Give your agent a way to follow the work
A long-running project needs more than a remembered conversation. Before continuing, the agent should inspect what exists, what can run and what the previous work produced.
Wildo provides a separate command for each of those activities. They connect the agent’s next action to project evidence and distinguish doing work from accepting its result.
Example — Resume development with evidence
Ask the agent to read the current state, explain the available next work, run the chosen step and show the result for review. The conversation starts from the project rather than assuming that last session’s plan was completed.
For engineers
Keep inspection, execution and acceptance distinct
These are Claude Code commands shipped in the wildo plugin. Each one calls wildo console verbs, which is also how a person or another agent can drive the same work:
| Command | Responsibility |
|---|---|
/wildo:start | Check the stack and open the vision interview in a new application |
/wildo:status | Read current project and journey evidence |
/wildo:next | Explain eligible work and unmet prerequisites |
/wildo:run | Execute a selected method and inspect its result |
/wildo:build | Dispatch build-plan tasks to coding workers |
/wildo:answer | Answer the questions the companion waits on |
/wildo:accept | Review finished work and commit what the person accepts |
/wildo:why | Explain a refusal or a stuck task from its records |
/wildo:doctor | Check the development stack, the companion and compile health |
Starting work spends model money, so /wildo:run and /wildo:build start only when a person invokes them.
Readiness and sequencing are different. Missing required inputs cannot be bypassed; a sequencing-only restriction can allow an explicitly selected run. The current response supplies the reason to interpret.
Drive the same work from a terminal
These are the console verbs the commands call, with an illustrative playbook reference and question id. Every verb takes --json and exits 2 when the development companion is not running.
wildo creation run <playbook-ref> --brief "Implement the selected change using the current product definition."
wildo creation feed --follow --exit-on-change
wildo creation questions
wildo creation answer <question-id> --revision <revision> "<creator-answer>"
Discover methods with wildo context journey and the next build task with wildo build next. The answer is matched to the question the run asked, and an answer written for an earlier question is refused.
Inspect the kind of result returned
A run may request further input, produce specification files, compose application structure, dispatch coding work or return an assessment without an artifact. Do not assume every successful response creates a specification.
Read the result, inspect the affected files and run the relevant checks. A judge report is review input. Acceptance follows the declared autonomy policy and the project’s accepted state; a run’s own response does not approve itself.