57 57 "prompt": "Investigate this bug without modifying files: {args.bug}. Reproduce or trace the failing path, inspect callers and shared routes, identify the likely root cause, and list relevant files. Use read/grep/find/bash as needed. Return JSON matching Investigation.",
66 68 "prompt": "Create the smallest regression test for this bug: {args.bug}. Investigation: {investigate.results}. Do not fix production code. Run the focused test and set ready=true only when it fails for the expected reason, not from a test error. Return JSON matching Regression.",
76 80 "prompt": "Fix the root cause of this bug: {args.bug}. Investigation: {investigate.results}. Regression evidence: {regression.results}. Prefer the shared route over caller-specific patches. Make the minimum production change, run the regression test and relevant checks, and set fixed=true only when they pass. Then return JSON matching Fix.",
87 93 "prompt": "Adversarially verify the bug fix from the {item} perspective. Bug: {args.bug}. Investigation: {investigate.results}. Regression: {regression.results}. Fix: {fix.results}. Inspect the diff and run focused checks as needed; do not modify files. Set passed=false and list concrete findings if the fix is incomplete, the regression test is weak, or new risk was introduced. Echo dimension as {item}. Return JSON matching Verdict.",
100 100 "prompt": "Research this codebase to implement the feature: {args.task}. Identify the files you'd touch, the existing patterns/conventions to follow, and any constraints. Inspect with read/grep/find/bash (e.g. `git log`, `git grep`); do not modify anything. Return JSON matching the Research schema.",
109 111 "prompt": "Produce an implementation plan for: {args.task}. Ground it in this research: {research.results}. Give concrete ordered tasks, a one-paragraph approach, and the real risks. Read further if needed; do not modify anything. Return JSON matching the Plan schema.",
118 122 "prompt": "Adversarially critique this plan BEFORE any code is written. Task: {args.task}. Plan: {plan.results}. Research: {research.results}. Hunt for missing steps, wrong assumptions, and unhandled edge cases. Set approved=false and list each blocking issue if the plan isn't ready; default approved=false when uncertain. Do not modify anything. Return JSON matching the Critique schema.",
128 134 "prompt": "Implement the feature: {args.task}. Follow the plan and resolve EVERY blocking item from the critique. Plan: {plan.results}. Critique: {critique.results}. Make the actual edits with the edit/write tools, keep the change focused, and follow the conventions surfaced in research. Run focused checks and set implemented=true only when the changes and checks complete. Then return JSON matching the Implementation schema (list the files you changed and a concise summary).",
139 147 "prompt": "Review the just-made implementation of {args.task} from the {item} perspective. What changed: {implement.results}. Inspect the working tree with read/grep and `git diff`; do not modify anything. Set passed=false and list each blocking {item} problem you find. Echo the `dimension` field as \"{item}\" so the verdict ties back. Return JSON matching the Verdict schema.",
44 44 "prompt": "Review the requested target for {item} issues. Target: {args.target}\n\nThe target may be a PR, branch/ref, recent changes, folder, or file. If target is empty, inspect the working tree and recent local changes. Use git/read/grep/find/ls as appropriate; for PRs or branches, inspect the relevant diff; for folders/files, inspect that scope. Do not modify files. Return JSON matching the Findings schema, and set each finding's `file` to the path it concerns.",
54 56 "prompt": "Adversarially try to refute this finding. Default to real=false if uncertain. Echo the finding's `file` verbatim in your verdict's `file` field so the verdict ties back to its finding.\n\nFinding: {item}",