common/references/handoff-modes.md.

Load ../ccf-common/references/task-modes.md before deciding exploratory, quick, or standard mode.

Treat unpublished ideas, draft abstracts, methods, and results as private material. Load ../ccf-common/references/privacy-and-evidence.md before using private text in queries. Use public-safe keywords, public method names, venue names, and user-approved query text.

Load ../ccf-common/references/review-output-standards.md when monitor results feed idea scoring, paper review, or score-risk language.

Workflow

1. Load references/monitoring-workflow.md. 2. If project state is available (ccfa.yaml), read the current idea, target venue, claims, and tracked competitors. 3. Choose the mode and time window. If the user says "latest", "recent", "new", "this week", or "today", verify the current date and use an explicit date range. 4. Execute the requested monitoring mode using public-safe queries and source-quality exclusions from the shared policy. 5. Classify overlaps by problem, mechanism, evidence, benchmark, dataset, claim, and venue positioning. 6. Report findings with precise implications for other CCFA skills. 7. Offer optional handoff to ccf-literature-searcher for deep retrieval, ccf-idea-reviewer for score impact, ccf-idea-optimizer for differentiation, or ccf-paper-writer for related-work integration.

Output Contract

For each monitoring execution, report in this structure:

## Literature Monitoring Report: [Mode] - [Topic/Venue]

**As of:** [YYYY-MM-DD]
**Time range:** [from] to [to]
**Sources scanned:** [arXiv categories / venues / labs / APIs]
**Papers scanned:** [count]
**High-relevance papers:** [count]
**Overall signal:** RELAX / RESEARCH / FOLLOW-UP

| Title | Source | Overlap level | Overlap type | Evidence basis | Action |
|:---|:---|:---:|:---|:---|:---:|
| [title] | [arXiv/OpenReview/venue] | None/Low/Medium/High | Problem/Method/Evidence/Benchmark | [abstract/intro/result] | RELAX/RESEARCH/FOLLOW-UP |

**Actionable flags:**
- RELAX: no material overlap found in the scanned window; continue, but do not claim full novelty proof.
- RESEARCH: partial overlap or possible close work; deep retrieve via `ccf-literature-searcher`.
- FOLLOW-UP: significant overlap in problem, mechanism, and evidence; route to `ccf-idea-reviewer` or `ccf-idea-optimizer` before writing stronger novelty claims.

**Handoff signals:**
- `ccf-literature-searcher`: [papers or clusters to retrieve deeply]
- `ccf-idea-reviewer`: [novelty or score implications]
- `ccf-idea-optimizer`: [differentiation or rescue direction]
- `ccf-paper-writer`: [related-work or positioning update]
- `ccf-integrity-auditor`: [citation/attribution conflict]

**Output self-check:** [date range explicit, source basis stated, table valid, no unsupported novelty conclusion]

References

• references/monitoring-workflow.md — Workflow, api-key-management, search strategy.
• references/report-template.md — Output template and formatting.
• ../ccf-common/references/source-registry.yaml — Competitor and arxiv links.

Monitoring Workflow

1. Setup

• Identify target categories (e.g., cs.CV, cs.CL, cs.LG, stat.ML), venues, labs, authors, repositories, datasets, benchmarks, or keyword clusters.
• Determine time range: most recent period (last 7 days, 14 days, 30 days), or since a specific date.
• If the user wants novelty check, read the current idea from ccfa.yaml when available. Err on the side of inclusion for potential overlap: if a new paper has a similar problem, insight, mechanism, dataset, or claim, report it with uncertainty rather than hiding it.
• If the user's idea is private, transform it into public-safe query terms before searching.

2. Search Strategy

• Use arXiv API, OpenReview, official venue/proceedings pages, Semantic Scholar, OpenAlex, DBLP, Crossref, and project pages according to the shared source registry.
• Fetch recent papers in target categories or from the requested venues/labs.
• For each high-relevance candidate, read at least the title and abstract. Do not claim high overlap without checking the abstract and, when accessible, the introduction or method summary.
• If the user has specific keywords, add them to the search or use them as filters after the broad search.
• Deduplicate by normalized title and stable URL.

3. Overlap Detection

• Compare each new paper's research question and main claims with the user's idea.
• Check six axes:
• problem/task,
• claimed gap,
• core mechanism,
• evidence type,
• dataset/benchmark,
• venue positioning.
• Classify each overlap:
• None (RELAX): different problem, method, and evaluation; no immediate action.
• Low (RESEARCH): shared motivation or setting, but different method or claim; retrieve if relevant.
• Medium (RESEARCH): overlapping mechanism, benchmark, or claim; deep search and differentiation are needed.
• High (FOLLOW-UP): same problem, similar method, and similar evidence path; major novelty or positioning revision may be needed.
• Assign an overlap score only as a diagnostic:

Overlap score (0-5):
0 = no relation
1 = same broad area
2 = same problem or dataset
3 = same problem plus related mechanism
4 = same problem plus similar mechanism or claim
5 = same problem, mechanism, and evidence path

4. Reporting

• Use the report template from references/report-template.md.
• Be honest about uncertainty: report what you don't know, not what you suspect.
• Format all high-relevance papers as a table with title, source, date, overlap score, overlap type, evidence basis, and recommended action.
• Include exact date range and source basis.

5. Follow-up and Persistence

• Store the monitoring output in the project directory or output/literature-monitor/ when file output is requested or a project folder exists.
• Propose a summary for ccfa.yaml under artifacts.literature_monitor_last_run; do not silently overwrite project state unless explicitly asked.
• Flag conflicting papers for next use: add them to the project literature cache.

6. Handoff Rules

• Send RESEARCH items to ccf-literature-searcher for deep retrieval and closest-work clustering.
• Send FOLLOW-UP items to ccf-idea-reviewer when the user needs a score impact or to ccf-idea-optimizer when the user wants a rescue/differentiation route.
• Send new citation/positioning candidates to ccf-paper-writer only after stable source links are available.
• Send attribution or already-cited-paper conflicts to ccf-integrity-auditor.

Monitoring Report Template

For all monitoring modes, use this structure. Output must be complete, factual, and parsable.

Report Header

## Literature Monitoring Report: [Monitoring Mode] - [Topic/Category]

**Date:** [YYYY-MM-DD]
**Time Range:** [from] to [to]
**Sources Scanned:** [arXiv categories / OpenReview venues / proceedings / labs / project pages]
**Query Basis:** [public-safe keywords or user-approved query text]
**Papers Scanned:** [count]
**High-Relevance Papers:** [count]
**Overall Signal:** RELAX / RESEARCH / FOLLOW-UP
**Confidence:** [1-5 with reason]

Source Coverage

| Source | Scope | Count | Notes |
|:---|:---|---:|:---|
| arXiv | [categories/queries] | [count] | [coverage limits] |
| OpenReview / venue | [venues] | [count] | [coverage limits] |
| Labs / competitors | [names] | [count] | [coverage limits] |

Executive Summary

One paragraph summary of the monitoring result, including total papers scanned, number of high-relevance papers, overlap distribution, and the safest next action. Do not claim that an idea is novel solely because no overlap appeared in the scanned window.

High-Relevance Papers

Table of papers that show potential overlap with the user's idea.

Detailed Findings

For each high-relevance paper, provide:

• Title: [full title]
• URL: [arXiv, OpenReview, proceedings, DOI, or project link]
• Source/date: [source and publication/preprint date]
• Key contribution: [what the paper claims or contributes]
• Overlap rationale: [why this overlaps or differs from the user's idea]
• Uncertainty: [what was not checked]
• Action: [RELAX | RESEARCH | FOLLOW-UP with reasoning]

Trend Map

Summarize emerging trends and open gaps in the scanned sources. Separate observed trends from inferred opportunities.

Flags for Other Skills

List explicit flags for other CCFA skills to act upon:

• @ccf-literature-searcher: [papers or clusters to retrieve deeply]
• @ccf-idea-reviewer: [novelty implications for idea score]
• @ccf-idea-optimizer: [differentiation or rescue routes]
• @ccf-paper-writer: [related-work or positioning updates]
• @ccf-integrity-auditor: [potential dependency or attribution issues]

Output Self-Check

Date range explicit:
Source basis stated:
Private text avoided or user-approved:
Table columns valid:
No unsupported novelty conclusion:
Handoff flags concrete: