- Home
- Skills
- Technical Documentation
- Release Docs Proof — README & API QA
Works with the AI Agents you already use
Release Docs Proof — README & API QA
Catch broken README and API examples before release. Get the stale line, the evidence and the exact correction.
Free
Release Docs Proof — README & API QA
Example session with this skill installed
Synthetic local example. Our greeting CLI moved from v1 to v2. Check the existing quickstart against the supplied code and intended change. Fix README.md only if it is stale; leave implementation and contract unchanged.
README.md currently tells a new user to run:
python3 greeting.py --name Ada --json
It also says the default output is plain text. CHANGE.md specifies that v2 replaces --json with --format json and keeps that plain-text default. The supplied greeting.py accepts --name and --format {text,json}.
Run the safe local examples, show the exact doc correction and rerun the corrected command. Target fixture revision: d58e2b9713963bdbc536440dcf273d24403db3db.
- Read your context and instructions
- Compiled the release docs proof
Quickstart repaired
Actual synthetic local run on the supplied greeting CLI. The two affected documentation claims are now verified.
| Check | Before | Correction and recheck |
|---|---|---|
README.md:5 quickstart | --json was rejected; exit 2 with unrecognized arguments: --json. | Replaced it with --format json; corrected command exited 0. |
README.md:8 plain-text default | Confirmed by greeting.py:6 and an actual default invocation. | Kept unchanged; output was Hello, Ada!. |
The intended v2 change in CHANGE.md:3 and the implemented --format option agree. The correction is:
-python3 greeting.py --name Ada --json
+python3 greeting.py --name Ada --format json
Corrected command output
{"message": "Hello, Ada!"}
Documentation gate: PASS for these two claims. Only README.md changed. Implementation and the intended-change contract remained unchanged.
The stale command, help text, default invocation and corrected command all ran locally. Installation, external APIs, deployment and the complete product release gate were outside this example.
Connects securely to your tools. The creator never sees your data.
About this skill
The code changed, but the quickstart still uses a removed flag. A new user meets an error on their first command.
Release Docs Proof checks the instructions people will follow against the version you intend to ship. Supply the existing README, quickstart, CLI/API or MCP reference, PR description or release notes, together with the relevant change and product contract.
Find the break, then repair the instruction
Your agent traces each affected claim to code, contracts and safe observed examples. It checks commands, arguments, defaults, compatibility statements and migration steps. You get the document location, what the reader expects, the evidence and an exact correction. If you requested doc fixes, it applies the small patch and reruns the affected checks.
Conflicting code and intended behavior remain a decision to resolve; the workflow will not quietly document an accidental implementation. Commands requiring unavailable access or unsafe side effects stay unverified.
The final result is a documentation-only PASS, HOLD or NEEDS EVIDENCE. Use the ledger to close the specific onboarding or reference gaps before publication. This instruction-only skill reviews existing documentation; it does not establish deployment or whole-product readiness.
It cannot infer intended behavior from conflicting code and product claims. Risky examples remain unverified without suitable authorization and isolation. Partial checkout or missing test environment narrows proof. A docs PASS does not certify deployment or product release.
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
1 use
Times used to date
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 25 days ago
- Free with an Agensi account
Permissions
Read project and documentation files; local shell for safe checks. Write documentation files only when the user requests fixes. Browser or connected MCP access only when needed for authorized evidence.