Spec Amend
Surgically update specs, design, tasks, and test files to maintain a living, auditable contract when requirements change.
$9
Works with the AI Agents you already use
Spec Amend
Example session with this skill installed
Amend the export-users specs: include display_name in the CSV too.
- Read your context and instructions
- Compiled the spec amend
I read spec.md, design.md, tasks.md and testing.md, then made the smallest edits that reflect the new requirement. No source code was touched. The change is recorded in the amend log.
What changed
| File | Change |
|---|---|
spec.md | Requirement 1 now lists display_name; one scenario updated |
design.md | displayName added to the UserExportRow contract |
tasks.md | New task T6 added; done tasks stay ticked |
testing.md | TC1 expected header updated |
Diffs
-1. The system SHALL export id, email and created_at for every active user.
+1. The system SHALL export id, email, display_name and created_at for every active user.
interface UserExportRow {
id: string;
email: string;
+ displayName: string;
createdAt: string;
}
- [x] T3. add serializer for UserExportRow
+- [ ] T6. add displayName column to the serializer and the user query
T3 stays ticked because the code exists. T6 tells the implementer what to adjust.
Amend log (end of spec.md)
## Amend log
- 2026-01-15 - Include display_name too - files: spec.md, design.md, tasks.md, testing.md
Next steps
- Review the four small diffs.
- Run the implementation step so T6 is built against the updated contract.
PASS
Connects securely to your tools. The creator never sees your data.
About this skill
Specs often rot the moment a requirement changes, leading to a permanent drift between documentation and code. This skill enforces a "specs-first" change management workflow, ensuring that every shift in project scope is surgically recorded across your entire documentation suite before a single line of code is touched.
What it does
- Surgical editing updates only the specific requirements, technical decisions, or tasks affected by new feedback.
- Audit logging appends verbatim user feedback and timestamps to a persistent amend log in the spec file.
- Task invalidation manages task lists by marking obsolete items and adding new ones without erasing history.
- Coherence checking verifies that changes are reflected consistently across spec, design, tasks, and testing files.
- Scope guarding detects when a change is too large for an amendment and triggers a full proposal restart.
How it works
- Analyze input against existing
spec.md,design.md,tasks.md, andtesting.mdfiles. - Identify gaps in logic or ambiguity in the feedback, stopping for clarification if needed.
- Execute edits using non-destructive operations to maintain a clear audit trail of what changed and why.
- Log the event with an ISO date and the exact feedback text.
- Issue a verdict of
PASSorGAPbased on the internal coherence of the updated files.
Frameworks & tools
This tool is language-agnostic and works with any codebase using Markdown-based documentation (spec.md, design.md, tasks.md, testing.md).
Why this beats prompting it yourself
Generic prompts often cause LLMs to rewrite entire files, losing context and history. This skill enforces strict surgical edit rules and prevents the "unticking" of completed tasks, maintaining an honest project history that generic instructions usually ignore.
Use cases
- Update a feature requirement after a stakeholder changes their mind mid-sprint.
- Adjust technical design decisions without losing the original context of the architecture.
- Re-align testing suites and task lists when a specific edge case is redefined.
- Maintain a regulatory or project-governance audit trail of why requirements changed.
Known limitations
Does not edit source code. Requires existing Markdown documentation files to function. Restricted to surgical edits rather than full-scale project pivots.
Does not edit source code. Requires existing Markdown documentation. Explicitly stops if feedback is ambiguous or requires a total project restart.
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
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 19 days ago