Skip to content

Prompt Index

47 prompts. Find the right one, then go to the Full Prompt Pack to copy it. For usage context and forensic evidence, see 3.2 Prompt Reference.

PL — Planning (13)

ID Name One-liner
PL-01 First Principles Prefix for any significant decision. Pause and think.
PL-02 Plan Draft Generate the first comprehensive plan from scratch.
PL-03 Praise Push I "Barely scratches the surface." Push past default quality.
PL-04 Praise Push II "Still a far cry." Push harder with encouragement.
PL-05 Praise Push III "I KNOW you can do better." Final emotional push.
PL-06 Plan Critique Single-model analytical critique with git-diff output.
PL-07 Multi-Model Synthesis Merge competing plans from 3+ models into one.
PL-08 Integrate Critique Apply review feedback into the plan in-place.
PL-09 Dueling Wizards Two models compete, score, synthesize the winner.
PL-10 Innovation Boost One transformative addition. Not incremental.
PL-11 Premortem Imagine total failure in 6 months. Fix the plan.
PL-12 Project Opinion Honest assessment: is this even a good idea?
PL-13 Alien Artifact Injection Inject mathematically optimal constructs (BOCPD, VOI, etc.).

BD — Beads (3)

ID Name One-liner
BD-01 Plan to Beads Convert the plan into an exhaustive bead graph.
BD-02 QA the Beads Review bead quality. Repeat until changes flatline.
BD-03 BV Triage Use bv --robot-triage to pick the next bead.

EX — Execution (6)

ID Name One-liner
EX-01 Execute Beads Main loop. Read AGENTS.md, check mail, execute next bead.
EX-02 Mail Check & Continue Check agent mail, respond, then keep shipping.
EX-03 Agent Introduction Fresh spawn: read everything, register, start working.
EX-04 Post-Compaction Refresh After context loss: re-read AGENTS.md.
EX-05 Full Push Do ALL remaining work. Every bead. Every test.
EX-06 Git Commit Commit agent only. Group changes, write messages, push.

RV — Review (9)

ID Name One-liner
RV-01 Self-Review After each bead: re-read your code with fresh eyes.
RV-02 Deep Review Randomly explore files, trace flows, find bugs. Numbered sessions.
RV-03 Cross-Agent Review Review other agents' code across the whole project.
RV-04 McCarthy Hunt "The bugs are there. Find them." Paranoid escalation.
RV-05 Stakes Escalation "Your family's life depends on this code."
RV-06 CVE Probe Research recent CVEs, probe for similar vulnerabilities.
RV-07 Stub Eliminator Multi-phase: systematic scan for all stubs/mocks/TODOs, then plan & resolve.
RV-08 UBS Scan Run ubs . and analyze every finding.
RV-09 Random Inspect Pick 5 random files. Read. Trace. Fix.

QA — Quality & Ideation (8)

ID Name One-liner
QA-01 Stripe-Level UI World-class UI/UX. Desktop and mobile separately.
QA-02 E2E Pipeline Complete end-to-end testing. No mocks. No fake data.
QA-03 UX Audit Find every rough edge and unintuitive behavior. Fix all.
QA-04 Root-Cause Fix Don't patch symptoms. Find and fix the underlying issue.
QA-05 Deploy & Verify Deploy, visit with Playwright, screenshot, fix errors.
QA-06 Idea Wizard 30→5 Generate 30 ideas, winnow to 5 best.
QA-07 100-to-10 Filter Generate 100 ideas, present only 10 best.
QA-08 Deep Performance Audit Profile-first optimization with 7-step methodology.

MT — Meta (8)

ID Name One-liner
MT-01 Deep Project Primer Read AGENTS.md + README.md, understand everything.
MT-02 System Weaknesses What are the weakest parts? What needs fresh ideas?
MT-03 README Reviser Update docs to reflect current state. No "we added X."
MT-04 De-Slopifier Kill AI slop: em-dashes, "Here's why", LLM patterns.
MT-05 Code Reorganizer Propose a reorganization plan before touching anything.
MT-06 CLI Error Tolerance Make CLI forgiving: honor intent, helpful errors.
MT-07 Dependency Analysis Write COMPREHENSIVE_ANALYSIS before integrating anything.
MT-08 Agent Feedback Rate <TOOL_NAME> across dimensions. Scores, signal-to-noise, recommendations.