القائمة
Scaffold a new Forge tool end-to-end — type, implementation, registration, permission/risk classification, and unit test. Use when adding anything to src/tools/.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
Deep review — random exploration + cross-agent review, alternating. Finds integration bugs that self-review misses. Run every 30–60 min during active implementation.
Overview of Forge's Agentic Coding Flywheel methodology. Use to route between plan, bead, and code space. See FLYWHEEL.md at the repo root.
End-of-session ritual. Close finished beads, file new ones for remaining work, run verify, commit in logical groupings, push. Work is not complete until git push succeeds.
Start a new Flywheel planning round. Produce initial markdown plan, route to competing frontier models, synthesize hybrid, refine 4–5×. Use before any coding begins on a new feature.
Translate a polished markdown plan into self-contained beads in .beads/beads.jsonl. Every bead must be executable without reopening the plan. Use after the plan has converged.
Polishing pass over all open beads. Run 4–6 times. "Check your beads N times, implement once." Use after plan-to-beads and between refinement rounds.
| name | add-tool |
| description | Scaffold a new Forge tool end-to-end — type, implementation, registration, permission/risk classification, and unit test. Use when adding anything to src/tools/. |
| argument-hint | <tool-name-kebab-case> |
Adding a tool touches four places. Do all four in one change, or the tool is half-registered.
Tool name: $ARGUMENTS (kebab-case, e.g. read-manifest).
src/tools/$ARGUMENTS.tsTool from ../types and ForgeRuntimeError from
../types/errors.zod schema for input and output.sideEffect: none | read | write | network | execrisk: low | medium | high | criticalsrc/sandbox/fs.ts helpers.src/sandbox/shell.ts +
classifyCommandRisk.requestPermission from
src/permissions/manager.ts before the side-effecting call, not
after.{ ok: true, data } / { ok: false, error }. Throw only for
programmer errors.src/tools/registry.ts (indirectly)Follow the existing pattern used by sibling tools: export a register
helper or a tool constant and wire it up wherever registerTool is
called at startup. Read one existing tool (e.g. src/tools/read-file.ts)
before writing yours.
test/unit/$ARGUMENTS.test.tsAt minimum, cover:
Use vi.mock for the permission manager and sandbox boundaries. No
real filesystem writes outside os.tmpdir().
If the tool is user-visible, add a row to the tools table in
docs/ARCHITECTURE.md (it lists all 18 tools today). Update the count in
README.md "At a glance" if it changed.
Run /verify when you're done. The test count should increase by the
number of tests you added.