Personal AI skills by Santhosh Gandhi — VC content creator · systems thinker · author · AI-enabled builder.
A collection of Claude Code skills I find useful and interesting. I open-source them in case others do too.
AI tools forget. You shouldn't have to repeat yourself.
Every session starts from zero unless you build systems that survive context resets. Every token wasted on verbose responses or unnecessary file reads is a token not spent on the actual problem.
I'm not a software engineer. I'm sharing what I'm exploring as someone who came to this from a different direction — and I believe that perspective is useful, especially for others who aren't engineers either.
| Skill | What it does | Invoke |
|---|---|---|
context-manager |
Three-layer context system. SETUP builds docs/context/ so any AI tool can resume without re-reading everything. Two files per domain — operational ({domain}.md) and reference ({domain}-ref.md). UPDATE distills sessions into long-term learnings. |
/context-manager |
arch-map |
Generates docs/ARCHITECTURE.md — Mermaid system diagram, plain English overview, data flow walkthrough, tech stack with rationale. For humans: share with stakeholders, collaborators, contributors. |
/arch-map |
frugal-token-usage |
Mid-session audit. Stops unnecessary Bash calls, enforces dedicated tools, cuts verbose responses. Run when a session feels bloated. | /frugal-token-usage |
foresight-intelligence |
Strategic foresight using IFTF methodology. Soft Predict (instant) and Hard Predict (deterministic Python pipeline, identical output every run). | /foresight-intelligence |
review |
Production-readiness review. Finds real bugs — logic errors, null access, data loss, auth gaps, race conditions, hardcoded secrets. Prioritised findings with exact file:line references. No style feedback. |
/review |
security-audit |
Security audit covering OWASP Top 10 — injection, auth gaps, secrets in code, missing validation, info leakage, CORS misconfig, missing rate limiting. Each finding includes attack vector and fix. | /security-audit |
whybroken |
Root-cause tracer. Forms a hypothesis, follows the execution path to the actual cause, proposes the precise fix. Never patches symptoms. | /whybroken |
post-ai-thinking |
Unbiased solution space explorer. Strips inherited human heuristics, enumerates all possible approaches across 7 paradigms (human heuristic, hardware-native, theoretical limits, biology-inspired, economic/incentive, AI-augmented, hybrid), then lets you choose. Auto-calibrates depth — quick scan for tool choices, full space for architecture decisions. Existing solutions are fully included. | /post-ai-thinking |
Tip for
frugal-token-usage: The rules work better as permanent standing orders than on-demand. Copy them into~/.claude/CLAUDE.md. A starter file is inclaude.md.example.
Requires Claude Code to be installed.
Just tell Claude Code what you want:
Install the context-manager skill from github.com/isanthoshgandhi/santhoshstack
Install the arch-map skill from github.com/isanthoshgandhi/santhoshstack
Install the frugal-token-usage skill from github.com/isanthoshgandhi/santhoshstack
Install the foresight-intelligence skill from github.com/isanthoshgandhi/santhoshstack
Install the review skill from github.com/isanthoshgandhi/santhoshstack
Install the security-audit skill from github.com/isanthoshgandhi/santhoshstack
Install the whybroken skill from github.com/isanthoshgandhi/santhoshstack
Install the post-ai-thinking skill from github.com/isanthoshgandhi/santhoshstack
Install all skills from github.com/isanthoshgandhi/santhoshstack
Claude will handle the rest. No commands, no paths, no platform differences.
Clone once:
git clone https://github.com/isanthoshgandhi/santhoshstack.gitCopy what you need:
Mac / Linux:
cp -r santhoshstack/skills/context-manager ~/.claude/skills/
cp -r santhoshstack/skills/frugal-token-usage ~/.claude/skills/
cp -r santhoshstack/skills/foresight-intelligence ~/.claude/skills/
# or all at once
cp -r santhoshstack/skills/* ~/.claude/skills/Windows (PowerShell):
Copy-Item -Recurse santhoshstack\skills\context-manager $env:USERPROFILE\.claude\skills\
Copy-Item -Recurse santhoshstack\skills\frugal-token-usage $env:USERPROFILE\.claude\skills\
Copy-Item -Recurse santhoshstack\skills\foresight-intelligence $env:USERPROFILE\.claude\skills\
# or all at once
Copy-Item -Recurse santhoshstack\skills\* $env:USERPROFILE\.claude\skills\Verify:
ls ~/.claude/skills/ # Mac / Linuxls $env:USERPROFILE\.claude\skills\ # WindowsNo restart needed.
Writing & Media
- Medium — Writing on venture capital, startups, and AI
- YouTube — VC with Santhosh — VC content
- Instagram — @vcwithsanthosh — VC content
Book
Open source
- venture-capital-intelligence — Startup screening, pitch decks, cap tables, financial modeling
- awesome-vc-opensource — Curated list of open-source tools for VC and startups
MIT License