sandbaseai/workbuddy-skill/deep-project-primer

Claude Code and agent skill by sandbaseai/workbuddy-skill

Trust tier 3
  • community
Source check pending

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

Build an evidence-backed project baseline before significant work by inspecting repository guidance, architecture, runtime, tests, and delivery constraints.

Install

npx vskill@latest install sandbaseai/workbuddy-skill/deep-project-primer

SKILL.md

The same SKILL.md was published earlier as sandbaseai/workbuddy-skill/deep-project-primer.

Deep Project Primer

Use this skill at the start of work in an unfamiliar repository, after context loss, or before a consequential architectural or delivery change. The goal is not to read every file indiscriminately; it is to establish the smallest reliable model of the project needed for the requested work and to expose what is still unknown.

Establish repository authority

Begin at the repository root and inspect, in order:

  1. AGENTS.md, CONTRIBUTING, SECURITY, and other local instruction files,

following the most specific applicable guidance.

  1. README and documented development, test, build, release, and deployment

commands.

  1. Package manifests, lockfiles, workspace boundaries, generated-file markers,

CI workflows, container or infrastructure definitions, and entry points.

  1. The requested feature's owning modules, public interfaces, persistence and

integration boundaries, tests, and recent relevant history.

Record the repository commit, branch, working-tree state, runtime assumptions, and commands actually executed. Distinguish facts observed in files or command output from inferences and recommendations. Do not treat an outdated README, an example configuration, or a green unrelated workflow as proof of current behavior.

Build the project model

Map the request to:

  • user or operator outcome and non-goals;
  • entry points, data/control flow, and ownership boundaries;
  • public contracts, compatibility obligations, and feature flags;
  • dependencies, external services, credentials, and authorization boundaries;
  • state, migrations, queues, caches, retries, and failure behavior;
  • test layers, fixtures, observability, rollout, and rollback paths.

For every important claim, keep a pointer to the file, symbol, command, test, or runtime observation that supports it. Identify stale, contradictory, or missing evidence explicitly. Ask for clarification only when an ambiguity would change scope or create a meaningful safety or compatibility risk; otherwise make a bounded assumption and label it.

Verify before changing

Select the cheapest repository-native checks that validate the baseline: a targeted test or inspect command, build/type check, lint or format check, and a safe runtime or integration probe when one is available. Never run unknown scripts, install unreviewed dependencies, expose secrets, or mutate production systems merely to complete reconnaissance. Treat generated files and lockfiles according to repository policy, and keep the initial diff empty unless the task explicitly requires a change.

Before a major change, state the affected contracts, likely regression paths, required approvals, verification evidence, and rollback or recovery condition. If the requested work crosses an authority boundary, stop at that boundary with an actionable handoff instead of guessing access or permission.

Context-restoration handoff

Return a compact, reproducible primer:

text
Repository / commit / branch / working-tree state:
Applicable instructions and constraints:
Request, outcome, and non-goals:
Architecture and ownership map:
Relevant files, contracts, dependencies, and data flow:
Commands run and observed results:
Tests, runtime evidence, and freshness:
Assumptions, unknowns, and risks:
Proposed next action / owner / verification:

The primer is complete only when another agent can restart from it without repeating broad exploration, while still knowing which claims require fresh verification. Keep sensitive values out of the handoff and preserve links to redacted evidence rather than copying secrets or customer data.

Security

  • Security scanPassed, 90 of 100Pattern checks on the skill's files.
  • Model reviewPassed

Quality evaluation

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

Source

Popularity

7-day trend score
0

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/sandbaseai/workbuddy-skill/deep-project-primer/badge)](https://verified-skill.com/skills/sandbaseai/workbuddy-skill/deep-project-primer)

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