Guide multi-step coding tasks with deepseek-v4-pro or deepseek-v4-flash in thinking mode. Verify files and symbols, write plans with confirmed file paths and line numbers, and use a verifier agent before changes spanning three or more files.
Use when working with deepseek-v4-pro or deepseek-v4-flash in thinking mode on multi-step or plan-driven tasks. Provides rules to prevent stale references, unverified plan assumptions, and vague plan output.
Review the instructions the agent will receive before you install.
Rules for multi-step V4 thinking-mode workflows. Each rule prevents a specific, observable failure class.
Before referencing a file path, function, or type in code or plan output,
call grep_files or read_file to confirm it exists in the workspace.
# Bad: edit_file path="src/config/loader.rs" (assumed from memory)
# Good: grep_files pattern="pub fn load_config" → confirms src/config/mod.rs:42
# then reference src/config/mod.rs:42
Failure avoided: edit_file errors on non-existent paths; LSP diagnostics
on hallucinated symbols.
Before executing a plan that touches 3+ files, spawn a deepseek-v4-flash
sub-agent (thinking off) to read the target files and confirm path/symbol
assumptions still hold.
agent type="verifier" model="deepseek-v4-flash"
prompt: "Read these files and confirm: [list assumptions]. Report mismatches."
Failure avoided: multi-step edits fail partway because file structure changed since the plan was drafted.
In plan-mode output, replace vague location pointers with path:line
references drawn from a prior grep_files result.
# Bad: "Update the retry logic in the client module"
# Good: "Update retry loop at crates/tui/src/client.rs:187"
Failure avoided: agent-mode execution cannot locate the intended edit target when plan directions are imprecise.
No reviews yet. Tell others what you thought.
No login needed. Share how it worked for you, ask a question, or note where it did not run.
No comments yet. Be the first to write one.
Split a large goal into clearly bounded tasks that can run in parallel. Coordinate existing agent and worktree tools, then combine and verify the results. The skill starts only when explicitly requested.
Generate independent options for design, implementation, explanations, or debugging, then compare them using one shared rubric. Code candidates use separate working copies and undergo independent verification before a winner is presented for review.
Check Codewhale release readiness through ordered tests, builds, version checks, and manual terminal checks. Record command results and observed behavior, including responsiveness, route isolation, and follow-up input handling, while explicitly identifying any skipped checks.
Operate desktop apps and isolated browsers on registered computers by inspecting interface controls, entering information, and checking results. The skill also covers screenshots and recording, with separate routing for the user's signed-in Chrome through Chromewhale.
Check a local Codewhale checkout before contributing: inspect branch and sync status, build the CLI, and run the repository’s CI checks. Present recent commits and changelog entries from local files, with clear status for each step.
Builds and installs Codewhale with a source revision identifier, then verifies the installed version in a fresh shell. It guides hands-on runtime checks for interface behavior, routing, approvals, and other changes that automated tests may miss.