prime
Builds a tailored primer for an AI coding session — the stack, style, constraints, threat model, and audience that should be loaded BEFORE any real task.
Skill by AQaddora
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/aqaddora-prime
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: security, productivity, agent-building
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/prime
What's inside
- Skill name
prime- Description
- Builds a tailored primer for an AI coding session — the stack, style, constraints, threat model, and audience that should be loaded BEFORE any real task. Triggers when the user says "prime me", "set up context", "build a primer", "start a new session", "I'm starting a new feature", or begins a fresh conversation without context. Reads the user's ~/ai-doctrine.md (if present) to seed the primer with rules they already established. Always offers to save high-leverage primers back to doctrine so the same setup carries to Cursor, Copilot, Codex, Gemini, and Windsurf via sync_to_agents.sh.
- License
- MIT
- Instructions
- 93 lines in SKILL.md · view on GitHub
Files (3)
- SKILL.md
- primer-format.md · references/
- README.md · scripts/
SKILL.md (first lines)
# Prime
The first letter of **PURE**. The cheapest leverage in AI-assisted coding.
A weak primer means the next 30 minutes are spent correcting the AI. A strong primer means the next 30 minutes are spent shipping. This skill builds the strong one.
## When to Use This Skill
Trigger when the user:
- Says "prime me", "set up context", "build a primer", "start a new session"
- Opens a new conversation and hasn't given any stack info
- Starts working on a new feature, repo, or task type
- Says "I'm switching projects" or "different context now"
- Asks to "load my doctrine" before a task
## How It Works (4-Step Protocol)
### Step 1 — Read existing context
1. Check `~/ai-doctrine.md` for the user's *Default Primer* section + any **Prime**-category rules.
2. Check `./CLAUDE.md`, `./AGENTS.md`, or `./.cursorrules` in the current project for project-specific primers.
3. If both exist, merge: project-specific overrides personal defaults where they conflict.
### Step 2 — Identify the gap
Ask the user (one question at a time, max 3) only for fields that are missing:
- **Stack** — languages, frameworks, versions
- **Style** — naming conventions, formatting, testing approach
- **Constraints** — performance targets, latency budgets, deadlines
- **Threat model** — production code? hostile inputs? sensitive data?
- **Audience** — who reads/runs the output
### Step 3 — Render the primer
Output the primer as a single fenced block the user can copy into any chat. Format defined in `references/primer-format.md`.
### Step 4 — Offer doctrine save
If the user adjusted defaults during the conversation (e.g., "actually use Pydantic v2 for this one"), end by offering to save the new default to `~/ai-doctrine.md` via `doctrine-keeper`. This carries the change to every other AI tool on the user's machine.
## Files in This Skill
| File | Purpose |
| --- | --- |
| `SKILL.md` | This file |
| `references/primer-format.md` | Canonical primer template — TODO: fill in with annotated examples |
| `scripts/README.md` | Scripts placeholder (logic to be built in a dedicated session) |
## Operational Rules
1. **Never assume.** If a field is missing, ask once. If the user says "you decide", pick a sensible default and flag it explicitly so they can override.
2. **One question at a time.** Don't fire a 5-question survey. Walk through fields in priority order.
3. **Cite the doctrine.** When pulling defaults from `~/ai-doctrine.md`, name the rule so the user knows where it came from.
4. **Output is copy-pasteable.** The primer must work pasted into a fresh Claude/Cursor/Copilot chat with zero edits.
5. **Suggest doctrine candidates.** Any primer field that took >1 message to establish should be a save candidate.
## Doctrine Integration
Every primer this skill builds is a doctrine candidate. After rendering the primer, the skill checks:
- Did the user supply a new field value not in their doctrine? → Offer to save as a **Prime** rule.
- Did the user accept a sensible default we suggested? → Offer to save as the new default.
- Did the user reject a default ("nah, use bun not npm here")? → Offer to save as a **Prime** rule for this stack.
Save flow routes through `doctrine-keeper`, which appends to `~/ai-doctrine.md`. After the next `sync_to_agents.sh`, every coding tool the user opens reads the new default.
## Example Invocation
**User:** "Prime me for a FastAPI feature."
**Skill activates and produces:**
```
Stack: Python 3.11, FastAPI 0.110, Pydantic v2, asyncpg, PostgreSQL 16
Style: type hints everywhere, pytest, ruff, async/await for all I/O
Constraints: production code, sub-100ms p95 latency, all inputs assumed hostile
Audience: senior backend engineers (review-ready code)
Threat model: public endpoint, rate limiting required, OWASP Top 10 considered
I will give you tasks one at a time. Confirm context, then wait for the task.
```
> *Pulled from `~/ai-doctrine.md` — Default Primer (5 rules) + Prime-category rules #03, #07, #14.*
> 💡 Save adjustment? You overrode "FastAPI 0.108" → "FastAPI 0.110" today. Save as the new default in your doctrine?
---
*Part of the [Vibe Coding Mastery](https://github.com/AQaddora/vibe-coding-mastery) skill suite. MIT licensed.*From AQaddora/pure-skill-suite, MIT.
More from AQaddora/pure-skill-suite
- 0.519score
understand
AQaddora
Clarifies scope before code is written.
SkillCFreshness: C★ 3Claude Code +9 - 0.519score
execute
AQaddora
Delegates real work to a coding agent (Claude Code, Cursor agent mode, Codex, Aider) and tracks the run end-to-end.
SkillCFreshness: C★ 3Claude Code +9 - 0.519score
handoff
AQaddora
Migrates an in-flight AI coding session to a fresh chat (or a larger-context model) without losing state.
SkillCFreshness: C★ 3Claude Code +9 - 0.519score
organize-agents
AQaddora
Maintains a registry of agent runs and outputs across every coding agent the user works with (Claude Code, Cursor, Codex, Aider, Gemini CLI, Windsurf).
SkillCFreshness: C★ 3Claude Code +9 - 0.519score
pure-orchestrator
AQaddora
Runs the full PURE Loop (Prime → Understand → Refine → Execute) on a feature request, calling each phase's skill in sequence with explicit checkpoints for…
SkillCFreshness: C★ 3Claude Code +9 - 0.519score
refine
AQaddora
Runs systematic self-review on AI-generated output before it's accepted.
SkillCFreshness: C★ 3Claude Code +9 - 0.519score
doctrine-keeper
AQaddora
Captures insights, rules, and learnings from any AI coding conversation into the user's personal ai-doctrine.md file — a single source of truth that is then…
SkillCFreshness: C★ 3Claude Code +9 - 0.514score
report-back
AQaddora
The mandatory completion-report skill.
SkillCFreshness: C★ 3Claude Code +9 - 0.514score
status-beacon
AQaddora
Lightweight mid-flight progress logging for long agent runs.
SkillCFreshness: C★ 3Claude Code +9 - 0.508score
3d-modeler
AQaddora
Generates 3D interior models from photos and injects them into Mapbox GL JS map instances.
SkillCFreshness: C★ 3Claude Code +9 - 0.503score
montage-creator
AQaddora
Generates dynamic video montages and edits video content for social distribution.
SkillCFreshness: C★ 3Claude Code +9 - 0.274score
handoff-receiver
AQaddora
No description from the publisher yet.
SkillCFreshness: C★ 3Claude Code +9
Signals
- Stars
- 3
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- May 26, 2026 (5 months ago)
- Last release
- May 31, 2026 (5 months ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/AQaddora/pure-skill-suite
- Website
- aqaddoura.com/blog/pure
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from AQaddora
- 0.519score
handoff
AQaddora
Migrates an in-flight AI coding session to a fresh chat (or a larger-context model) without losing state.
SkillCFreshness: C★ 3Claude Code +9 - 0.519score
pure-orchestrator
AQaddora
Runs the full PURE Loop (Prime → Understand → Refine → Execute) on a feature request, calling each phase's skill in sequence with explicit checkpoints for…
SkillCFreshness: C★ 3Claude Code +9 - 0.519score
doctrine-keeper
AQaddora
Captures insights, rules, and learnings from any AI coding conversation into the user's personal ai-doctrine.md file — a single source of truth that is then…
SkillCFreshness: C★ 3Claude Code +9 - 0.519score
execute
AQaddora
Delegates real work to a coding agent (Claude Code, Cursor agent mode, Codex, Aider) and tracks the run end-to-end.
SkillCFreshness: C★ 3Claude Code +9 - 0.519score
organize-agents
AQaddora
Maintains a registry of agent runs and outputs across every coding agent the user works with (Claude Code, Cursor, Codex, Aider, Gemini CLI, Windsurf).
SkillCFreshness: C★ 3Claude Code +9 - 0.519score
refine
AQaddora
Runs systematic self-review on AI-generated output before it's accepted.
SkillCFreshness: C★ 3Claude Code +9
Cite this page
prime skill. Agent Mart. https://agentmart.dev/skills/aqaddora-prime (last updated 2026-10-24)