More screenshots
Works with the AI Agents you already use
🛰️ Model Release Impact Mapper
Turn sourced model-release or retirement notices into a scoped review queue: affected workflows, excluded deployment routes, unknowns and deadlines.
Free
🛰️ Model Release Impact Mapper
Example session with this skill installed
Use Model Release Impact Mapper with the included DeepSeek V4-Pro portfolio fixture. Map the release’s Responses API support, flexible reasoning effort, and dated pricing window against the declared products. Return JSON and Markdown; separate facts from inferred impact; do not claim a regression, compatibility, demand, or revenue.
- Read your context and instructions
- Compiled the model release impact
Model Release Impact Mapper
Posture: HOLD
Release: DeepSeek V4-Pro by DeepSeek (CN)
Impact coverage: 0/3 item-change pairs resolved
Prioritized retest queue
| Priority | Portfolio item | Change | Domain | Score |
|---|---|---|---|---|
| URGENT | LLM Usage Cost Attribution Reconciler | pricing-window | pricing | 13 |
| HIGH | JustHandled Agent Gateway | reasoning-effort | reasoning | 9 |
| HIGH | LLM Usage Cost Attribution Reconciler | responses-api | api-contract | 9 |
Findings
| Level | Code | Scope | Finding |
|---|---|---|---|
| BLOCK | LIVE_HIGH_RISK_REVALIDATION_PENDING | cost reconciler pricing window | Inferred revalidation obligation; no regression result is claimed. |
| WARN | CLAIM_REVALIDATION_REQUIRED | gateway reasoning contract | The buyer-facing claim intersects the documented release domain. |
| WARN | REVALIDATION_PENDING | Responses API fixture | Evidence remains unresolved. |
Boundaries
- Mapping identifies evidence-backed retest obligations; it does not prove regression, compatibility, safety, demand, or revenue.
- Omitted portfolio dependencies and claims remain outside the assessed scope.
- No provider adapter, paid model call, product mutation, publication, or price change is performed.
Result SHA-256: ee8255891a9dfa1a7815c4bc46dae97057dfe93f47c7e1e960c86fe51dd75a5c
Next steps
- Run each bounded revalidation.
- Attach evidence to the affected claims and artifacts.
- Keep those claims unconfirmed until the checks resolve.
Connects securely to your tools. The creator never sees your data.
About this skill
Know what needs review after a model change
Model Release Impact Mapper v1.1.0 turns supplied release facts and portfolio records into local JSON and Markdown reports. It separates documented facts from inferred impact, with CLEAR, REVIEW or HOLD posture and an evidence-linked retest queue.
Two modes
The existing broad release-impact mode maps declared API, pricing, reasoning, permission, protocol and output-contract changes. The new optional retirement mode matches exact provider, platform and model scope, separates announcement from effective date, and keeps deadlines and unknown inventory visible.
Before you start
Agensi provides agent-connected access, not a ZIP download. Your connected agent needs a Python 3.10+ execution environment and must retrieve the supporting references and script. On October 16, 2026, we tested retirement-mode delivery through Agensi and execution in Claude’s Python sandbox using standard input, without saving the script. A synthetic case and the bundled retirement fixture produced the expected scoped results. This does not certify other hosts or every workflow. A successful connection or instruction load is not a completed run. Bring the source notice and reviewed portfolio/deployment records. It does not gather news, authenticate sources, discover every dependency, call models or evaluate replacements.
Run in your connected agent
Have the connected agent retrieve the required files from Agensi and use its permitted in-session execution route; do not save reusable skill files into your project. Read the input-contract reference for generic release mapping. For retirement notices, read references/retirement-contract.md and use scripts/map_retirements.py with that separate input contract. No provider account or API key is needed.
Retirement example
In a fictional scoped handoff, the same model on Hosted A can match a Hosted A retirement notice while its Direct API deployment is excluded. Unresolved deployment identity remains unknown. A due deadline can produce HOLD, but this is not a failed regression test. Required regression observations remain not run until real evidence exists.
Existing demo
The included DeepSeek portfolio example and video demonstrate the generic release-impact mode, not retirement-mode execution or current provider availability. Included helpers, contracts, fixtures and tests support both modes. Suggested replacements remain untested.
Relies on owner-supplied release facts and declared portfolio domains; it does not browse or authenticate sources. A mapped impact is a retest obligation, not proof of a regression. CLEAR covers only the supplied manifest and evidence. Semantic tool, authorization, agent-control, and protocol checks require both a compatible or sealed runner state and runner-contract evidence. It does not measure demand, run paid model calls, edit products, change prices, or publish updates.
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
This skill has been used 6 times.
Trust & safety
Security scan
Security scan passed 1 month ago
- Free with an Agensi account
Permissions
File Scopes
Runs one bundled local Python helper. Reads one user-supplied UTF-8 JSON packet and writes only the two report paths selected by the user. It does not browse release notes, call models, connect to external hosts, modify the portfolio, or create provider adapters.