- Home
- Skills
- Code Quality & Review
- briefback
Works with the AI Agents you already use
briefback
A structured development flow for complex features using Research, Grill, Plan, and Review phases.
Free
briefback
Example session with this skill installed
I need to implement a new multi-tenant auth system. Let's start with briefback research on our existing session logic.
- Read your context and instructions
- Compiled the briefback
I have activated Briefback in research mode. I am scanning the existing codebase to map our session management, middleware, and database schema for the new multi-tenant auth system.
Research Map: Existing Session Logic
| Component | Path | Responsibility |
|---|---|---|
| Session Middleware | src/middleware/auth.ts | Validates JWTs and sets request context. |
| Token Service | src/services/token.service.ts | Handles signing, rotation, and revocation logic. |
| User Model | src/models/user.model.ts | Defines user schema; currently lacks tenant_id. |
| Redis Store | config/redis.conf | Manages server-side session persistence. |
Observations
- Current session logic assumes a single global user namespace.
- Middleware retrieves user data by
subclaim without tenant validation. - The
Usertable requires a migration to supporttenant_idpartitioning.
Next steps
- Run
briefback grillto define the multi-tenancy isolation strategy (silo vs. pooled). - Initialize the
.briefback/auth-migration/directory to track implementation stages. - Draft a migration plan for the existing session store.
Connects securely to your tools. The creator never sees your data.
About this skill
Large software changes often collapse into chaotic, unreviewed code blobs or stalled implementation cycles. Briefback forces structure onto non-trivial development tasks by enforcing a disciplined Research-to-Commit delivery loop. It prevents premature coding by ensuring architecture and scope decisions are settled before a single line of implementation begins.
What it does
- Enforces lifecycle stages by routing work through Research, Grill, Plan, Stage, Review, and Commit phases.
- Settles architecture decisions via the Grill mode to resolve scope and product logic before execution.
- Sequences complex implementations into durable, trackable stages with clear review boundaries.
- Maintains context continuity by generating handoff artifacts that preserve state across sessions.
- Applies test-driven logic through a dedicated TDD mode for critical implementation slices.
How it works
- Invocation: The skill activates automatically for any change larger than a quick fix or via explicit commands like
/briefback research. - Phase Selection: It identifies the earliest necessary phase based on current project state and
.briefback/artifacts. - Execution: It loads only the specific playbook required for the current mode to minimize context noise.
- Iterative Delivery: It moves through the Stage-Review-Commit loop, repeating until the planned work is complete.
Frameworks & tools
This skill is language-agnostic and works with any stack. It manages local state using the .briefback/ directory to store project guidance, plans, and research maps.
Why this beats prompting it yourself
Generic prompts often jump straight to code, missing critical edge cases or architectural flaws. Briefback locks the agent into a specific methodology that mandates planning and review, ensuring complex features are delivered in logical, tested increments rather than one massive, fragile PR.
Use cases
- Mapping unfamiliar third-party APIs or legacy modules using Research mode.
- Resolving ambiguous product requirements through structured Grilling sessions.
- Breaking down a multi-week feature into a durable sequence of implementable stages.
- Transitioning unfinished work to another developer or a fresh session via Handoff.
Known limitations
Requires a local file system to store .briefback/ artifacts. Not intended for simple one-line bug fixes or trivial documentation updates.
Requires local filesystem access to store .briefback/ artifacts. Best suited for multi-stage features rather than trivial patches.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 20 days ago
- Free with an Agensi account
Requires