c31-daily-intel
Skill1 starsChanged 3 months agoMITView on GitHub
Daily RL-driven intelligence briefing for Chian
In this repository56 other files it gives its agents.
- C31-1stSkill
- C31-adopt-projectSkill
- C31-brainstormSkill
- C31-coding-disciplineSkill
- C31-compound-refreshSkill
- C31-compoundSkill
- C31-context-engineeringSkill
- C31-curatorSkill
- C31-debugSkill
- c31-devSkill
- C31-grillSkill
- C31-lfgSkill
- C31-planSkill
- C31-researchSkill
- C31-specSkill
- C31-stormSkill
- C31-strategySkill
- C31-workflow-bug-reproductionSkill
- C31-workSkill
- rl-workflowSkill
- C31-cvideoSkill
- c31-loopSkill
- c31-sxsSkill
- teachSkill
- video-content-analysisSkill
- agent-reachSkill
- kimiimSkill
- skillhub-preferenceSkill
- c31-communitySkill
- c31-growSkill
- c31-gutcheckSkill
- c31-marketSkill
- c31-mvpSkill
- c31-priceSkill
- c31-processSkill
- c31-sellSkill
- c31-validateSkill
- c31-valuesSkill
- growth-hackerSkill
- c31-improve-codebase-architectureSkill
- C31-multi-reviewSkill
- C31-review-adversarialSkill
- C31-review-architectureSkill
- C31-review-securitySkill
- C31-reviewSkill
- c31-wizardSkill
- external-privacy-guardSkill
- find-skillsSkill
- gsd-map-codebaseSkill
- gsd-new-projectSkill
- gsd-progressSkill
- gsd-quickSkill
- gsd-shipSkill
- humanize-aiSkill
- time-awarenessSkill
- worker-safetySkill
The file
We split this skill into the rules it gives an agent
11 rules. About 0 are common, 10 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈10
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 133
All KB signals must answer: "Should this be added to investment-os, or enhance existing cross-references?"
- Distinctive: line 156
Only include signals with decision value
- Distinctive: line 157
Every signal must have a source URL
- Distinctive: line 158
No meta language ("Here's today's briefing")
- Distinctive: line 159
If a project has no new signals, skip it entirely
- Distinctive: line 161
If a tool launched + someone claims stats, report as two separate signals or clearly distinguish: "[Tool] launched" vs "[Person] reports …
- Distinctive: line 162
Never imply Tool X validated Claim Y unless the source explicitly states the validation
- Distinctive: line 164
When multiple sources touch the same topic, keep facts segregated — merge only if the same source makes both claims
- Very specific: line 171
Default to Claimed or Launched, never upgrade to Validated without explicit evidence
- Distinctive: line 190
No feedback → no change
- Distinctive: line 209
All core keywords = 1.0, all explore keywords = 0.5
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
C31 Daily Intel Briefing
Schedule
Daily at 08:00 Asia/Shanghai, delivered to C31 Signal group
Input
memory/session_state.json→ pending_tasks (P0/P1)memory/.planning/daily-intel-weights.json→ keyword weights
Project Definitions
investment_os_kb
core:
- new primary source discovery
- investment concept framework update
- latest Howard Marks memo Oaktree
- Warren Buffett annual letter new
- knowledge graph cross reference enhancement
explore:
- behavioral finance new research
- quantitative fundamental hybrid framework
- AI knowledge management investing
- investment education innovation
- reflexivity Soros new writingproject_q
core:
- Qullamaggie strategy backtest
- quantitative trading framework
- market regime detection
- factor investing momentum
- position sizing risk management
explore:
- volatility targeting
- trend following systems
- order flow analysis
- portfolio optimization
- alternative data tradingchian_io
core:
- personal brand growth strategy
- content product monetization
- SEO organic traffic growth
- knowledge product positioning
- digital presence building
explore:
- newsletter growth tactics
- audience building indie hacker
- content repurposing strategy
- paid community models
- info product marketingnewsletter
core:
- newsletter content strategy
- paid subscriber growth
- email marketing engagement
- content distribution channels
- audience retention tactics
explore:
- podcast newsletter synergy
- referral program growth
- sponsorship model pricing
- content curation strategy
- automated content pipelinehenkaku_joi_ito
core:
- Joi Ito Henkaku Center
- MIT Media Lab innovation
- Japanese startup ecosystem
- university industry collaboration
- creative commons digital governance
explore:
- Chiba University SDS data science
- Kyoto University innovation lab
- academic entrepreneurship Japan
- research commercialization models
- open innovation policysilicon_boardroom
core:
- podcast product design
- interview based content strategy
- media startup monetization
- audio content production
- guest booking strategy
explore:
- video podcast production
- community driven media
- subscription media models
- cross platform content strategy
- live event productionSearch Strategy (RL: 90% exploitation, 10% exploration)
Daily allocation: 8 searches
- 5 exploitation → highest-weight keywords across all 6 projects (weighted by project weight × keyword weight, pick top 5)
- 2 adjacent explore → pick from explore pools of projects with highest aggregate weight
- 1 cross-project explore → randomly pick a keyword from any explore pool, weighted inversely (prefer low-weight projects)
Search rules:
- Serial execution, 15s interval between searches
- Limit results to past 7 days
- Skip if no new signal found (don't report "no news")
KB-specific search rules (investment_os_kb):
- For "new primary source discovery": search for newly published investment classics, recently digitized archives, or newly translated works
- For "latest Howard Marks memo": search Howard Marks / Oaktree for latest memo (prioritize this if no new memo in past 30 days in KB)
- For "investment concept framework update": search for new frameworks/models gaining traction in value/macro/growth investing
- For "knowledge graph cross reference enhancement": search for articles that explicitly connect multiple investors' frameworks (e.g., Buffett + Munger on moats, Marks + Taleb on tail risk)
- All KB signals must answer: "Should this be added to investment-os, or enhance existing cross-references?"
Output Format
📌 Today
[P0 task 1] → [suggested next action]
[P0 task 2] → [suggested next action]
📡 Signal — [project name]
- [signal 1] — [source URL]
- [signal 2] — [source URL]
📡 Signal — [project name]
...
🧠 KB — investment-os
- [发现类型]: [具体内容] → [建议动作]
Source: [URL]Output rules:
- Total ≤ 500 words
- Only include signals with decision value
- Every signal must have a source URL
- No meta language ("Here's today's briefing")
- If a project has no new signals, skip it entirely
- Source attribution rules (anti-conflation):
- If a tool launched + someone claims stats, report as two separate signals or clearly distinguish: "[Tool] launched" vs "[Person] reports [stat]"
- Never imply Tool X validated Claim Y unless the source explicitly states the validation
- Distinguish: "backtester available" (tool launch) ≠ "strategy validated" (independent verification)
- When multiple sources touch the same topic, keep facts segregated — merge only if the same source makes both claims
- Example of WRONG: "核心策略验证:楼梯模式+VCP+Episodic Pivots,胜率20-35%" (conflates tool launch + self-reported stats)
- Example of RIGHT: "Qullamaggie backtester工具上线 [URL]" / "Qullamaggie自述历史胜率20-35%,盈亏比3-5倍 [URL]"
- Claim strength tagging (internal check before output):
Validated= independent study/data confirms itClaimed= someone says it, no independent verification foundLaunched= product/tool released- Default to
ClaimedorLaunched, never upgrade toValidatedwithout explicit evidence
KB signal format (when investment_os_kb produces findings):
- 标注发现类型:
New Source/New Concept/Cross-Ref Enhancement/Content Update/Framework Gap - 建议动作必须具体:
- New Source → "加入 [collection_name] 候选队列"
- New Concept → "创建概念节点 '[concept_name]' 并关联 [existing_concepts]"
- Cross-Ref Enhancement → "在 [source_A] 和 [source_B] 之间建立 '[relationship]' 链接"
- Content Update → "更新 [existing_source] 的 [field]"
- Framework Gap → "investment-os 缺少 '[framework]' 覆盖,建议评估加入"
RL Feedback Loop
Weight file: memory/.planning/daily-intel-weights.json
Adjustment rules:
- User asks "展开讲讲" / "详细讲讲" for a signal → keyword +0.2
- User asks "多搜[领域]" → all keywords in that project +0.05
- User says "不感兴趣" / "跳过" → keyword -0.3
- No feedback → no change
- Weekly normalization: divide all weights by max, cap at 2.0, floor at 0.1
- Keywords below 0.3 → moved to explore pool
- New keywords added via exploration start at 0.5
KB-specific feedback rules (investment_os_kb):
- User says "加入" / "收录" / "这个要加" for a KB signal → keyword +0.3, 并在
memory/.planning/kb-candidates.json中记录候选 - User says "下次多找这类 source" → "new primary source discovery" +0.2
- User says "concept 不够" / "缺概念" → "investment concept framework update" +0.2
- User says "cross-reference 太少" → "knowledge graph cross reference enhancement" +0.2
- User says "memo 更新了吗" / "怎么没抓到最新 memo" → "latest Howard Marks memo Oaktree" +0.4 (高优先级修复)
- KB 候选在 7 天内未被用户处理 → 自动从 candidates 移除,keyword 权重 -0.1
KB candidate tracking:
- 文件:
memory/.planning/kb-candidates.json - 结构:
{ "candidate_id": { "type": "source|concept|cross_ref|update", "description": "...", "source_url": "...", "suggested_action": "...", "discovered_date": "YYYY-MM-DD", "status": "pending|approved|rejected" } } - 每日 briefing 产出时自动更新此文件
Initial weights (uniform): All core keywords = 1.0, all explore keywords = 0.5
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

