hjsang/crisp_reasoning_compression/neon-access

Claude Code and agent skill by hjsang/crisp_reasoning_compression

Trust tier 3
  • community
Source check pending

Checking that the SKILL.md file is still in the repository.

Run arbitrary PostgreSQL SQL against the Miles Neon database through the repository-authorized GitHub workflow and return the encrypted result. Use when a repository writer asks to query, inspect, update, migrate, or otherwise operate on Neon without exposing NEON_DATABASE_URL. The SQL is not classified, rewritten, split, or restricted; the database role remains the authorization boundary.

Install

npx vskill@latest install hjsang/crisp_reasoning_compression/neon-access

SKILL.md

Neon Access

Use this skill when a writer asks to run SQL, for example: “Use $neon-access to run SELECT now().” The workflow is the only route to Neon; never ask for or expose NEON_DATABASE_URL.

Run SQL

  1. Treat an explicit request as authorization to generate the exact SQL and a short audit reason, then dispatch without a second confirmation. Ask only when unresolved ambiguity would materially change the SQL. Treat every statement as potentially state-changing because the skill does not inspect or classify it.
  2. Save the exact UTF-8 SQL in a temporary .sql file. Preserve the text byte-for-byte; do not normalize, split, wrap, retry, or add a transaction.
  3. From the repository root, run:

``bash python3 .claude/skills/neon-access/scripts/run_neon_workflow.py \ --reason '<short audit reason>' \ /absolute/path/to/request.sql ``

  1. Return the decoded JSON result and workflow URL. If the result status is error, report the database error and possible partial effects; never retry automatically.

Boundaries

  • The caller and workflow require GitHub write or admin permission; the workflow checks both the original and triggering actors.
  • The workflow reads the existing secrets.NEON_DATABASE_URL. Users do not retrieve or provide this secret.
  • Any PostgreSQL SQL accepted by the server is allowed, including multiple statements, DDL, DML, functions, and explicit transaction control.
  • psql backslash commands are client commands, not PostgreSQL SQL, and are unsupported.
  • Execution uses autocommit and performs no implicit transaction wrapping. A later error can leave earlier statements committed.
  • Returned rows are capped at 4 MiB; this does not limit or cancel SQL execution.
  • Results and database errors are encrypted for the one-time local key before artifact upload. Public logs and summaries contain metadata only.

Security

  • Security scanPassed, 95 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

Repository stars
69
Repository forks
7
7-day trend score
0

Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/hjsang/crisp_reasoning_compression/neon-access/badge)](https://verified-skill.com/skills/hjsang/crisp_reasoning_compression/neon-access)

Something wrong with this skill or its page? Tell us.