You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
To enhance the spec-kit workflow, I propose an /audit command. It would serve as an automated verification step to be run after the /implement command has finished, ensuring the final output is fully compliant and complete.
How it Would Work
The /audit command would validate the codebase against the project's foundational documents:
Specification Compliance: Checks if the code meets all requirements outlined in spec.md and plan.md.
Constitution Adherence: Verifies the code follows all principles defined in constitution.md.
The command would produce two artifacts:
audit.md: A human-readable report documenting all findings, which is updated with each run.
tasks.md: New remediation tasks are appended to the original tasks.md file. These tasks use the format **RT###** (e.g., **RT001**) to distinguish them from the original tasks (**T###**).
The /audit command can be run iteratively. After the remediation tasks are completed via /implement, another audit can be run until 100% compliance is achieved.
Example Workflow
Core Development: A user runs the standard workflow:
/specify -> /plan -> /tasks -> /implement
The agent completes all **T###** tasks in tasks.md.
Initial Audit: The user runs the verification:
> /audit
The agent analyzes the code and generates audit.md and appends new tasks to tasks.md.
Updated tasks.md:
# Tasks-[x]**T001** Create `src/models/user.js` with a User schema.
-[x]**T002** Implement the `POST /api/users` endpoint.
-[ ]**RT001** Add `lastName` field to the User model as required by `plan.md`.
-[ ]**RT002** Refactor database logic out of the controller into a new service layer to comply with `constitution.md`.
Remediation: The user runs /implement again to fix the audit findings. This cycle can be repeated as needed.
Key Benefits
Closes the Loop: Provides automated verification after implementation.
Actionable Feedback: Generates executable remediation tasks directly into the existing tasks.md.
Enforces Quality: Ensures strict adherence to both the project plan and its architectural rules.
Maintains a Record: The audit.md file serves as a persistent log of compliance checks.
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
-
To enhance the
spec-kit
workflow, I propose an/audit
command. It would serve as an automated verification step to be run after the/implement
command has finished, ensuring the final output is fully compliant and complete.How it Would Work
The
/audit
command would validate the codebase against the project's foundational documents:spec.md
andplan.md
.constitution.md
.The command would produce two artifacts:
audit.md
: A human-readable report documenting all findings, which is updated with each run.tasks.md
: New remediation tasks are appended to the originaltasks.md
file. These tasks use the format**RT###**
(e.g.,**RT001**
) to distinguish them from the original tasks (**T###**
).The
/audit
command can be run iteratively. After the remediation tasks are completed via/implement
, another audit can be run until 100% compliance is achieved.Example Workflow
Core Development: A user runs the standard workflow:
The agent completes all
**T###**
tasks intasks.md
.Initial Audit: The user runs the verification:
The agent analyzes the code and generates
audit.md
and appends new tasks totasks.md
.Updated
tasks.md
:Remediation: The user runs
/implement
again to fix the audit findings. This cycle can be repeated as needed.Key Benefits
tasks.md
.audit.md
file serves as a persistent log of compliance checks.Beta Was this translation helpful? Give feedback.
All reactions