Works with the AI Agents you already use
Evidence Freshness Check
Check whether earlier verification still applies after files, configuration, versions, or external state change; identify stale and unverifiable claims.
Free
Evidence Freshness Check
Example session with this skill installed
Check the freshness of the 'Auth Middleware validation' claim from the previous session. Files in src/auth/ and the .env config have changed.
- Read your context and instructions
- Compiled the evidence freshness check
I have audited the 'Auth Middleware validation' claim against the current state of the repository. While the middleware logic remains largely intact, the configuration drift and dependency updates have invalidated the previous verification.
Evidence Freshness Report
| Claim | Scope | Status | Reason |
|---|---|---|---|
| JWT Secret Validation | .env | Stale | JWT_SECRET_KEY digest changed; previous entropy check invalid. |
| Middleware Unit Tests | src/auth/*.ts | Current | File digests for unit tests and implementation match previous record. |
| Bcrypt Dependency | package-lock.json | Unverifiable | Version mismatch (v5.0.1 -> v5.1.0); binary compatibility not checked. |
| Role-Based Access | src/auth/roles.ts | Stale | New exported constants found that were not present in previous scope. |
Next steps
- Rerun the entropy validation script for the updated
JWT_SECRET_KEYin.env. - Perform a compatibility check for
bcryptv5.1.0 to ensure native bindings are functional. - Update the test suite to include coverage for the new roles defined in
roles.ts.
Connects securely to your tools. The creator never sees your data.
About this skill
Check whether earlier evidence still supports a claim after its files, configuration, or external state changes.
Use this when reusing an earlier check after changes or when evidence crosses a session boundary. It is unnecessary for a fresh, direct observation whose scope is already clear.
For each important claim, record the inspected target, version or observable state, relevant dependencies, observation, check time, source kind, and conditions that would invalidate it. Separate direct observation, a user's report, and an agent assertion. None should silently stand in for another.
Check an immutable snapshot or pause changes to the selected files while comparing them. Reading several files is not an atomic view of the filesystem. If the files change during inspection, report unverifiable and repeat the check against a stable source.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for LLM Ops.
Cannot observe inaccessible external state or undeclared dependencies. Does not grant execution permissions. Atomic filesystem views are not guaranteed without snapshots.
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 buying a skill.
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 to use
No account required to browse
Trust & safety
Security scan
Security scan passed 11 days ago
- Free with an Agensi account